The Observability settings menu
An LLM observability tool records the calls made by the Gen AI orchestrator for each answer: the condensed question, the retrieved documents, the prompts sent to the LLM and its answers, with their latency, token consumption and cost. It helps understand a bad answer, and follow the cost and the performance of the RAG.
To access this page, you need the admin role (more details on roles in security).
Configuration

- Observability activation: enables or disables the traces for the bot.
- Observability provider: the tool that receives the traces (see the list of observability providers).
For Langfuse:
- Public key and Secret key: the API keys of a Langfuse project (Settings > API Keys in Langfuse),
- Url: the address of the Langfuse server, as reached by the Gen AI orchestrator,
- Public url (optional): the address of the Langfuse server as seen from the users' browser, when it differs from Url (for instance an internal Docker or Kubernetes host name). It is used for the links to the traces.
The settings can be exported (optionally with sensitive data, such as the keys) and imported, to copy them between bots or environments, and deleted.
Viewing the traces
Once observability is enabled, each generated answer has a link to its trace (View observability details):
- in Analytics > Dialogs, on the bot answers,
- in the playground, on the answers of the LLM.
Running Langfuse with Docker
The tock-docker repository provides a Docker Compose file
that starts Langfuse (version 3) and its dependencies:
git clone https://github.com/theopenconversationkit/tock-docker.git && cd tock-docker
docker compose -f docker-compose-langfuse-v3-only.yml up -d
Langfuse is then available at http://localhost:3000: create an account, an organization
and a project, then its API keys. In the observability settings, set Url to an address that the orchestrator
container can reach (for instance http://host.docker.internal:3000), and Public url to http://localhost:3000.
Change the default passwords and secrets of the Compose file before using it beyond a local test.