First chat
After startup, the web UI opens, but chat needs a selected language model. Without a model, a message can fail with a configuration error.
1. Sign in
Open http://localhost:9090 and use the initial local administrator account:
admin@example.comgiga_agent_admin
2. Add a connection
Open Settings → Connections and add a provider connection. A connection stores access parameters such as keys, base URLs, and provider-specific options.
3. Add a language model
Open Settings → LLMs, create a model linked to the connection, then select it in Settings → General as the default model.
4. Send a short message
Briefly explain what you can do in this installation.
If the answer streams back, the basic setup works.
What to configure next
| Task | Enable | Where |
|---|---|---|
| RAG over documents | Embeddings and, if needed, Qdrant. | RAG |
| Long-term memory | User embeddings. | Memory |
| Python and shell execution | A sandbox runtime. | Tools, Sandbox and security |
| Image generation | Image generator provider. | Images |
| Web search | Search provider. | External services |
| GitHub, VK, weather, subagents | Related secrets. | External services, Subagents |