SOVEREIGN CODING
Cursor-grade AI coding. Your source code never leaves your VPC.
Certified local models, governed through one gateway, signed on every completion. Give your developers a modern AI coding assistant without sending a single line of client code to someone else’s cloud.
SOC 2 · ISO 27001 · Aligned with all 26 RBI FREE-AI recommendations · Trust Receipt on every call
THE PROBLEM
Your developers want AI. Your clients' contracts say no.
Every IT services firm, BPO, and regulated team hits the same wall. The productivity case for AI coding is settled — but hosted assistants route source code through clouds you don't control, in regions you didn't choose. For a team under client NDAs, data-residency clauses, or RBI and DPDP obligations, that isn't a feature trade-off. It's a breach waiting for an audit.
So the work goes one of two ways. It happens in the shadows — developers pasting client code into personal accounts — or it doesn't happen at all. Neither is a position you can defend to a client's security team.
HOW IT WORKS
Local models. One governed gateway. A receipt for every line.
Sovereign Coding isn't a new model and it isn't a new IDE. It's the governance layer between the two — assembled from the same control plane that already runs regulated agents in production.
CERTIFIED MODELS
Pick from a governed catalog of local coding models in Model Hub — each one carrying provenance, evaluation history, and license posture. A certified asset, not a raw download off the internet.
GOVERNED GATEWAY
Every completion routes through one OpenAI-compatible endpoint. Policy is enforced at the point of the call — what context can leave a repo, what gets redacted, what's logged. Routing is policy-aware: cheap autocomplete stays on a local card, a heavy agentic task routes to a bigger node — right model, right GPU, right governance.
SOVEREIGN DEPLOYMENT
Models and inference run on your infrastructure, in your jurisdiction, under your keys. Nothing crosses a border you didn't draw.
TRUST RECEIPT
Every completion is signed — model, region, policy decisions, redactions — in the open AgentBOM format. Hand a client's auditor cryptographic proof, not a verbal assurance.
YOUR EDITOR, UNCHANGED
Your developers keep their editor. You get the control plane.
The gateway speaks the OpenAI-compatible protocol your developers’ tools already expect. Point Continue, Cline, or Zed at your sovereign endpoint and they run entirely against models inside your perimeter — no context leaves the building. Cursor is supported via custom endpoint; talk to us about keeping its workflow fully in-jurisdiction.
- Continue
- Cline
- Zed
- Cursor (custom endpoint)
BUILT FOR
For the teams who answer to someone else's auditor.
IT Services & SI partners
Deploy AI coding across your delivery bench without breaching a single client's source-control clause.
BPOs & GCCs
Give offshore teams modern tooling while every keystroke stays inside the contracted geography.
Regulated in-house teams
Banks, insurers, and health systems shipping their own software under RBI, DPDP, and HIPAA obligations.
THE CERTIFIED CATALOG
Start on one GPU. Scale when the work demands it.
We certify a focused set of open coding models and publish what diligence needs — eval results, provenance, license posture, policy bindings. We also tell you the truth most lists won't: what it costs to run. The catalog is tiered by GPU class, so you start where your hardware already is and scale only the heavy tasks.
Tier 1 · Runs on a single 24GB GPU — your everyday bench
Covers the vast majority of developer hours: autocomplete, edits, explanations, debugging.
- Qwen3.6 27BApache 2.0
the all-round workhorse, ~22GB VRAM.
- Codestral / Devstral SmallApache 2.0
fast fill-in-the-middle autocomplete on a single card.
- North Mini Code 1.0Apache 2.0
coding-specific repo, terminal, and review work.
- Qwen3.5 9BApache 2.0
for laptops and thin hardware.
Tier 2 · Single 80GB GPU — departmental
- Gemma 4 31BGemma license
multimodal: screenshots, UI bugs, diagrams. (License carries an acceptable-use policy — published, not hidden.)
- EXAONE 4.5 33Bverify terms
document-heavy, visual coding workflows.
Tier 3 · Multi-GPU node — owned or rented in-jurisdiction
For the heaviest long-horizon agentic work. Most teams burst to this by the hour, not buy it.
- GLM-5.2MIT
frontier agentic engineering, 8-hour autonomous runs (~744B MoE; needs a node).
- DeepSeek V4MIT
top LiveCodeBench / long-context tasks.
Provenance SKU (any tier)
- StarCoder2BigCode OpenRAIL
trained on openly published, license-filtered code. The pick for clients who need to audit the training corpus for IP risk.
Don’t see your model? Bring it — we certify customer-nominated models, including India-origin sovereign models, into the same governance contract.
WHY NOT JUST RUN IT YOURSELF
Anyone can download a model. Almost no one can prove what it did.
Ollama gets you a model on a laptop. It doesn’t get you a certification record, a policy plane, redaction, or a signed receipt your client’s security team will accept. Sovereign Coding is the difference between “we run AI locally” and “here is cryptographic proof of exactly which model touched your code, where, and under what policy.” That proof is the product.
FAQ
Frequently asked questions.
- Does Sovereign Coding send my source code to the cloud?
- No. Models and inference run on your own infrastructure, in your jurisdiction, under your keys. Source code, prompts, completions, repository context, and model weights never cross your VPC boundary. The only thing handed out is a signed Trust Receipt you give to an auditor.
- Which editors and IDEs does Sovereign Coding work with?
- The gateway speaks the OpenAI-compatible protocol your developers' tools already expect. Continue, Cline, and Zed point at your sovereign endpoint and run entirely against models inside your perimeter. Cursor is supported via a custom endpoint — talk to us about keeping its workflow fully in-jurisdiction.
- How is this different from just running Ollama or a local model myself?
- Downloading a model gets you inference on a laptop. It does not give you a certification record, a policy plane, redaction, or a signed receipt your client's security team will accept. Sovereign Coding is the governance layer between the model and the editor: cryptographic proof of exactly which model touched your code, where, and under what policy. That proof is the product.
- What is a Trust Receipt?
- Every completion is signed — model, region, policy decisions, redactions — in the open AgentBOM format. It is cryptographic evidence of what ran, rather than a verbal assurance, so you can hand a client's auditor proof on demand.
- What models are in the certified catalog, and what hardware do they need?
- The catalog is a focused, tiered shortlist of open coding models published with eval results, provenance, and license posture. Tier 1 runs on a single 24GB GPU and covers most developer hours; Tier 2 is a single 80GB GPU for departmental multimodal work; Tier 3 is a multi-GPU node for the heaviest agentic work, which most teams rent by the hour. You can also nominate your own model — including India-origin sovereign models — for certification.
- Does Sovereign Coding help with RBI, DPDP, HIPAA, or GDPR obligations?
- It is built for teams under data-residency and client-confidentiality constraints. Because inference stays inside your perimeter and every call is governed and signed, it supports deployments that must keep source code in-region. AgentAnywhere is SOC 2 and ISO 27001 certified, independently assessed for HIPAA and GDPR, and aligned with all 26 RBI FREE-AI recommendations. It is a control, not a substitute for your own compliance program.
- Is Sovereign Coding generally available?
- It is in design-partner release. A small group of services firms and regulated teams is being onboarded to shape the certified catalog and the integration before general availability. Design partners get priority model certification, hands-on deployment, and pilot pricing. Pilots are starting Q3 2026.
SOC 2 · ISO 27001 · Independently assessed for HIPAA & GDPR · Aligned with all 26 RBI FREE-AI recommendations · Trust Receipts (signed AgentBOM)
DESIGN-PARTNER PROGRAM
Pilots are onboarding now.
Sovereign Coding is in design-partner release. We’re onboarding a small group of services firms and regulated teams to shape the certified catalog and the integration before general availability. Design partners get priority model certification, hands-on deployment, and pilot pricing.
What we’ll ask
- Work email
- Company
- Role
- Developers on the bench
- Primary IDE
- GPU footprint (optional)
Pilots starting Q3 2026. We reply within two business days.
Give your developers AI. Give your clients proof.
Sovereign Coding is in design-partner release. Bring your bench and your toughest client clause; we'll scope a pilot against your stack.