Ask Venn About Your Data

Search, tail, and summarize logs, metrics in plain language, check dataflow health, and get documentation answers without leaving the chat.

Search logs in plain language

Ask for what you want to see; Venn builds and runs the search and shows matching log lines inline in the chat, color-coded by severity.

• "Show me errors from the payments app in prod in the last 2 hours"

• "Find logs mentioning timeout in the checkout namespace"

• "Any warnings from nginx today?"

How it behaves:

• If you do not give a time range, Venn searches the last hour by default.

• Results are paginated: 100 lines per page, up to 1,000 lines per search. Ask for "more" or "the next page" to continue the same search.

• Alongside log lines, results include a breakdown of the fields in the result set (services, severities, hosts) and, where available, recurring message patterns with counts, so you can see the shape of the traffic at a glance.

Power users: you can give Venn precise field-level filters and it will use an expression instead of a plain keyword, for example: "search the prod namespace where severityString is ERROR and message matches timeout". Boolean combinations and regular expressions are supported.

Tail logs live

Ask Venn to watch live traffic for a bounded window:

• "Tail the prod namespace for 60 seconds"

• "Watch the checkout app for errors for two minutes"

Venn collects up to 120 seconds or 500 lines, whichever comes first, then returns the batch together with a short summary: severity counts, the busiest applications and hosts, and repeated patterns. Tailing is a single bounded batch, not a continuous stream.

Summarize what you found

After a search, ask:

• "Summarise these logs"

• "Summarise the last search, focus on errors only"

Venn produces a triage-ready summary: grouped errors, recurring patterns, and likely-cause hints. You can steer the focus (errors, performance, security). Summaries work on your most recent search results; after roughly 10 minutes Venn re-runs the search first.

Check dataflow health

• "Which dataflows have the most errors in the last 24 hours?"

Venn scans error and warning volumes per dataflow over your chosen window (15 minutes up to 7 days) and lists the worst offenders first, with a severity breakdown. Venn also explains what makes up the totals: on many environments a large "critical" count comes from synthetic check states rather than application errors, and Venn calls that out instead of quoting a scary number.

Documentation answers in the same chat

Ask how-to and concept questions right in the conversation:

• "How do I configure a Kafka source?"

• "What is a stream rule?"

Venn answers from the Apica documentation with links to the relevant pages, and decides on its own when a question needs the docs and when it needs your live data.