Help
A few useful answers
Turn a brief into research, design and a working app—with your chosen AI models collaborating in one workspace.
Which AI providers can I use?
OpenAI, Anthropic (Claude), Google (Gemini), DeepSeek, Qwen (Alibaba Cloud Model Studio), Kimi (Moonshot AI) and MiniMax, through your own API keys. Choose a model for each role on your AI team; only the providers your team uses need a key, and each provider bills its own account. What a model can do differs: research needs sourced web search, and app builds need specific capabilities for each role, so not every model can fill every role. Each integration is implemented from its provider’s official documentation and tested against documented examples; none has yet been exercised against its live service from this deployment.
What do I need to start?
An account and an API key for at least one supported provider. Chat works with the model you choose. For a build, you assign a connected model to each role the task needs; one provider can fill several roles when its models support the work. A build is complete only when TGTHR’s own checks pass in an isolated sandbox.
Who pays for the model usage?
Your providers. Each API key you add is billed by its provider to the API account behind it. TGTHR has no plan of its own and does not resell model usage. Costs shown in TGTHR are estimates; usage a provider does not report stays marked as unknown.
What does “verified” mean?
Before a build is handed over, a separate step builds the app and checks it in a browser. If checks fail, the engineering role repairs the code and it is checked again, within your limits. Passing is not a guarantee that an app has no bugs.
What can I inspect and keep?
Documents, decisions, source exports, check evidence, revisions and usage stay connected to your project. Generated apps run in an isolated preview when the deployment supports it.
After you sign in, the full guide in your workspace covers connections, readiness, how a build works, limits, previews and troubleshooting: open the in-app help.