Release Notes

The complete Pervaziv AI public release archive now lives here.

September 2026

Sep 10, 2026

Cortex 5.7

Managed Cloud Execution for Enterprise AI with Cortex Cloud

From connected intent to durable cloud execution across mobile, browser, and Visual Studio Code
Enterprise AI is moving from answering questions to helping people complete meaningful work.
That shift changes the role of the environment around the AI. A useful answer can be generated almost anywhere. Engineering work is different. It depends on source code, project state, development tools, permissions, compute, continuity, and a clear understanding of where an action is allowed to run.

Cortex Connect established a continuous experience across mobile, browser, and Visual Studio Code. It allowed a request to begin where the need appeared, remain visible across work surfaces, and reach the connected development workspace where the relevant project context lived.

Cortex Cloud takes that experience to the next level.
With this release, users can choose between Local execution and Cortex Cloud directly from the Cortex experience. Local keeps eligible work close to the connected development environment. Cortex Cloud extends the workflow into a managed cloud execution environment designed for durable, remote, project-aware work.

The choice is simple at the surface. Underneath, Cortex coordinates the intelligence, workflow state, project version, execution environment, progress, and result as one connected experience.

This is the base Cortex Cloud release: a foundation for cloud execution across Cortex, focused on continuity, control, and a clear path from user intent to completed work.

Sep 4, 2026

Cortex 5.6.5

Delivering Long-Running Workflows Without Losing Context

Smarter Context, More Reliable AI
AI assistants are moving beyond isolated questions and short conversations. Customers increasingly rely on them for long-running development tasks, document analysis, research, troubleshooting, and workflows involving multiple files, tools, and decisions.
As these interactions grow, so does the amount of context the AI must process. Requirements, attachments, code, validation results, earlier decisions, and conversational history all compete for a finite context window.
Context compaction helps solve this problem by reducing lower-value historical material while preserving the facts and instructions required to continue safely.
Across our evaluated semantic-compaction benchmark scenarios, Cortex reduced eligible supplemental context by 46% while maintaining complete required-fact recall and successful source retrieval.

Why context management matters
Long conversations create several challenges:
– Important requirements can become buried beneath repetitive content.
– Earlier decisions may conflict with newer instructions.
– Large attachments and tool outputs can consume substantial context.
– Similar identifiers and file paths can become confused.
– Generic summaries can omit critical details or introduce unsupported claims.
– Eventually, an interaction can approach the model’s context limit.

Simply retaining everything is inefficient. Indiscriminately deleting older material is unsafe.
Effective context management must retain the information that determines correctness while reducing material that no longer needs to appear in full.

Sep 1, 2026

Cortex 5.6

From Intent to Action with Three Cortex Routers
Coordinating models, search, and engineering expertise through one Enterprise AI experience
Enterprise AI is moving beyond the single-model assistant.

Software work spans coding, testing, security, research, review, remediation, and verification. Although these activities may begin in the same chat interface, they require different forms of intelligence and context.

Cortex addresses this through three complementary capabilities:
– Model Routing connects requests with the appropriate Cortex intelligence.
– Search Routing provides access to current public information across multiple search services.
– Skill Routing brings focused engineering practices into relevant workflows.

Together, they help Cortex understand not only what a user is asking, but also what type of intelligence, information, and engineering expertise can best support the outcome.
The result is a simpler experience for users and a more adaptable AI foundation for organizations.

One experience, specialized intelligence
AI platforms are becoming more capable, but added capability can create added complexity.
Users may be asked to choose between models, agents, tools, search modes, and large collections of specialized instructions. That exposes the internal architecture of the AI system to the person simply trying to complete a task.

Cortex takes a different approach.
Developers and enterprise users express what they want in natural language. Cortex coordinates the appropriate capabilities behind a unified interface.

August 2026

Aug 25, 2026

Cortex 5.5

From specialized intelligence to connected execution across every work surface

Enterprise AI is becoming part of how modern teams understand, build, secure, and improve software. But the way people work has never been limited to a single screen.
An idea may begin on a phone between meetings. Research may happen in a browser. A production question may surface while reviewing an issue or reading documentation. The implementation itself may need to happen in Visual Studio Code, where the code, development tools, and project context already live.

Until now, those moments have often been disconnected. AI can answer a question on one surface, but the user must carry the context and the work back to another. Conversations become fragmented. Intent gets lost. Progress becomes difficult to follow. The user is left acting as the bridge between insight and execution.

Cortex Connect changes that experience.
Cortex Connect allows a user to begin with Cortex on mobile or in the browser and securely continue the work in a connected Visual Studio Code workspace. The request, progress, and outcome remain part of one coordinated Cortex experience, helping enterprise AI move beyond isolated conversations toward useful, governed action.

It is a natural next step in the evolution of the Cortex AI Model Ensemble and the planning capabilities introduced with Cortex Planner. The Ensemble brings together specialized intelligence. Cortex Router helps direct the request. Cortex Planner organizes the work. Cortex Connect closes the distance between those capabilities and the development workspace where the work can be completed.

Aug 21, 2026

Cortex-Planner-1.8 Cortex 5.4.5

AI-assisted software delivery is moving beyond individual code suggestions. The next frontier is coordinated execution: understanding the work, organizing the right expertise, respecting dependencies, and validating outcomes before change reaches production.

With the release of Cortex Planner, Cortex adds its ninth specialized model.

Cortex Planner 1.8 brings a new layer of intelligence to the Cortex experience. It transforms a development request into a clear, governed execution plan—helping teams move from “what needs to happen” to “how the work should be coordinated” with greater confidence, visibility, and control.

From routing work to planning outcomes
Cortex Router established a smarter way to direct work across Cortex’s specialized AI capabilities. It evaluates a request and helps identify the right path and the right expertise for the task.

Cortex Planner extends that vision.
Where routing determines where work should go, planning defines what must happen next. Cortex Planner interprets the objective, available context, and permitted actions to create an organized plan for analysis, implementation, testing, security review, integration, and validation.

This is an important step in the evolution of AI-assisted development. Complex engineering work is rarely a single action. A request to add a feature, resolve a defect, investigate a security concern, or improve an existing service may involve several distinct activities—some independent, some sequential, and some requiring careful review before they can move forward.

Cortex Planner helps make that work visible, intentional, and easier to govern.

Aug 7-13, 2026

Cortex-Router-1.7 Cortex 5.4

Introducing Cortex Router for the AI Model Ensemble

AI-assisted development is moving from isolated chat interactions to real work across the software lifecycle. A developer may begin with a quick question, move into an investigation, ask for a code change, request a security review, and then need to understand what happened before a release. The experience should feel continuous. But the work itself is not one kind of task, and it should not be treated as one.
Today, we are publicly announcing Cortex Router 1.7.

Cortex Router is the coordinating entry point for Cortex. It helps recognize the kind of work a developer is asking to do and guides that request toward the most appropriate specialized model or next experience in Cortex. It is designed to make AI-assisted development feel less like selecting from a menu of disconnected capabilities and more like working through a single, coherent engineering environment.

This release builds on the Cortex AI Model Ensemble story. We introduced the ensemble as a collection of specialized models, each focused on a meaningful part of the development workflow rather than asking one general-purpose model to carry every responsibility. We later added AI Code Verify as the seventh specialized model, expanding the ensemble’s ability to support greater confidence around code changes. Cortex Router is the connective layer that helps customers realize the value of those capabilities in the flow of work.

The result is a clearer access layer for the ensemble: one that helps turn a developer’s intent into the right next step, while keeping the experience direct, understandable, and business-ready.

Cortex Enterprise AI For Everyone With Free Tier
AI is rapidly becoming part of how modern organizations design, build, secure, and deliver software. But for many teams, the first question is still practical: Where do we start?

Cortex was built to answer that question with a clear path forward. It brings AI assistance, security insight, and connected enterprise context into the development workflow, then grows with your team as your needs expand.

Today, we are making that first step easier: Cortex Enterprise AI is now available through a Free Tier. New users can experience basic Cortex chat in VS Code with no payment information required. It is a focused, low-friction way to see how Cortex fits into the work developers already do every day.

This is the next chapter in the Cortex Enterprise AI story. At the heart of Cortex is the Cortex AI Ensemble: a coordinated set of AI capabilities designed to support coding, security, analysis, and connected enterprise work. The Free Tier makes it easier for developers and teams to experience that intelligence in action before selecting the package that fits their business.

Aug 5-6, 2026

Cortex 5.3.7-5.3.9

5.3.7 Advancing Cortex Reliability for Secure, Enterprise AI Development

AI coding assistants are rapidly becoming more capable, but enterprise adoption depends on more than code generation quality.

Development teams need AI systems that can understand complex requests, stay aligned across longer tasks, work safely within project boundaries, recover from failures, and provide results that can be reviewed with confidence.
Pervaziv AI has completed a major reliability initiative for Cortex focused on strengthening these capabilities across coding, security, validation, and agentic development workflows.

The improvements help Cortex operate with greater continuity, precision, security, and accountability as tasks move from simple suggestions to multi-step engineering work.

5.3.9 Building a More Trustworthy Cortex for Enterprise Software Delivery
Enterprise teams do not need AI that merely writes code. They need AI that fits into real software-delivery practices: clear ownership, dependable validation, controlled access, useful continuity, and evidence that helps people decide what should happen next.

This Cortex update advances that goal. It strengthens the end-to-end coding-agent experience—from request through code change, validation, review, recovery, and learning—while keeping developers and organizations in control.

From code generation to accountable delivery
Cortex is evolving from a coding assistant into a more complete development partner. It can help teams understand requests, prepare code changes, run appropriate checks, summarize outcomes, and present the resulting work for review.

The experience now makes the status of a coding task easier to understand at a glance. Developers can see the files involved, the scope of a change, validation status, and whether a change is still awaiting review.
This is important in enterprise environments, where a natural-language statement that work is “done” is not enough. Teams need a concise, reviewable record of what the agent actually produced and what evidence supports it.

Aug 3-4, 2026

Cortex-Verify-1.6 Cortex 5.3.5-5.3.6

5.3.5 Cortex Adds Personal Coding Style and Test Design Specification.
AI-assisted software development is becoming a normal part of how organizations build, maintain, secure, and deliver software. But adoption alone is not the goal. The real opportunity is to help teams move faster without losing the engineering standards, quality practices, and organizational context that make software dependable.
That is why Cortex has recently introduced two major capabilities across VS Code, Browser and Mobile: Personal Coding Style and Test Design Specification

5.3.6 Introducing AI Code Verify, Seventh Model in the AI Ensemble
Bringing clearer validation to AI-assisted software delivery
AI-assisted development is changing how organizations build software. Teams can move from an idea to a proposed implementation faster than ever, with AI helping to explain code, generate changes, investigate issues, recommend improvements, and create tests.

But faster output does not automatically mean more dependable software.
For enterprise teams, the question is not simply whether AI can propose a patch. It is whether the proposed change is aligned with the intended requirement, consistent with engineering standards, supported by meaningful tests, and ready for the right level of human review. That is the purpose behind Cortex Verify 1.6, the newest addition to the Cortex AI Model Ensemble.

July 2026

July 24, 2026

Cortex 5.3

Cortex Connects Salesforce CRM with Engineering and Cloud Workflows

Expanding Cortex Enterprise AI with Salesforce CRM, the 10th MCP connection and 46th AI Agent in Cortex’s growing enterprise ecosystem.

Cortex already connects software teams to the systems where engineering and operational context lives: GitHub, Atlassian, Slack, Azure DevOps, Microsoft 365, Google Workspace, Google Cloud, AWS, and Azure. Salesforce adds the customer and revenue dimension to this connected enterprise workspace.

This release builds on Cortex’s multicloud direction: connecting the tools used to build, secure, operate, and support modern software in one AI-driven experience. Cortex 3.2 introduced AWS integration, and Cortex 3.5 expanded that vision across AWS, Azure, and Google Cloud. Salesforce CRM now brings customer relationship intelligence into the same workflow.

From engineering context to customer context. Enterprise work rarely stays within a single system.
A product team may investigate a customer-reported issue in Salesforce, review the related support discussion, trace the implementation in GitHub, validate a deployment in Azure or AWS, and update a work item in Jira.
Traditionally, that process involves repeated context switching, manual searching, and handoffs between technical and business teams.

With Salesforce CRM support in Cortex, authorized users can bring relevant CRM context into their existing AI-assisted workflows. Cortex can help teams explore Salesforce objects, locate accounts and contacts, review opportunities and cases, and work with CRM records through a conversational experience.
The goal is not to replace Salesforce. It is to make Salesforce context available where technical and operational decisions are already being made.

July 16, 2026

Cortex-Safety-1.4 Cortex-Code-1.5 Cortex 5.2

Specialized On-Device and Backend Models for Secure, Private AI-Assisted Development

Cortex 5.2 introduces the Cortex AI Model Ensemble: a coordinated family of specialized AI models designed to support secure software development across both on-device and backend environments.

AI-assisted development should not depend on one broad model trying to solve every task. Privacy protection, prompt-injection defense, security analysis, safety decisioning, remediation, and broad coding assistance each require different model behavior, latency profiles, security boundaries, and evaluation criteria.

The Cortex AI Model Ensemble brings these capabilities together as a layered AI system. On-Device Models operate close to the developer and sensitive context, while Backend Models provide deeper reasoning, security analysis, safety-aware orchestration, and broader coding assistance.

Together, they help Cortex deliver AI that is more private, more secure, more capable, and more practical for enterprise software development.

The Cortex AI Model Ensemble includes six specialized models:
Cortex-LLM 1.0 — foundational secure AI capability for structured software-security workflows
Cortex Privacy 1.1 — on-device sensitive-data detection
Cortex Prompt Guard 1.2 — on-device prompt-injection and instruction-risk classification
Cortex Analysis 1.3 — backend security analysis and structured findings
Cortex Safety 1.4 — backend safety-aware decisioning for secure AI workflows
Cortex Code 1.5 — a broad coding agent for software-development tasks

This is not a collection of disconnected models. It is an ensemble architecture in which each capability contributes at the point where it is most effective.

Local models provide rapid preflight controls before sensitive or untrusted content enters a larger AI workflow. Backend Models provide the contextual reasoning needed for analysis, secure development assistance, safety validation, and broad coding tasks. This division enables Cortex to apply the right level of AI capability without sending every request through the same model path.

July 8, 2026

Cortex-Privacy-1.1 Cortex-Prompt-Guard-1.2 Cortex 5.1

Building Private, Low-Latency AI Controls for Developer Workflows

AI-assisted software development is becoming part of everyday engineering work. Developers use AI to review code, explain issues, summarize context, generate fixes, and move faster across increasingly complex software systems.
But enterprise adoption still depends on a very practical question: where does sensitive context go?

Not every AI decision should require a remote model call. Some security and privacy controls are most valuable when they run directly inside the user experience, close to the developer, before sensitive content is sent anywhere else.

That is the direction behind our on-device local model work: specialized AI controls that help protect developer data, detect prompt-injection risk, and support secure private distribution without adding friction to the development workflow.

These local models are designed to run where Cortex users already work: inside VS Code and across browsers – Chrome, Safari, Edge, and Firefox. The goal is not to ask developers to change tools. The goal is to make privacy and safety controls available directly inside the surfaces where AI-assisted development happens.

The theme is simple: keep high-frequency safety decisions local whenever possible, and reserve larger remote or agentic systems for tasks that truly need deeper reasoning.

This work centers on three core capabilities:
– Cortex Privacy
– Cortex Prompt Guard
– Cortex Secure Distribution

Together, these capabilities help create a more secure foundation for AI-assisted development. They combine local inference, product-ready ML behavior, versioned model delivery, integrity verification, and enterprise-friendly lifecycle management.

July 1, 2026

Cortex-LLM-1.0 Cortex 5.0

Building Specialized AI Capabilities for Secure Software Development

We have been expanding our security-focused AI capabilities to support a more complete secure development lifecycle. We are thrilled to announce that we have trained and built our first, very own AI Model and we are releasing it as Cortex-LLM 1.0. The goal is straightforward: help engineering teams identify security issues earlier, understand them more clearly, and move toward remediation with less manual overhead.

This work brings together two complementary capabilities. One is focused on security analysis: reviewing selected code context and producing structured security findings that can be consumed by downstream systems. The other is focused on security remediation: helping convert a validated issue into a practical, targeted code change.
Together, these capabilities support a closed-loop workflow: analyze code, validate findings, recommend action, apply fixes, and re-check results.

Cortex-LLM 1.0 shows a strong balance of capability and efficiency across both benchmark views. When plotted on secure instruction-following performance versus runtime performance using CyberSecEval from the PurpleLlama project, Cortex-LLM 1.0 lands in the desirable quadrant, combining strong security-aligned behavior with practical execution speed. 

The same pattern holds for code completion accuracy versus runtime performance on HumanEval, where Cortex-LLM 1.0 again leads the comparison by delivering high coding accuracy without sacrificing responsiveness. Together, these results indicate that Cortex-LLM 1.0 is not only competitive on model quality, but also well-positioned for real-world developer workflows where both correctness and latency matter.

The latest work shows promising progress. Structured security analysis is becoming more reliable, especially when paired with validation and runtime normalization. Remediation behavior is being kept focused on explicit security fixes rather than broad code generation. Evaluation has become more realistic, with separate checks for curated behavior, held-out generalization, false positives, false negatives, and output validity.

June 2026

June 24, 2026

Cortex 4.9

With Safari support, Cortex 4.9 now reaches users across Chrome, Microsoft Edge, Firefox, and Safari, giving organizations a broader and more consistent way to bring secure AI assistance into everyday browser workflows.

This is an important milestone for the Cortex Enterprise AI Control Layer. Modern work does not happen in one place. Developers review code in repositories, security teams investigate findings in dashboards, business users move across web applications, and enterprise teams increasingly expect AI to be available inside the tools they already use.

Cortex has already brought AI-powered coding, security review, enterprise context, and productivity workflows to IDEs, browsers, cloud-connected systems, and mobile environments. Safari support completes a major part of that browser journey by extending Cortex to Apple-first teams and macOS users who rely on Safari as part of their daily workflow.

Features of this release:
– A Native Path for Apple Users
– Apple Sign-In and Subscription Support
– A Broader Apple Ecosystem Expansion
– Built for Enterprise Trust, Not Just Access
– Secure AI Assistance in the Browser
– Enterprise AI Wherever Work Happens

For Apple-first users, Safari support removes an important gap. For enterprises, it simplifies rollout across mixed environments. For security and development teams, it brings Cortex closer to the actual flow of work.

June 15, 2026

Cortex 4.7

Safer Validation, Smarter Remediation, and Enterprise-Ready Security Workflows

Cortex 4.7 advances the product in a direction that matters deeply for enterprise AI adoption: moving from AI-assisted coding to AI-assisted engineering control. As coding agents become more capable, the question for organizations is no longer whether AI can generate code, suggest fixes, or summarize security issues. The real question is whether those actions can be validated, governed, traced, and trusted inside real software delivery environments.

Minor releases
4.7.1: Stronger Post-Change Validation for AI-Assisted Development
4.7.2: Repository-Aware Guidance and Governed Workflow Orchestration
4.7.3: Structured Validation and Iterative Validation Lifecycle
4.7.4: Security Analysis and Remediation Built Around Code Context
4.7.5: Dependency and Infrastructure Risk Review

A More Enterprise-Ready AI Development Lifecycle
The broader story of Cortex 4.7 is control. AI coding systems are becoming more powerful, but power without control creates risk. Enterprises need to know what the AI changed, what was validated, what failed, what security risks remain, and what evidence supports a decision to merge, reject, remediate, or accept risk.

This release helps reduce the gap between AI-generated suggestions and production-ready outcomes. It supports safer coding workflows, stronger validation practices, more contextual security remediation, and better handling of supply chain and infrastructure risk. That combination is essential for organizations that want to scale AI-assisted development without weakening their security posture.

June 15, 2026

Cortex 4.8

Enterprise AI wherever work happens. Secure AI assistance, enterprise context, and security workflows now available on iPhone.

Cortex 4.8 extends the Enterprise AI Control Layer to iPhone, bringing the Cortex mobile experience to iOS users with secure AI assistance, connected enterprise context, privacy-aware workflows, and continuity across devices.

With Cortex already available across browsers, IDEs, enterprise workflows, and Android, this release expands the platform to another critical part of the modern workday: the iPhone. Users can now continue conversations, review ideas, ask operational questions, assess security concerns, and access enterprise-aware AI assistance while away from their desktop environment.

Cortex 4.8 builds on the Android foundation introduced in Cortex 4.5 and brings the same enterprise-first approach to iOS: mobility without sacrificing context, security, privacy, or trust.

Release features:
– Enterprise AI Workspace on iPhone
– Subscriptions and Sign In Built for iPhone
– Privacy First, Before Send
– Connected Enterprise Context
– Security Workflows On The Go
– Enterprise Continuity Across Devices
– iPhone Ready Experience
– Reliability and Trust
– Enterprise AI Wherever Work Happens

June 3, 2026

Cortex 4.5

Enterprise AI Control Layer for Android

Enterprise AI wherever work happens. Secure AI assistance, enterprise context, and security workflows now available on Android.

Cortex 4.5 extends the Enterprise AI Control Layer beyond browsers and IDEs, bringing secure AI assistance, connected enterprise context, security workflows, and chat continuity directly to Android devices. This release enables users to access Cortex wherever work happens, whether reviewing ideas between meetings, responding to operational questions, continuing an existing conversation, assessing security risks, or preparing next steps while away from a workstation.

Cortex 4.5 released with these key features:
– Enterprise AI Workspace on Android
– Privacy First, Before Send
– Connected Enterprise Context
– Security Workflows On The Go
– Enterprise Continuity Across Devices
– Mobile Ready Attachments, Voice, and Export
– Refined Android Experience
– Reliability and Trust

Cortex 4.5 represents another step forward in the evolution of the Enterprise AI Control Layer.
Previous Cortex releases expanded AI assistance across IDEs, browsers, enterprise systems, cloud environments, security workflows, and privacy-aware operations. Android extends that same vision to the mobile edge, ensuring that users can access Cortex whenever work requires insight, context, or assistance.

Whether reviewing a security concern, continuing an existing conversation, accessing enterprise context, evaluating architectural decisions, or simply capturing ideas while away from a workstation, Cortex remains available as a trusted AI companion.
Cortex 4.5 brings secure AI assistance, enterprise context, and intelligent workflows closer to where modern work actually happens: everywhere.

June 2, 2026

Cortex 4.4

A More Connected AI Security Workspace

Cortex has advanced from a collection of security features into a more unified AI security workspace. The latest updates strengthen the experience across the console, VSCode, and browser extension, giving teams more ways to review software risk without disrupting how they already work.

Cortex 4.4 released with several features:
– Developer-Native Security In The IDE
– AI Summaries For Faster Security Decisions
– A Stronger Console Experience
– Usage Controls Designed For Commercial Scale
– Browser Expansion with Security Improvements
– Cleaner Reporting And Sharing
– A More Enterprise-Ready Product Surface

The post-4.3 work makes Cortex feel more like a coordinated platform. The console provides the central workspace, VSCode brings security closer to developers, and the browser extension extends AI assistance into web-based workflows. Cortex helps teams discover, understand, and act on software security risk wherever work is already happening.

It combines AI-assisted review, repository-level analysis, browser-based assistance, developer-native workflows, cleaner reporting, and commercial-grade usage controls into a more complete security workspace for modern engineering teams.

May 2026

May 28, 2026

Cortex 4.3

Cortex 4.3 advances the Enterprise AI Control Layer with a strong focus on reliability, session continuity, secure authentication, and smoother day-to-day usage across VS Code, browser-based workflows, and the Cortex web experience.

The release builds on the foundation introduced in Cortex 4.2, where AI threat modeling, security review, and developer-side intelligence became more deeply embedded into engineering workflows. Cortex 4.3 takes the next step: making those workflows more dependable across multiple environments, multiple sessions, multiple connected services, and longer-running AI interactions.

Cortex now behaves more consistently across tools, preserves work more reliably, and reduces friction when teams move between development, browser, and console-based experiences.

Highlights of this release:
– Built for Real Enterprise Usage
– Session Continuity Across VS Code, Browser, and Console
– More Reliable Real-Time AI Interaction
– Cleaner, More Focused Chat Experience
– Safer Agentic Code Change Workflows
– Improved Validation and Project-Aware Workflows
– More Dependable Authentication Across Tools
– Connected Enterprise Services and MCP Readiness
– Browser Experience: More Context, Better Continuity
– Stronger Authentication and Session Backbone
– Privacy-Oriented Error Handling
– Better Quality Foundation

Cortex 4.3 moves that vision forward by improving the connective tissue between experiences. It makes Cortex more resilient, more continuous, and more practical for teams using AI across real development and security workflows.

For developers, Cortex becomes easier to trust during active implementation. For security teams, Cortex becomes more consistent during review, validation, and remediation. For enterprise leaders, Cortex becomes a stronger platform for bringing AI into controlled software delivery workflows.

Cortex 4.3 is a release about dependability, continuity, and enterprise-grade workflow maturity.

May 18, 2026

Cortex 4.2

Cortex now feels more like a dependable engineering and security partner inside the developer environment. The update focuses on three outcomes: better understanding of the user’s project, smoother handling of code changes, and a more controlled experience when the assistant takes action.

Cortex now has a stronger foundation for understanding project context. Instead of relying only on the visible file or manually attached snippets, it can use structured workspace signals to decide what information is relevant to a request.

🔐 AI Threat Model and AI Security Review
The release introduces two new security workflows: AI Threat Model and AI Security Review. AI Threat Model helps teams reason about how a system could be attacked before issues become incidents, including sensitive assets, trust boundaries, exposed interfaces, data movement, and likely abuse paths. AI Security Review focuses on implementation risk, helping developers inspect code for common vulnerability classes, unsafe assumptions, weak controls, and security gaps that may not be obvious during normal development.

💻 More Natural Coding Workflows, 🛡️ Safer Local Action Handling, 🔎 Improved Privacy and Context Protection and ⚡ Refined Chat and Tooling Experience are all part of this release.

This moves Cortex closer to a production-ready agentic coding and security workflow. It improves context quality, makes code edits easier to trust, reduces noisy UI behavior, and sets up a cleaner foundation for future capabilities such as richer file operations, git-aware workflows, deeper security analysis, and higher-level implementation plans.

May 14, 2026

Cortex 4.1

⚡ Real Time AI Interaction for Modern Development Teams

Cortex 4.1 continues the evolution of Cortex into a more responsive and enterprise ready AI workspace designed for real world development and security workflows. Building on the streamlined workspace experience introduced in Cortex 3.8 and the enterprise AI control layer capabilities introduced in Cortex 4.0, this release focuses heavily on speed, reliability, usability, and operational maturity across VS Code environments.

A major highlight of Cortex 4.1 is a significantly more interactive AI experience. Developers can now engage with Cortex in real time through a smoother and more dynamic workflow experience that reduces friction during coding, troubleshooting, security analysis, and AI assisted development tasks.

The overall workspace experience has also been refined to better support long running conversations, larger development workflows, and continuous interaction with AI systems across enterprise environments.

May 11, 2026

Cortex 4.0

🚀 A brand new communication protocol and visually pleasing redesign

We have made Cortex 4.0 to be 2.5x faster in several coding tasks, with the overall improvement being 1.5x faster for all the agentic workflows we measured. We have ensured that Cortex can go from an exponential increase in response time to a logarithmic scale.
In addition, we are offering a completely overhauled and redesigned user interface in browsers – Google Chrome, Microsoft Edge and Mozilla Firefox and the most widely used Microsoft VS Code platform.

As complexity increases, we noticed that in certain cases, Cortex’s response took much longer than expected and we have addressed those with a brand new communication protocol. Diligently we captured all the inputs/outputs, and analyzed the cases where slower responses were observed.

We wanted to make Cortex more interactive. ✨ A brand new design emerged! With Cortex 4.0, the end to end communication protocol has been completely overhauled to make it more immersive, responsive and interactive. You get instantaneous feedback for your queries with suggestions for improvements – all enhanced by AI.

In addition, both the browser and VS Code version of Cortex look identical. We have strived to make Cortex, the most beautiful Coding, Security and AI Agent yet! ✨ With an eye for design elements, we have ensured Cortex’s intelligence is augmented with enhanced visual elements that make it stand out from the rest of the pack.

New design elements include:
🎯 A vastly redesigned chat interface
⚡ Simplified workflows
🧭 More intuitive context menus
🌈 Color combination of various responses
🛠️ Easy to use coding and security outputs and patches
📋 Copy, paste and edit capabilities
🎙️ Voice input and output modalities
🔗 Enhanced MCP interactions

May 4, 2026

Cortex 3.8

Cortex has evolved into a more focused and streamlined AI workspace inside VS Code. The experience is now centered around a cleaner Cortex identity, a dedicated chat surface, and a more intuitive interaction model that helps developers stay in flow while working across code, security, and AI-assisted tasks.

In this release, we have overhauled UI interface that moves Cortex closer to being a dependable AI workbench for secure software development. Cortex 3.8 modifies all of the visual elements to make it user friendly and adds native theme support for visual appeal.

A major theme of the release is productivity. The chat experience is easier to use, easier to scan, and better suited for real development workflows. Users can manage context more deliberately, interact with responses more efficiently, and access key actions without leaving the conversation. This reduces friction and helps teams move faster from question → insight → action 🚀

The chat interface received a broad usability upgrade. Message bubbles were refined for better density and readability, with improved spacing, aligned controls, theme-aware styling, and cleaner menus. Common actions are now easier to discover through compact pull-out menus with aligned icons and labels, including security actions, chat management, and export options. Context-file handling was also improved so users can clearly choose between no context, pinned files, and current-editor-file workflows without surprising automatic behavior.

The update also improves collaboration and knowledge capture. Cortex conversations can now be preserved and shared more easily, making AI-assisted work more useful beyond a single session. MCP and provider authentication flows were made more robust. This is valuable for audits, handoffs, reviews, documentation, and team alignment, especially when AI outputs need to become part of broader engineering or security workflows.

Reliability and trust were also strengthened. Authentication and connected-service behavior is more predictable, with fewer unnecessary interruptions and clearer handling of access or usage limits. For enterprise users, this helps reduce confusion, avoids avoidable sign-in loops, and creates a smoother path for using multiple integrations safely.

April 2026

April 30, 2026

Cortex 3.7

We are now expanding the capabilities of Cortex to scan for personally identifiable information (PII) and sanitize web page context and attachments that are sent as context to the backend LLMs. This release includes three scanners all running within your local environment, so your data never leaves the confines of your development machine.

🛡️ The three scanners include:
⚡ Quick Scan to detect/redact sensitive data
🧠 Deep AI Privacy Scan to detect/redact PII
📁 File Sanitizer to ensure safety of input files

⚙️ Core Capabilities
🔍 Detect sensitive patterns like API keys, tokens, passwords, private keys, database URLs, email addresses, IP addresses and many other
🧠 Deep scan uses advanced AI-powered scanning to detect personal information in chat messages, attachment and context files
⚠️ Intelligently scans and then asks the user whether to cancel, redact, or send anyway when sensitive content is found
📄 File sanitizer can sanitize PDFs, images and other attachments for harmful or malicious content, PII, API keys and other sensitive data
🔄 Performs a two phase sanitization of content to identify harmful content and render the final output

April 27, 2026

Cortex 3.6

Multi-Model > Multi-Agent > Multi-Cloud > Multi-Modal with Cortex 3.6. 🚀 After abstracting complexities in models, agents, cloud environments, we are releasing the latest innovation in Cortex that can assist you even better with voice 🎤, image 🖼️, documents 📄, text 💬 or similar input modalities. You can dictate instructions and receive responses and listen to content seamlessly. 🎧

We’ve strived hard to make Cortex more accessible and user friendly with a common graphical user interface across 3 browsers 🌐 – Google Chrome, Microsoft Edge, Mozilla Firefox and 3 IDEs 💻 – Microsoft VSCode, VisualStudio and IntelliJ Jetbrains IDEA.

All operations such as signing in 🔐, chat request and responses 💬, running scans 🔍, viewing security reports 📊, coding workflows 👨‍💻, MCP connections 🔗 and configurations ⚙️ are all integrated intuitively across various environments.

This release also includes the performance improvements detailed here. To summarize the performance work: The optimization work delivered consistent, substantial gains across all 5 operations 💪, reducing average response time from ~530ms to ~82ms ⚡ — a roughly 6.5x improvement in latency 🚀.

The most impactful change transformed a previously slow operation into one of the fastest 🔥. These results directly reduce memory and latency pressure 📉, especially around high-frequency storage reads, which were previously identified as a recurring bottleneck.

April 20, 2026

Cortex 3.5

Cortex 3.5 🚀 supports Microsoft Azure, Google Cloud and AWS expanding Enterprise AI context to MAX-level in Cortex. We released Cortex 2.5 with Enterprise AI earlier in January with connections to GitHub, Atlassian, Slack, Azure DevOps, Microsoft 365, Google Workspace and Google search. Cortex 3.2, launched two weeks ago, added AWS support to Cortex.

This new release brings 40+ AI Agents, 12+ connectors and 300+ operational capabilities with MCP connections ✨. Cortex is entering a new era with secure connections to all your cloud services, in addition to the most popular code versioning ecosystems, Project Management and Developer Productivity suites.

🌐 Multicloud Support
With Cortex 3.2, we added popular cloud services from Amazon such as IAM, RDS, S3, CloudWatch, EC2, DynamoDB, VPC, along with limited support for CloudFront, SNS and AWS Lambda.

Cortex 3.5, now adds Google Cloud MCP with the capability to control and obtain context from:
– ⚙️ Compute & Containers: Compute Engine, GKE, Cloud Run, App Engine, and Cloud Functions.
– 📊 Data & Analytics: BigQuery, Dataflow, Dataproc, Pub/Sub, and Cloud Composer.
– 🤖 AI & Machine Learning: Vertex AI, Vertex AI Search, Cloud GPUs, and Dialogflow.
– 💾 Storage & Databases: Cloud Storage, Cloud SQL, AlloyDB, Spanner, Bigtable, and Firestore.
– 🔐 Networking & Security: VPC, Cloud Load Balancing, IAM, Secret Manager, and Cloud Armor.
– 📡 Operations & Management: Cloud Logging, Monitoring, Trace, Asset Inventory, and Resource Manager.
– 🧰 Developer Tools: Artifact Registry, Cloud Build, and Database Migration Service.

On the Microsoft Azure front, Cortex can control and obtain context from the most popular Azure services such as:
– Compute: Azure Virtual Machines, Azure Container Instances (ACI).
– Kubernetes: Azure Kubernetes Service (AKS).
– Serverless: Azure Functions, Azure App Service.
– Databases: Azure SQL Database, Cosmos DB, Azure Database for PostgreSQL.
– AI/ML: Azure OpenAI Service, Azure AI Search.
– Storage: Azure Blob Storage, Azure Files.
– Networking: Virtual Networks (VNet), Network Security Groups (NSG). ☁️

You can control these services, query and manage state seamlessly from the chat interface 💬. Critical enterprise context and data from all these services can be seamlessly ingested into your coding and security workflow to make them even more effective.

April 8, 2026

Cortex 3.2

🚀 Expanding Enterprise AI context in Cortex with AWS

With Cortex 3.2, you can connect to the popular cloud services from Amazon such as IAM, RDS, S3, CloudWatch, EC2, DynamoDB, VPC, along with limited support for CloudFront, SNS and AWS Lambda. ⚙️

You can control these services, query and manage state seamlessly from the chat interface. 💬 Critical enterprise context and data from all these services can be seamlessly ingested into your coding and security workflows to make them even more effective. 🔐

🤖 Agentic AI:
Cortex features powerful Agentic AI innovations, incorporating over ten enterprise connectors and more than thirty specialized AI agents to perform complex backend operations efficiently.

🔐 Secure Code Development:
Cortex can help you write production-ready, secure code by adhering to security best practices and industry standards. Cortex can analyze your code for potential vulnerabilities and suggest improvements.

🧪 Vulnerability Assessment:
Cortex can assess your code for security weaknesses and provide detailed explanations of security measures to mitigate risks.

🔗 Platform Integrations:
Cortex can assist with operations across a wide range of development and collaboration platforms, ensuring secure and efficient workflows. This includes:
☁️ Cloud Services: AWS
🔄 Version Control & CI/CD: GitHub, Azure DevOps
📋 Project Management & Documentation: Jira, Confluence
💬 Communication & Collaboration: Slack, Microsoft 365 (including Teams and Outlook), Google Workspace
🧠 Code Analysis and Modification:
Cortex can analyze code content, identify security issues, and perform secure code modifications.

Cortex is an enterprise-grade AI coding and security assistant designed to help organizations build, operate, and scale software securely. 🚀
It combines deep code intelligence with operational automation across the tools engineering and security teams use every day.

April 3, 2026

Cortex 3.1

📊 We’ve spent the last 10 months evaluating model responses for chat, coding and security queries and amassed a tremendous amount of learnings. Based on heuristic analysis and by utilizing tried and tested strategies in computing, we are making Cortex even better.

🛠️ Making Cortex more resilient, available in case of failures and adding backup methodologies to ensure users are always served with their queries was quite challenging. Since our architecture was built from the ground up to handle such cases, we were able to make updates to entire stack to make Cortex better.

🔐 We also ensured that Cortex is even more secure by performing Red-Teaming experiments and by adding stress tests. This strengthens Cortex and protects the software from threats posed by bad actors. At Pervaziv AI, we are a security first organization and these updates came naturally onto our roadmap.

🧩 Session management changes with capabilities to list, rename and delete sessions have been added. This functionality is plugged in end to end across VSCode, Browsers and the backend Pervaziv-LLM endpoints.

💻 Finally, the VSCode extension has been updated to improve context handling and incorporate all of the above capabilities seamlessly. We have also fixed minor issues and bugs in our frontend and middleware to make Cortex more user friendly.

March 2026

March 25, 2026

AI Code Review 2.1

Code Review 2.1 is here with an updated GitHub Action that performs a security scan of your entire repository 🔍 and provides results in the Actions and Security tabs, a new capability to create a Pull Request with AI generated code suggestions in addition to a detailed vulnerability and security analysis in Pervaziv AI’s Console.

This product is available for customers with Premium and above subscription. This release follows the AI Code Review 2.0 GitHub App, which went live on GitHub Marketplace on March 12, 2026 🗓️, keeping up with our incredible pace.

We have further reduced the friction of dealing with multiple platforms by bringing security scan reports, rankings, code suggestions, Pull Request creation and insights directly into GitHub 🔗. Users can still review the more comprehensive security report in DevSecOps console 💡.

Capabilities
🤖 AI Enhancements: Our security scans, reviews and code suggestions are enhanced with Gen AI and we provide you with the best possible outcome with all our intelligence built into Pervaziv-LLM. 
Get additional information on the 2.0 release in this blog and the 1.0 release here.

March 17, 2026

Cortex 3.0

We’re introducing Cortex 3.0 🚀 in 3 browsers – Google Chrome, Microsoft Edge and Mozilla Firefox, in addition to the 3 IDEs – Microsoft VSCode, Visual Studio and Jetbrains IntelliJ IDEA. We’ve launched Cortex in 6 marketplaces with one unified experience 🌐, making Cortex the world’s first cross-browser and cross-IDE Coding and Security AI Agent.

Usage ⚡
In Cortex chat, you can get answers for your coding and security related queries 💬, run scans 🔍 and connect to your Enterprise suites such GitHub, Atlassian, Slack, Azure DevOps, Microsoft 365, Google Workspace and search the web 🌐 for more information.
With Cortex 3.0, you get a familiar user interface that you are used to on the IDEs 🖥️. We’ve strived hard to make Cortex similar on all 6 platforms that we offer currently (3 browsers and 3 IDEs) 🔄.

Capabilities 🚀
Key capabilities of Cortex 3.0 includes:
🤖 Provides coding and security related answers enhanced by AI
💻 Write new code in 15+ programming languages
🔐 Scans files for security issues and provides a vulnerability report
🔎 Connects to Pervaziv AI DevSecOps console for full security review and AI generated code suggestions
🔗 Integrated with 6 MCP Agents – GitHub, Atlassian, Slack, Azure DevOps, Microsoft 365, Google Workspace
🌍 Performs web search for additional context and relevant answers
🧠 Retention options to Save/restore of chats so you can get historical context
🔒 Option to delete chat response storage to enhance privacy
🎯 Option to provide personalized responses you based on your browser content

Benefits 🌟
Cortex 3.0 offers significant additional benefits to complement the prior Cortex releases in IDEs:
🤖 AI Enhancements: Connects to Pervaziv-LLM to get you vetted answers
🔐 Integrated Security: First of its kind to offer security analysis in a chat interface
📊 Security Metrics: Vulnerability report includes several key security metrics for all the files your repository
🌐 User Context: You can send context right from any of your webpages to enrich code and security reports
🔑 Seamless Authentication: Login to you favorite enterprise services with authentication from trusted providers
🔒 Privacy: Login to your favorite enterprise services with authentication

March 12, 2026

AI Code Review 2.0 DevSecOps 2.8

AI Code Review 2.0 comes with a brand new GitHub Action that performs a security scan of your entire repository 🔍 and provides results in the Actions and Security tabs, plus a detailed report in Pervaziv AI’s Console.

This product is bundled with our DevSecOps suite version 2.8 📦 and is available for customers with Premium and above subscription. This release follows the AI Code Review 1.0 GitHub App, which went live on GitHub Marketplace on Feb 13, 2026 🗓️.

Since the last 1.0 release, we spent time understanding the Enterprise user’s pain points with security workflows 🛡️. We gathered analytics in our DevSecOps console 📊 and analyzed how developers and security analysts perform their everyday tasks.

With that insightful knowledge, we developed this release, packed with more features ⚙️. We have reduced the friction of dealing with multiple platforms even further by bringing security scan reports, rankings, and insights directly into GitHub 🔗. Users can still review the more comprehensive security report and get code suggestions in our DevSecOps console 💡.

Capabilities
Key capabilities of this AI Code Review 2.0 are briefly bulleted below.
⚙️ Automate code scanning from within GitHub
🛠️ Quick feedback on Actions page regarding the status of the run
🔐 Scans entire repository for security issues 
✨ Integrates with AI Code Review that fixes code and provides suggestions using Generative AI
📊 Connects to Pervaziv AI DevSecOps console for full security review and AI generated code suggestions

Benefits
AI Code Review 2.0 Action offers significant benefits to complement the 1.0 AI Review App:
💻 Code security: Scans your entire repository for security vulnerabilities on every push or on a schedule.
🛡️ Security Metrics: Vulnerability report includes several key security metrics for all the files your repository
🧩 Streamline Debugging: Automates debugging, reducing manual code reviews.
📚 Query Knowledge Bases: Access public and private knowledge bases seamlessly
🌍 Scalability: We can review large repositories and provide remediations for 15+ programming languages
🤖 AI Enhancements: Our security scans, reviews and code suggestions are enhanced with Gen AI and we provide you with the best possible outcome with all our intelligence built into Pervaziv-LLM

March 6, 2026

Cortex 2.7

🚀 Feature rich AI/ML release! Cortex 2.7 updates 6 Enterprise AI MCP Agents with huge performance gains, improved chat session management, updated context handling, and updated authentication for 2 MCP backend connectors.

🤖 Cortex 2.7: VSCode and Backend: We added support for enhanced session management so users can have more control over their chats and responses. 💬 Updated Atlassian MCP connections to make it cross functional and portable across platforms.

This release also includes improved context tracking across the entire repository. 📂 We include cross-file context in performing security scans and while generating code. This largely expands Cortex’s role as a full-fledged coding agent and brings it on par with other well-established coding agents.

AI Enhancements: We achieved substantial accuracy improvements across the production MCP agents, with the most significant gains reaching +61 percentage points for the Slack agent and +43 percentage points for the Google Workspace agent. 📈
The Slack agent demonstrated the most dramatic improvement, jumping from near-zero accuracy (2.34%) to 64.06%, a 2,637% relative increase. 🔥

Similarly, Google Workspace and GitHub agents showed substantial gains exceeding 200% relative improvement.
We have released numbers and charts for a number of Agent capabilities plotted with improvement percentage. 📊 This varied significantly across the six supported MCP agents. Notably, the MS365 agent, with 44 capabilities, was the only agent to show negative performance change (-1.04 percentage points).

February 2026

February 20, 2026

Cortex 2.6 and DevSecOps 2.7

We are releasing Cortex 2.6 with largely expanded context handling capabilities and improved backend MCP connectors. DevSecOps 2.7 adds new features for Developer Certification with scoring models, hierarchical organizational dashboard with visually pleasing leaderboard charts and tables.

🤖 Cortex 2.6: Context and Reasoning
Cortex can now take in multiple files for context and can analyze and fix those files – all at the same time. We started with a single file context, but as we progressed with our analysis, we realized that Pervaziv-LLM is capable of doing more. We can also create new code files with your queries. The patch handling has been improved with our custom formatting and markers to make it expandable across files.

We’ve expanded the context, code understanding and reasoning in our LLMs to provide you with customized security scan reports and remediations. The reasoning mode is much more intelligent in its responses and is capable of providing you with very complex code suggestions. We’re getting closer to making Cortex a full-fledged coding agent. 🚀

We’ve improved the Slack connector to make it more reliable and cross-platform. We’ve made it more secure by moving the authentication to a secure server from the client-side, so you get the added benefit of being more secure. 🔐✨

🛡️ DevSecOps 2.7: Developer Certification
We released Developer Certification in version 2.5 and now we’ve added several new features to it to benefit developers, administrators, managers and the organization as a whole. This release includes 📈 leaderboard charts, KPI summary cards, distribution charts, risk quadrants and more.

With our Enterprise context, we realized how difficult it is to do a quantitative and analytical performance analysis of your developers. So we added KPI cards to capture developer scores based on their commits, Pull Requests and on the security side – how many insecure commits they’ve made (or not). Our custom AI-based score, then ranks them in a leaderboard and sorts them by their scores. Let the scoring and games begin! 🎮

February 13, 2026

AI Code Review 1.0 DevSecOps 2.6

🛡️ DevSecOps 2.6: Launching AI Code Review on GitHub
We are thrilled to announce that our Pervaziv AI Code Review is officially live on the GitHub Marketplace! 🚀 We can perform security and AI analysis of your code, all within GitHub. AI Code Review is designed to improve developer productivity and simplify workflows by seamlessly integrating advanced AI capabilities into your code review process.

This is our 8th product offering and is bundled under our DevSecOps product. This powerful new tool is included for free with the Pervaziv AI’s Premium and Enterprise Subscriptions! 🎉

Key capabilities of this product are briefly bulleted below.
⚙️ Automate code review
🛠️ Real-time debugging assistance
🔐 Scan code for security issues 
✨ Generate code fixes using Gen AI
📊 Adds an “Overall Risk” label to the PR

AI Code Review offers significant benefits to enhance your software development lifecycle. Again, briefly:
💻 Code quality: Provides vulnerability report for files changed in the Pull Request.
🧩 Streamline Debugging: Automates debugging, reducing manual code reviews.
📚 Query Knowledge Bases: Access public and private knowledge bases seamlessly
🌍 Scalability: We can review large PRs and provide remediations for 15+ programming languages
🤖 AI Enhancements: Our review and code suggestions are enhanced with Gen AI and we provide you with the best possible outcome with all our intelligence built into Pervaziv-LLM
🛡️ Security Metrics: Vulnerability report includes several key security metrics for all the files changed in your Pull Request

Subscription-Based Access: Available for Premium and above subscriptions.

January 2026

January 28, 2026

Cortex 2.5

🤖 Cortex 2.5: Reimagine Enterprise AI
Welcome to the future of Enterprise AI! At Pervaziv AI, we’re excited to introduce Cortex 2.5, reimagining how you approach your everyday tasks in an AI native workplace. Cortex 2.5 immensely expands its capabilities to seamlessly integrate with your entire enterprise stack, making it an indispensable asset for your organization.

We started with the tools we use extensively – GitHub, Atlassian and Slack. We set out to make your experience better with the Pervaziv AI partners by previously releasing Cortex 1.5.

And now we are introducing Cortex 2.5 that adds significant new enterprise connectivity to further enhance your AI capabilities.

Cortex 2.5 connects with GitHub, Atlassian, Slack, Azure DevOps, Microsoft 365, Google Workspace, websearch and more! In addition to being a Security and Coding agent, Cortex 2.5 can now manage operations across the gamut of your enterprise stack.

– With Azure DevOps, Cortex can manage projects, work items, teams, pipelines, repositories, builds, test plans, wikis, iterations, list advance security alerts and commits.
– With Microsoft 365, Cortex can interact with Calendar, Outlook, OneDrive, Excel, Contacts, Teams and Sharepoint.
– With Google Workspace, Cortex can interact with Google Drive, Calendar, Gmail, Slides, Sheets and Documents.

We are revolutionizing Enterprise AI by improving Cortex’s ability to connect to a broad range of enterprise suites, giving you unparalleled visibility and control. You can connect to these suites securely with your credentials and ingest context directly into your coding and security workflows.

Give your Enterprise AI the massive boost it needs with Cortex 2.5. You can install Cortex in 3 Marketplaces – Microsoft VSCode, Microsoft Visual Studio and IntelliJ Jetbrains marketplace. We can’t wait to see how you innovate with Cortex.

January 27, 2026

Cortex 2.2

🤖 Cortex 2.2: We’re adding personalization, memory features, additional privacy and retention to Cortex 2.2. From initial release to memory, Cortex received Personalization feature within 162 days 🚀 fastest of all Coding Agents. 

We now support long-term and cross-chat memory to deliver more personalized responses. Users can choose whether Cortex may personalize replies using insights from previous conversations.
A new option allows users to clear all stored memory at any time. By disabling save chat and personalization, a chat is logically in Incognito mode, ensuring conversations are not saved on server or used for future personalization.

🧪 Backend: Storage management to manage user chats and personalization. Premium and above subscription changes for supporting personalization.

We announced Cortex in August of 2025 and for the next few months, we have constantly added new features and innovations. With the latest Memory Feature added today, we have exceeded the pace of other well known coding agents.

January 22, 2026

Cortex 2.1

🤖 Cortex 2.1: Users can now choose to store Cortex chat history either locally or on our secure remote server.
If chosen to store on the secure remote server, Cortex chat history persists across sessions. With these changes, average latency for responses improved by ~1.6x. Users can also notice a significant improvement in quality and stability.
We have added several UI changes to Extension Settings.

January 16, 2026

Cortex 2.0

Capturing IDEs! Cortex 2.0 can now work with VSCode, Visual Studio and IntelliJ environments. We’ve made Cortex even more accessible with the launch of our IntelliJ IDEA plugin. After analyzing the IDE usage share, we set out to conquer the most popular ones. Cortex can be installed from the three respective Marketplaces today!

We have assimilated the share of Developer IDEs used today and ranked the IDEs by their normalized share as a percentage. We embarked on a journey to support 3 out of the top 5 most popular ones. This gives us a whopping 72% Developer Market Share with all three IDEs!

🤖 Cortex 2.0: Cortex has a similar look and feel in all three environments. We strived hard to make it seamless for our users to switch between platforms if they choose to do so. There are subtle differences in the usage based on the native support that was available for us from these IDEs. Authentication and authorization in IntelliJ is similar to VSCode and Visual Studio plugins and is integrated with our DevSecOps suite. Users can chat with Pervaziv-LLM and get their coding and security questions answered. They can also scan their files for vulnerabilities and get suggested fixes from our AI agents. UI frontend seamlessly integrates with native IntelliJ features such as command handling, file handling and session management. DevOps and release infrastructure is designed similar to our other apps hosted in marketplaces.

Users still get the same availability, reliability and security from Pervaziv-Security 🛡️ and Pervaziv-Backend. Users also get the best of Pervaziv-LLM 🤖 in all three IDEs.

January 12, 2026

DevSecOps 2.5

Advanced Developer Certification now available with 🛡️ DevSecOps Suite 2.5! Rank all your developers to get a Certification Score.

Firstly, we used our tools internally to track Weekly lines of code added per repository, Weekly Commits across all repositories, Weekly Pull Requests across all repositories and other low level metrics. We can then run these metrics across a time horizon – last 7-30 days, last 3-12 months, or up to last 1-2 years.

Secondly, when users scan a project, we capture scores such as EPSS, CVSS and assign a severity metric to each of the security issues found. We then associate the code with these corresponding vulnerability statistics. In addition to this mapping, we track users who introduced the vulnerability in the code and assign a custom ranking to the user based on their security practices. We aggregate this score across all their repositories and assign them an Advanced Developer Certification. In addition, we can also show how many High, Medium and Low vulnerabilities the user has introduced across all their code repositories.

Thirdly, if you subscribe to our DevSecOps as an Organizational user (instead of an individual user), you get additional benefits. We have added a hierarchical Organizational Structure view for users, their projects and project dependencies. When a user with the role of an Organization Admin logs in, they are able to view the Advanced Developer Certification for all the developers under their organization, sort them by their scores and view other low level details. This is a key feature that helps your Organization and Developers adhere to security best practices. Over time our reports will help improve the developer’s certification score and in turn improve the security of their projects, their organization and their company as a whole.

January 8, 2026

Cortex 1.7

Cortex 🤖 is Better, Safer, Faster, Easier with the 1.7 release! 🚀 We are also introducing 🎯 tiered responses from Cortex and Pervaziv-LLM based on your subscription package. Install from VSCode Marketplace.

We released Cortex 1.5 with Model Context Protocol (MCP) on November 10, 2025. 📣 Since then we have captured feedback from users in the field and analyzed all of the improvements that can be made.

💡 With this valuable feedback, we are bringing you the latest Cortex release so you can be better at coding and security workflows, be safer from threats and malware, get responses faster than ever before, and enjoy an easier way to connect with your backend applications such as GiHub, Slack and Atlassian/Jira 🔗.

⚡ We have introduced a tiered response framework — Trial and Base subscriptions get you the responses in a reasonable timeframe with acceptable accuracy from the base Pervaziv-LLM. If you upgrade to our Premium and Enterprise packages, you will be 🤯 blown away with the speed and accuracy of our responses from Pervaziv-LLM. Give it a shot! 🚀

🛠️ We have also squashed several bugs, stabilized the chat API, made modifications to the UI in the chat panel, and updated the Code diff window output formats with changes to make it more user friendly ✨.

January 7, 2026

DevSecOps 2.4

Binaries Demystified! DevSecOps 2.4 can now scan, unpack and analyze Windows packages/binaries (in addition to previously released Unix ones) for security issues and vulnerabilities. We’ve covered the entire gamut of what’s feasible to reconstruct with the help of AI.

Previously in DevSecOps 2.2, we released the Package Analyzer with support for UNIX-based packages and binaries. With this latest release, we have added support for Windows binaries. This is a key breakthrough since Windows binaries are blackboxes for the most part and unpacking and reconstructing were largely manual and time taking until this point without our software. We have automated and added AI enhancements to the life cycle of Package Analysis and we can save many hours of tedious manual processes that are generally error prone to a large extent. 

Error handling and cleanup operations for the package analyzer backend are updated with additional optimizations and increased maintainability.

We have made AI risk score calculation updates to make the score more accurate and have an even spread across the broad spectrum of vulnerability classes that we have tested.

Updated MCP connection support in our backend with several minor fixes to increase the reliability of our responses and ensure the queries are handled seamlessly.

December 2025

December 3, 2025

DevSecOps 2.3

We’re introducing Pervaziv AI’s Risk Assessment augmented with Application Security Posture Management (ASPM) 🔐. We have enhanced software audits and assessments with AI 🤖.

Pervaziv AI’s solution pairs Risk Assessment with ASPM. This system unifies findings from all our security tools, including SAST, SCA, Software Bill of Materials, and Dependencies. AI correlates data to surface the highest-impact risks instantly. Continuous posture monitoring replaces time-consuming manual spreadsheet analysis and tedious tracking and summarization workflows 📊.

The outcome is AI-Driven assessment that reduces false positives and provides clear remediations. You get a holistic view of your security posture with risk progression over time 📈.

November 2025

November 24, 2025

Cortex 1.6

Capturing Developer Tools! We have expanded Cortex to work with Visual Studio in addition to VSCode. This is the first and only Security and AI coding agent in the Marketplace! Developers can install Cortex in Visual Studio 2022 starting today. ✨

🤖 Cortex 1.6: Developed the Visual Studio extension from the ground-up to be on parity with our VSCode product. Cortex chat interacts with Pervaziv-LLM to get you validated responses for your coding and security related questions. Cortex can run security scans, identify vulnerabilities, provide remediations and prioritize issues with Pervaziv AI’s categorizations. Cortex provides deeper insights into security issues with CWE, CVE, priority, and other metrics, along with a hierarchical view of issues sorted by priority. 

Cortex manages authentication and authorization, and users can save and restore chat sessions securely, request new code, or modify existing code from the chat interface. It delivers structured and unstructured outputs from our smart LLM backend and lets developers integrate suggestions seamlessly into their codebase. 🚀

🛡️ DevSecOps 2.2: We have enhanced our DevSecOps suite to support both frontend tools — VSCode and Visual Studio. Our architecture was designed to be extensible, allowing us to add additional tools with minimal effort. 🔧

🧪 Dev/Backend: Effortless authentication for users to log in with supported providers — GitHub, Okta, Azure EntraID, Google, Bitbucket, Cognito, Gitlab. We’ve developed build and release infrastructure that is extensible for deploying new features across both Marketplaces. ⚙️

🤖 Pervaziv-LLM: No change needed!
🛡️ Pervaziv-Security/DB: No change needed!

November 17, 2025

DevSecOps 2.2

No source code needed! We are thrilled to announce the launch of our brand new Package Analyzer ✨, a powerful tool designed to redefine how you handle software vulnerabilities. With this, we are also releasing our DevSecOps Suite 2.2 🚀.

🛡️ DevSecOps 2.2: Added custom AI scoring prioritized by Pervaziv AI, incorporating scores such as EPSS and CVSS published by NIST. This feature provides comprehensive scans and detailed reports to help you identify and understand potential threats. We can seamlessly scan Ubuntu, Debian, G.N.U and RPM based packages along with the embedded binaries within them 📦. We unpack, reconstruct binaries and provide a comprehensive analysis of the issues we find in these packages. We developed two new pages, one to add and list the users packages and another to view the scan reports of individual packages.

November 10, 2025

Cortex 1.5 and DevSecOps 2.1.2

Future Begins Now! Cortex integrates MCP Agents for Github, Slack and Atlassian directly into the chat interface. Imagine the possibilities with smart insights from your Enterprise stack aiding coding and security workflows.

🤖 Cortex 1.5: Added dedicated MCP Agents for Github, Slack and Atlassian as our first set of integrations. These are powerful tools that can bring in additional insights into your coding and security workflows. Developed seamless authentication for MCP Agents with one-click addition of context to chat. Improved presentation of structured and unstructured responses from various backend tools in agentic responses.

🛡️ DevSecOps 2.1.2: New search feature to search across pages, projects and their respective available reports. Support for custom callbacks for agentic workflows. Analytics and metrics for users at individual, organizational and company level have been added. In addition, usability and security insights are also being tracked on a timely basis. Enhancements to registration database, error handling and bug fixes have been committed. 

🤖 Pervaziv-LLM:  Updated to latest, cost effective LLMs for backend agents. Stabilization of error code handling and cost calculations for workflows. Bug fixes in metadata storage in DB and cost tracking improvements have been made.

🛡️ Pervaziv-Security/DB: Standardized error handling across repositories with ability to improve reliability. Appropriate cost tracking for various scans – security, deep and ML based scans have been implemented based on resource usage and time spent running the workloads.

🧪 Dev/Backend: Added support to dynamically capture platform and user analytics on a regular basis. Binary Analysis is integrated closely with DevSecOps suite with a new improved page where users can initiate new analysis and view the reports once the process completes. Speed up end to end response times for analysis workflows.

October 2025

October 21, 2025

DevSecOps 2.1.1 and Cortex 1.2

Stats Overload! 1 Million Lines of Code, 1.5 Million YouTube Views, 35000 watch hours, 200000 visits/wk, 3700 Commits, 800 PRs, 500 Jira Stories, 30 Sprints, 50000 LinkedIn impressions, 2000 followers, 21 contributors. We’re only getting started!

DevSecOps 2.1.1 and Cortex 1.2.0 are now available! We have updated both products with Agentic workflows with feedback from a large swath of users.

🤖 Cortex 1.2:  With improved Agentic updates, users can get better answers. Added the ability to apply all suggested code fixes together in a file. Made several performance improvements and bug fixes. Slightly improved login experience.

🛡️ DevSecOps 2.1.1: Added enhancements to the user approval process. Additional resource metrics for usage are being tracked. Refactored backend to be more modular and improved asynchronous response times.

🤖 Pervaziv-LLM: Stabilized our Agentic LLM prompts to make them more reliable. Updated the backend LLMs to newer, leaner set of models. Updated our cost calculation algorithm to include additional details.

🛡️ Pervaziv-Security/DB: Updated scripts for accurate Git Analytics.

🧪 Dev/Backend: Updates to scalability testing to support agentic workflows from various sources. Updated Binary Analyzer to parse a larger set of binaries. Updated backend LLM to perform faster and be more cost effective.

October 8, 2025

DevSecOps 2.1 and Cortex 1.1

🛡️ DevSecOps 2.1: Our Console now has a page to support Binary Analysis. Added enhanced versions of Code Editors to make it seamless and user friendly for DevOps and Deep Scan pages. Made CWE/CVE references more accessible to users.

🤖 Cortex 1.1:  Added additional providers and connectors to provide relevant answers to user’s queries. Option to scan large repositories now available in Premium and above subscriptions. Fixed a few bugs and added validations for input text.

🤖 Pervaziv-LLM: Scalability and connection reliability have been improved. Feasibility and comparisons were done with additional LLMs w.r.t their performance and cost (Report to be available soon!). Modified responses for code and chat related queries with optimizations on formatting/length and other usability parameters.

🛡️ Pervaziv-Security/DB: Enhanced Binary Analysis now available in Beta for Premium and above subscriptions. We use our innovative AI/ML approaches to look for vulnerabilities in binaries, no source code necessary! This is a feature some of our design partners asked for and we have delivered!

🧪 Dev/Backend: We have performed scalability testing with over 500 concurrent requests and users. We have a dedicated backend and test repositories to perform unit tests, end to end tests and scalability testing. More on these updates and a brand new scalable architecture to be announced in an upcoming release.

September 2025

September 10, 2025

DevSecOps 2.0.1 and Cortex 1.0.7

🤖 Cortex: Integrated login and authentication from VSCode to our backend with active session management. Provided several ways to interact with the product – to chat, run scans, obtain code suggestions and apply fixes. Cortex chat works seamlessly with VSCode workflows. Users can now provide context in addition to their text chat for queries to Cortex.

🖥️ DevSecOps: Seamless authentication/authorization of users for backend providers along with security fixes. Revamped our Registration page with several updates. Provided users more control of scan and code remediation progress. Optimized vulnerability scanning and reporting.
 
🤖 Pervaziv-LLM: Revamped the secure LLM hosting infrastructure with several security and performance improvements. Improved the interoperability with a larger set of LLMs. Pervaziv-LLM can seamlessly handle chat and code remediation requests. Made several improvements to track usage and billing. Embedding updates have been made to improve our LLM responses.

🛡️ Pervaziv-Security: Fixed vulnerabilities and bugs. We are proactive in testing our scans on our own product!! Enhanced vulnerability coverage by including Infrastructure as Code (IaC) scan into SCA reports. Tracking of scans is integrated tightly with our product.

🧪 Dev Branches: We have a breakthrough in binary scanning and no-source code analysis using our innovative agentic LLM approach. We are able to reconstruct select binaries for vulnerability assessment and we plan to iterate and integrate this approach with our product.

August 2025

August 18, 2025

DevSecOps 2.0 and Cortex 1.0

Introducing our smart AI coding agent and our DevSecOps Platform! At Pervaziv AI, we’re thrilled to introduce two powerful new products. Cortex is your intelligent coding and security AI Agent, that can help you write new code, scan and remediate issues in existing code expeditiously. Cortex brings security and code insights directly into your favorite development environment.

Say Hi to #Cortex, our smart AI coding agent and our #DevSecOps Platform! At Pervaziv AI, we’re thrilled to introduce two powerful new #products.

#Cortex is your intelligent coding and security AI #Agent, that can help you write new code, scan and remediate issues in existing code expeditiously. #Cortex brings security and code insights directly into your favorite development environment.

Next, we’re introducing our Unified #DevSecOps AI Platform, designed to give your entire team centralized visibility and control over your security posture.

July 2025

July 14, 2025

DevSecOps 1.0.7

🔐 Pervaziv-ASPM: Includes new security snapshots, overall risk management with advanced AI-based risk filtering and classification, commit-based risk progression, dependency risk progression, and support for report download. Users can now personalize results by tagging vulnerabilities and remediations with specific actions and filtering based on those tags. We are enhancing all of the reports with AI and this differentiates us from other incumbent solutions.

🤖 Pervaziv-LLM: We have made enhancements to RAG embeddings to improve code inferencing, migrated to a faster embedding model provider with optimal storage, improved context provisioning with faster retrieval times and moved to newer versions of LLM libraries.

🛡️ Pervaziv-Security: Improved protections and added additional management options for product configurations and subscriptions. Speed and efficiency of report fetches have been improved. Several workflow enhancements have been added to manage production and developer modes of the product. Expanded vulnerability coverage with deeper insights into package vulnerabilities for projects using JavaScript package managers. Enabled optimal storage of reports for larger repositories.

🖥️ Pervaziv-UI: Updates include Code Editor enhancements to make it user-friendly, support additional editing functions, track symbol references etc. Project and report search functionality have been added with additional summarization views, dynamic adjustment of window sizes and content.

🔐 Pervaziv-Auth/DB: Azure-Repos and Gitlab projects are now supported in addition to Github and Bitbucket. Users can authenticate and add projects hosted in all of these repositories. This opens up our product to a prospective list of new customers 🌍 and enables them to integrate their DevOps workflows. Enterprise updates have been made to manage team-based subscriptions and payments. Introduced several new roles as part of the enterprise suite which allows for customized login processes, access controls, and visibility scopes.

🌐 External Contributions: Our patches for Meta’s Purplellama benchmark have been merged successfully and we look forward to further collaboration with several other open source projects.

🧪 Dev Branches: Our internal developer branches have received several contributions from our exceptional employees and interns for a well planned list of upcoming features that are on our roadmap.

May 2025

May 28, 2025

DevSecOps 1.0.6

Several Updates to – LLM 🤖, Security 🔐, Authentication 🔑, DB 🗄️, UI 🎨, Backend ⚙️, External Demos 🎬 and Open Source Contributions 🌐.

The latest release includes several enhancements to pervaziv-llm, to make it more secure 🔐, more relevant 📌, with several validations for various operations ✅. We have added the capability to authenticate with additional providers such as Amazon Cognito ☁️, Atlassian Bitbucket 🧰 and Okta 🛡️. In addition to adding GitHub projects 🧑‍💻, users can now add Bitbucket projects, which was a big ask from several of our enterprise partners 💼.

We have made payment processing 💳 even more seamless and integrated with our product and subscription packages. Several user experience ✨ and performance 🚀 enhancements were made to our UI, SCA, and SBOM feature set. Across the board, we have enhanced the security 🔒 of our product by ensuring the entire stack is free of vulnerabilities 🛠️.

We have contributed back to the Open Source community 🌍 with our submissions to PurpleLlama benchmark. The changes include support for benchmarking the base and fine-tuned Gemini models 🧪. Our Q2 All-Hands meeting also included an exciting end-to-end demo 🎥 for our customers in our medical vertical 🏥.

This is a shortened version of the features that went into the latest release and only work done in the last 6 weeks.

April 2025

April 7, 2025

DevSecOps 1.0.5

Retrieval Augmented Generation (RAG) 🤖, Software Component Analysis (SCA) 🧑‍💻 and Software Bill of Materials (SBOM) support 📜. We have made significant progress as a team and delivered on both AI/ML 🤖 and the Software Security 🔒 domains.

SCA and SBOM features that were scheduled for later in our product lifecycle to be released in the current 1.0 v5 Release.

Customers are now able to perform SCA related tasks along with vulnerability detection immediately 🔍.
On the ML side of things, we went through several iterations and trials of LLM Optimizations.

We added that capability to provide a cohesive view of vulnerability detection and perform SCA and SBOM analysis seamlessly 🔎. Our SCA, true to our nature, is AI-first, enhanced with our latest ML-based innovations to reduce false positives ❌ and false negatives ❌.

In addition, we researched what industry-specific reports are being used for SBOM generation and incorporated Software Package Data Exchanges (SPDX) and CycloneDX formats for SBOM analysis and reports 📊. These standard formats will be familiar for customers who have used legacy SBOM tools and will find the AI enhancements quite valuable. Learn more about SPDX here and CycloneDX here.

March 2025

March 5, 2025

DevSecOps 1.0.4

We are also releasing the Pervaziv AI v4 version which now supports Amazon Web Services (AWS) along with Google Cloud and Microsoft Azure.

No two LLMs behave the same when it comes to APIs, tuning parameters, token sizes, accuracy 🎯, F1 scores 📊, precision, recall, and most importantly their costs 💰. With our extensive research over the past year 📅, we have established a solid baseline and knowledge of the LLM landscape 🌍. We also understand that this is a fast-changing and evolving industry 🔄 and we are poised to keep abreast of the latest updates to all of the ground-breaking models 🔬.

Cloud computing has been around for a while, but is incredibly complex to set up and to migrate applications across clouds 🌐. Our customers see value in the multi-cloud aspect of our software stack 🌍. On the other hand, some customers specifically asked for AWS support and we delivered on that promise ✅. We can now truly claim the multi-cloud aspect of Pervaziv AI that we set out to build initially.

On the enterprise side, we have integrations with various software products such as Github 🐙, Stripe 💳, WordPress 🌐, Mitre 🛡️, and many others. We believe these integrations are crucial and build out our startup ecosystem 🏢. As we mature our software stack, we intend to build upon additional integrations to support wider use-cases.

February 2025

February 11, 2025

DevSecOps 1.0.3

📈In terms of code contributions, we’ve achieved a monumental milestone with approximately 200,000 lines of code written! 😱 This is a massive effort from every single person on the team.

We’ve committed 1783 individual commits – each one improving the functionality, security, or usability of the product. Additionally, we’ve merged 365 pull requests into the mainline codebase, all of which went through thorough, professional code reviews to ensure the highest level of quality. This process has allowed us to maintain a well-organized and secure codebase that’s scalable and efficient. 🔥

November 2024

November 21, 2024

DevSecOps 1.0.2

This release has a gamut of updates – Security Enhancements, Database updates, Github/Cloud Integration updates, Subscription Management, Workflow automations, AI/ML and LLM enhancements and several bug fixes. 

September 2024

September 19, 2024

DevSecOps 1.0.1

We are excited to announce the availability of the Beta version of Pervaziv AI 1.0. This release provides over 25 features to scan, remediate, build and deploy applications in a multi-cloud environment.

After training on several thousands of vulnerability samples, our database is considerably large. Hence we are able to train, isolate and mitigate security issues in applications efficiently (more on the runtime performance in a soon to be released blog). Our vulnerability detection accuracy is about 430% better than traditional models.

We have also made our inferencing much faster by using prompt optimizations. Our inferencing is up to 65% quicker than traditional models as shown in the figure below. We expect to improve our accuracy and performance soon.

November 2023 – September 2024

Pervaziv AI Beta

Super stoked to launch Pervaziv AI – https://pervaziv.com – where customers can build, deploy and run applications with exceptional security in a multi-cloud environment. Security hardening is provided by pretrained generative AI.

During this time, we built out the Minimum Viable Product (MVP) after working with several enterprise customers in Med-devices, Medical Research, Fintech, Banking, Enterprise AI Platform, AI/ML SoC startup and others in the USA.

We welcome you to sign up for a demo and request a trial of the beta product. Let us make software security a first priority for organizations and ensure that the world is secure – one line of code at a time!