Errors in the support bot
Sarah Chen, an ML engineer at Acme Ltd, asked which support bot traces had errored today. Memtro listed them with their error type and message, duration and cost, so she could see they all failed on the same step.
Which traces in the support bot errored today, and what did they cost?
Comet Opik watches your LLM applications. Memtro reads projects, traces with latency, cost, token usage, errors, tags and feedback scores, experiments with status, scores and pass rate, prompts and their version counts, datasets, and the workspace's cost for a period. When asked and granted, it can comment on, score, tag or delete traces, manage projects, create prompts and commit versions, create or delete datasets, and cancel or delete experiments. Opik alerts can tell Memtro when errors, cost or latency pass a threshold, and a job can start on it. Trace inputs and outputs are never read.
Ask in plain English from chat, your editor, a scheduled job or your own agent. The model picks the right Comet Opik tool, reads what it needs, and answers with the source cited.
These tools change something in Comet Opik. The model is told to use them only when you ask for exactly that, and an administrator grants each one by name to the teams that may use it; a personal connection acts only after its owner allows it.
Turn events on in the connection's Events tab on the Connections page, then add the address it shows as a webhook in Comet Opik, choosing the same kinds; the tab shows Comet Opik's own steps. Each event can start a job, with the event as its input. Comet Opik sends a secret with every delivery, and one without it is refused.
Events and webhooksOnce Comet Opik is connected, these are the kinds of things people ask. Plain questions, no syntax. If the answer lives partly in another system you have connected, Memtro looks there too.
Which traces in the support bot errored today?
What did the workspace spend on LLM calls this week compared with last week?
Which experiments on the refund dataset had the best pass rate?
How many versions does the triage prompt have?
Tag the slowest trace from this morning as needs-review.
Sarah Chen, an ML engineer at Acme Ltd, asked which support bot traces had errored today. Memtro listed them with their error type and message, duration and cost, so she could see they all failed on the same step.
Northwind added an Opik alert for failed guardrails so Tom Reilly, its AI lead, hears at once. When a guardrail check failed on a trace, a job read that trace's tags, scores and error and gave Tom the details.
The teams that lean on Comet Opik most, a question each of them asks, and what they connect beside it.
A record in Comet Opik is rarely the whole story. Connect the systems that hold the rest and Memtro checks them together, then remembers what it learned so the next question starts further along.
Using something else for the same job, or both? Each connector has its own tools; connect as many as you like.
In Opik, open the user menu or Account settings and copy your API key. It acts as you in every workspace you belong to.
In Memtro, open Connections → Comet Opik and paste the Opik API key and the Workspace name as Opik shows it.
An administrator grants the trace, prompt and delete actions by name to the teams that may use them. Events are turned on from the connection's Events tab, and the address is then added as an alert's Endpoint URL in the Opik project.
Ask which traces in the support bot errored today.
Chat in the dashboard, the OpenAI-compatible API, scheduled jobs, and any MCP client: Claude Code, OpenCode, Cursor, Claude Desktop. The Comet Opik tools are the same everywhere.
Only when you ask, and only what it has been granted: add a comment to a trace; score a trace (adds or replaces a feedback score by name); remove a feedback score from a trace; add or remove tags on a trace; delete traces by id (permanent); create a project; rename a project or change its description or visibility; delete a project (permanent); create a prompt in the prompt library; commit a new version of a prompt; delete a prompt with its versions (permanent); create a dataset; delete a dataset (permanent); cancel running experiments; delete experiments by id (permanent). Every other tool reads, searches or lists. The Connections page labels the tools that act, an administrator grants each one by name to the teams that may use it, and a personal connection acts only after its owner allows it.
A personal connection serves only your own requests. A shared connection serves the organisation: everyone, or only the teams an admin chooses. Admins can always add shared connections; other members need the permission.
Credentials are encrypted at rest. Tool results are cached briefly so agent loops do not repeat identical calls, and the facts the model extracts are saved to your organisation's memory with provenance (system, record and link) so they can be checked and superseded.
Opik API key and workspace. In Opik, open the user menu or Account settings and copy your API key. It acts as you in every workspace you belong to.
Yes. Turn events on in the connection's Events tab on the Connections page, then add the address it shows as a webhook in Comet Opik, choosing the same kinds; the tab shows Comet Opik's own steps. A job set to start "When an event arrives" then runs once for each matching event, with the event as its input. The run treats the event as outside data, never as instructions, and anything the run remembers is kept as unverified.
Seven-day free trial, then one monthly price per organisation. Connectors, jobs and memory are never metered.
Sign in, add a provider key, create an API key and point your tools at Memtro. It takes about five minutes.
7-day free trial · card on file, nothing charged until it ends · cancel any time