The governed AI coding platform

    AI coding without guardrails is a liability.

    Every request your agents make flows through one enforcement layer. A real security perimeter that controls your AI stack.

    Try the coding agentSee enterprise governance
    Self-hosted VPCZero data egressNo training on your data
    Works with
    OpenCodeClaude CodeCursorCodexGSD2
    Proven at scale

    Trusted by engineering teams worldwide.

    Akamai
    BMW Group
    Cisco
    FICO
    Hugging Face
    Kakao
    PlayPlay
    Roku
    ShareChat
    Swisscom
    WeWork
    Akamai
    BMW Group
    Cisco
    FICO
    Hugging Face
    Kakao
    PlayPlay
    Roku
    ShareChat
    Swisscom
    WeWork

    "Kimchi is fundamentally changing how our engineering team thinks about AI-assisted development. The multi-model approach keeps the quality of a single frontier model while cutting token costs dramatically."

    DEKEL SHAVIT - SENIOR DIRECTOR OF ENGINEERING, AKAMAI

    [ GUARDRAILS ]

    Set the rules once.
    Kimchi enforces them everywhere.

    Every agent request passes through one enforcement layer for spending, data, deployment, and model routing.

    GOVERNANCE PERIMETER
    Incoming agent request
    refactor auth middleware
    Kimchi enforcement● 4 CONTROLS APPLIED
    BudgetAPPROVED

    $7,842 / $10,000

    Data boundaryENFORCED

    Egress blocked · EU VPC

    Training usageDISABLED

    Customer data excluded

    Model orchestrationAUTO-ROUTED

    Reasoning → execution

    ✓ Verified delivery
    PR READY · AUDITED

    BUDGET CONTROLS

    Stop overspend before it happens

    Set hard limits by user, team, API key, or organization. When the limit is reached, spending stops.

    ENFORCED BEFORE EXECUTION

    SELF-HOSTED VPC

    Your code stays in your perimeter

    Deploy in Kubernetes across AWS, GCP, Azure, or on-premises with zero model-data egress.

    DATA EGRESS BLOCKED

    DATA PRIVACY

    Your data is never training data

    Prompts, code, and outputs are excluded by architecture - not by an optional configuration.

    SERVERLESS + SELF-HOSTED

    MODEL ORCHESTRATION

    The right model for every task

    Automatically route planning, coding, and fast edits to specialized models based on quality and cost.

    50+ MODELS · ONE WORKFLOW
    Data sovereignty

    Your code. Your data. Your perimeter.

    Open-source models run inside your cloud account on AWS, GCP, or Azure. Prompts, completions, and code never leave your Cloud.  Nothing phones home. You decide what crosses the boundary; the default is nothing.

    • Self-hosted in your Kubernetes cluster: AWS, GCP, Azure, on-prem
    • Zero egress for model data. Control plane connectivity only.
    • 50+ open-weight models on your own GPUs
    • BYOK: bring your own OpenAI/Anthropic key, direct traffic, clearly surfaced
    No training policy

    We don't train on your data. Ever. In any mode.

    Nothing phones home. This isn't a configuration option. It's architectural.

    DEPLOYMENT MODES
    ServerlessKimchi GPUs (FR, IL, US)
    Self-hosted VPCZero egress
    Frontier BYOKYour key, direct
    Cost visibility

    Every dollar attributed. Every cap enforced.

    All AI coding spend lands in one governed line item - attributed per developer, per team, per model, in real time. Caps enforce at thresholds you set. A number you can defend to the CFO.

    • One governed line item instead of scattered per-seat subscriptions
    • Breakdown per-developer, per-team, per-model, per-tag attribution
    • Spend tracked against the caps you set, in real time
    app.kimchi.dev / cost-insights
    Cost insights - this month - $12,408 MTD - 312 developers
    BUDGET$25.0k50% used
    SESSIONS142,891+28% WoW
    AVG / DEV / MO$39.80vs $200+ per-seat
    BY DEPARTMENT
    product-eng$5,610142 devs
    platform-eng$3,94094 devs
    data-science$1,76038 devs
    sre-infra$1,09838 devs
    BY MODEL
    minimax-m3$7,57061% of spend
    kimi-k2.7$4,83839% of spend
    $ _
    Governance that enforces your policies

    Spend limits for total ease of mind.

    See exactly who, what, and where your AI spend is going. Per-developer, per-team, per-model, per-tag - in real time. Hard budget caps at every level: user, team, API key, org. Caps that enforce when you configure them, not alerts that just notify when it's too late.

    • Per API key set limits for CI/CD pipelines, bots, or contractors
    • Per user every developer has their own budget with real-time visibility
    • Per team finance gets a clean number; engineering keeps shipping
    • Per org hard ceiling across all keys, users, and teams so total spend never exceeds your number
    Budget controls
    APR 2026
    ORG • ACME INC80%
    $4,820 / $6,000 cap
    TEAM • PLATFORM ENG71%
    $284 / $400 limit
    Per developer
    ravi-sharma$38.20 / $50
    maria-t$29.40 / $50
    ci-pipeline-key$12.10 / $100
    ravi-sharma is at 76% of budget

    [ WHO IT'S FOR ]

    Built for speed. Governed by default.

    One governed AI coding platform, built around the way each team already works.

    PLATFORM ENGINEERING

    Run it your way

    Start serverless or deploy inside your VPC.

    • Open-weight models on your GPUs
    • One OpenAI-compatible API
    • Zero egress in self-hosted mode
    SECURITY TEAMS

    Govern every request

    Spending policy enforced before a request leaves your environment.

    • Zero training on your data
    • Enforced budget caps
    • Complete audit trail
    DEVELOPERS

    Keep your workflow

    Use Kimchi from the coding tools your team already relies on.

    • Claude Code, Cursor and VS Code
    • MCP servers and skills migrate
    • No workflow changes
    Enterprise controls backed by CAST AI's security program
    SOC 2 TYPE IIISO 27001PCI DSSGDPRTRUST CENTER →

    Start in 60 seconds. Govern from day one.

    Install the CLI, run /ferment, and hand off to the agent. No lock-in, no credit card.

    $ curl -fsSL https://github.com/getkimchi/kimchi/releases/latest/download/install.sh | bash
    Try the coding agentBook a demo
    50+ models, one platformHard budget caps that enforceOpen source CLI