Blog
Updates on product, data sovereignty, encrypted AI, and engineering.
-
Confidential work does not belong in a model you cannot check
A closed model can promise that your drafts are not used for training. TeeChat keeps the archive on your machine, and lets you verify the path before a prompt leaves.
-
Use Cline in VS Code with a TeeChat API key
Install VS Code and the Cline extension, choose Bring my own API key, then point OpenAI Compatible at TeeChat OpenAPI — same base URL and key flow as WorkBuddy.
-
What we found in TeeChat’s seal-sync protocol — and how we closed it
An internal pre-payment review found a critical key-migration flaw. We contained the ports, deployed the fixed version, rotated the serving keys, revoked the associated certificates, and are closing the finding.
-
Use your own Ollama model with your own web search
Local mode is as private and friendly as we can make it: chats stay as Markdown on your device. When the web is needed, you configure your own search.
-
Immutable golden images across every TeeChat hosted component
OpenAPI edge, routing gateway, and GPU inference engine now boot from immutable golden images. The production inference CVM no longer offers SSH. You can still verify published fingerprints before you send.
-
Run OpenClaw on TeeChat OpenAPI — chat and memory embeddings
Point OpenClaw at TeeChat’s OpenAI-compatible API for chat and memory embeddings. Prep a TeeChat API key (same flow as WorkBuddy), then configure the custom provider via onboard, configure, or the dashboard.
-
TeeChat OpenAPI is generally available — connect WorkBuddy in a few steps
Point everyday AI tools at TeeChat’s OpenAI-compatible API. This guide walks through Tencent WorkBuddy with product screenshots; OpenClaw has its own follow-up guide.
-
How to verify you are talking to TeeChat’s secure OpenAPI service
Challenge openapi.teechat.ai, verify hardware quotes against GitHub release hashes (with teechat.ai as network fallback), and pin measurements — so your OpenAI-compatible client can prove it hits a measured confidential-computing endpoint without giving up base_url + API key convenience.
-
Stop re-explaining the same project — keep checklists and hard research from fighting over one chat
Practical patterns for project folders: lasting facts that stick, a one-turn checklist, and parallel research that still returns one clean reply.
-
Beyond "opt out of training": how confidential computing defines data sovereignty in the AI era
How hardware confidential computing answers AI-era privacy sovereignty — and four myths professionals ask about.
-
Welcome to TeeChat
Your AI, your folder — you rule. Private chat with data sovereignty on teechat.ai.
-
How to verify confidential chat on TeeChat
Step-by-step verification — hardware proof, what the routing server does, and the path from your device to a verified AI engine.
-
Why we built TeeChat — privacy in the AI era
Real-life moments when AI chat needs privacy — and why data sovereignty matters for ordinary work.