Platform Configuration
Platform setup guidance for default and on-prem Outpost modes
This page separates the default flow (for most teams) from the on-prem flow (advanced).
Default (for most users)
If you use data pools in standard meinGPT cloud mode, you do not need to configure your own Outpost.
- Connect sources in meinGPT
- Select your data pool
- Use it in assistants and workflows
The underlying Outpost component is managed in the background.
On-Premise / Advanced (customer-managed Outpost)
Extra setup is only required when you operate your own Outpost runtime.
1. Create Outpost in meinGPT
- Open Company Settings > Outposts
- Click Add Outpost
- Enter a name
- Save ID and Secret Key from the dialog
Details: Create Outpost
2. Configure runtime
Set these values in your Outpost runtime:
VAULT_IDVAULT_SECRETMEINGPT_URL
Example:
VAULT_ID=...
VAULT_SECRET=...
MEINGPT_URL=https://app.meingpt.comAlso configure at least these sections in config/app_config.yaml:
vault.idvault.secretmeingpt_urlweaviate.*embedding_model.*
Details: Local Deployment and Configuration Reference
3. Configure data pools in meinGPT
Data pools are still created in meinGPT (name, source type, permissions). The Outpost fetches this configuration at runtime.
You can optionally define additional local data_pools in app_config.yaml (advanced).
4. Verify
- Check Outpost health
- Trigger sync
- Check status and logs
Details: Verification
Decision guide
- Fast start / no self-operation: default flow
- Own infrastructure / full control: on-prem flow