Multi-Model Support
Switch between leading AI providers and local models depending on the task, latency, workflow, or environment.
Building a Unified Model of software Systems
Your key to build AI software
Unlimited functionality, ease access and full control over your data.
Trusted by emerging startups and industry leaders
Feature highlights
01
5 min Quick Start
Get started with Cortex in under five minutes
02
4 Install Methods
Install Cortex the way that fits your workflow best
03
8+ Providers
Cortex supports a broad range of model providers and local setups
04
6 Core Features
Built around the core capabilities needed for real coding workflows
02 / Use Case
Built for real software work
From early ideas to production-ready workflows, use agents where they save the most time.
01
Understand unfamiliar code
Ask Cortex to explain architecture, inspect files, and map how a project works before you make changes.
02
Fix, edit, and iterate faster
Use Cortex to add validation, review recent changes, diagnose failing tests, and keep moving without leaving the terminal.
03
Run automations in CI
Execute headless tasks for linting, test runs, and scripted workflows with configurable autonomy and tool controls.
04
Connect tools & external context
Extend Cortex with MCP servers to work with databases, APIs, GitHub, and local files through one consistent workflow.
05
Automate GitHub workflows
Set up automated PR reviews, issue triage, and repository workflows directly through GitHub Actions.
06
Shape agents for your workflow
Create custom agents with specific prompts, restrictions, and capabilities tailored to how your team builds.
03 / How it Work
Two ways to {build}
Build software with one unified AI workflow.
Build Without Friction
Work across leading providers
Switch between leading AI providers and local models depending on the task, latency, workflow, or environment.
Scroll to Explore
05 / Agents
Works with the way modern teams build
Connect tools, use agents with context, and move from idea to execution faster.
06 / Subscription
One subscription.
All agents included.
Separately
In total
≈$89/mo
Pay for each AI separately across different subscriptions. Costs stack up and workflows stay fragmented.
- Multiple subscriptions
- Higher monthly cost
- Separate limits and billing
- Harder team management
- Chat GPT20$/mo
- Jasper29$/mo
- Bing AI15$/mo
- Writesonic25$/mo
- And many others
All-in-one
In total
$20/mo
Save 69$Get all AI tools in one unified plan. Lower cost with a seamless, connected workflow.
- Access every agent in one place
- Build across WEB and IDE
- Use VIBE inside IDE
- Connect workflows with MCP
- Keep execution secure and controlled
- Continue work across sessions
- Automate repeatable tasks without extra tools
Build with agents
Start building now
What is Cortex, exactly?
Cortex CLI is an open-source AI coding agent that runs directly in the terminal with access to the file system, terminal, and development tools. Your product layer extends that workflow into WEB and IDE experiences.
What’s the difference between WEB and IDE?
WEB is the browser workspace for reviewing and orchestrating work, while IDE keeps Cortex close to local code editing and agent execution.
What is VIBE inside IDE?
VIBE is the in-IDE flow for steering agents, reviewing context, and iterating on implementation without leaving the editor.
Which models are supported?
The mocked marketing site presents Cortex as model-flexible, with exact provider wiring deferred because this frontend has no backend integrations.
How does security work?
Security behavior is represented as local product copy only; agent permissions, filesystem access, and integrations remain controlled by the real Cortex runtime outside this mock frontend.
What is MCP used for?
MCP is described as the extension layer for connecting agents to approved tools and data sources while keeping this site fully local and mocked.
Can this fit into CI/CD workflows?
Cortex copy supports automation-ready workflows, but this landing app does not create real CI integrations or external calls.
Do I pay per agent?
The mocked plans are organized around workflow tiers, not per-agent billing. Real billing is intentionally out of scope for this frontend.
08 / Changelog