Bring AI Under One Governed Gateway
Connect every AI provider — hosted or self-hosted — through one gateway. Give every team the model they need, with budgets, audit, and chargeback baked in.
One Gateway for Every Provider
Vendor neutral by design
Register OpenAI, Anthropic, Azure OpenAI, and any OpenAI-compatible endpoint — including open models you run yourself — behind a single gateway. Your teams pick the best model for the job without procurement or DevOps overhead each time.
- Hosted providers: OpenAI, Anthropic, Azure OpenAI, Mistral, DeepSeek, and more
- Open and on-prem models via any OpenAI-compatible endpoint (vLLM, Ollama, TGI, NIM)
- Private model tunnels — weights stay inside your environment
- Org-wide or team-scoped providers with dynamic model discovery
Real Cost Control, Not a Surprise Bill
Every token tied to a budget
API keys are bound to allocations, so spend is capped before it starts. Finance gets per-team, per-project, per-model chargeback across AI, interactive sessions, and HPC — all in one dashboard, no spreadsheets.
- Hard budget caps per key — no runaway invoices
- Cost dashboards by user, model, team, and allocation
- Token, latency, and provider data on every request
- AI rolls up to the same chargeback as compute and storage
Governance Your Security Team Already Trusts
SSO, audit, and policy in one control plane
AI access inherits the same identity and policy posture as the rest of your stack. Scoped service keys, per-team model enablement, and a full audit trail mean Security and Compliance sign off once — not every time a new tool shows up.
- SSO/OIDC identity through ACTIVATE
- Scoped keys with expiry, last-used tracking, and instant revoke
- Per-team model enablement and provider sharing
- Full request audit trail for compliance and review
Drop-in OpenAI-Compatible API
Your existing tools just work
Point Cursor, Kilo Code, Continue, LangChain, or any in-house service at the gateway URL — same payloads, same SDKs, no rewrites. Anthropic and other providers are normalized to the same surface, so switching models is a config change.
- OpenAI-shaped chat completions and models endpoints
- Streaming, tool calling, vision, and reasoning models
- Anthropic and others normalized to one request shape
- Works with Cursor, Kilo Code, Open WebUI, LangChain, and more
A Workspace Your Team Will Actually Use
ChatGPT-grade experience, your governance
A full conversational workspace lives inside ACTIVATE — persistent threads, branching, file attachments, and team sharing — all under the same SSO, budgets, and audit trail as the rest of the platform. No shadow IT.
- One model picker spanning every registered provider
- Persistent conversations with branching and message editing
- Attach images, PDFs, Office docs, and CSVs
- Share threads with teammates view-only or collaborative
Terminal-Native for Developers
Chat, code, and providers from the CLI
The same governance lives in your shell. Drive chat, register providers, and discover models from a script — and soon, let a coding agent work against your providers, your budgets, and your audit trail.
- pw ai chats — interactive REPL with model picker
- pw ai providers — register and share providers from the terminal
- pw ai models — discover models across every provider
- pw code agentic coding coming soon
Built for the Whole Org
From finance and security to the developers shipping features — one gateway, one control plane
Allocation-Aware Keys
Keys bound to budgets stop runaway spend before it happens
Streaming + Tools
Streaming, parallel tool calls, vision, and reasoning models end-to-end
One Bill for AI + Compute
AI tokens, interactive sessions, and HPC jobs roll up to one finance view
CLI and IDE Ready
pw ai today, pw code soon — same gateway behind every developer surface
Ready to Govern Your AI Spend?
See how ACTIVATE AI unifies providers, caps budgets, and gives every team the AI they need — under one control plane.