Run an investigation

An investigation starts with a question and ends with an explanation you can check. This page covers the chat, the help desk sidebar and the API. For Slack, see Use the Troubleshooting Agent in Slack.

Ask in the chat#

1. In the Troubleshooting Agent, open Pluno.
2. Describe the problem. Mention the customer, the system or the error if you know them.
3. Optionally, select Attach to add files, such as a screenshot or a log export. You can attach up to 10 files of 10 MB each.
4. Send.

The chat opens once you have connected at least one app.

Write a good question#

The more specific you are, the faster Pluno finds the right place to look.

Watch it work#

While Pluno investigates, the Live events panel lists every step it takes: which system it queried and with what. Expand a step to see what came back.

Investigations take longer than a chat reply, because Pluno queries real systems one after another. If the connection drops, Pluno reconnects by itself. If a run fails, select Retry.

Read the result#

Pluno answers in plain text, leading with the root cause. A typical answer contains:

• The root cause, or the best current explanation.
• The evidence behind it, with links.
• A recommended fix and the next step to take.
• A note on any system it would have needed but could not reach.

Ask follow-up questions#

Reply in the same chat. Pluno keeps the context of the whole conversation, so you can say "check the same thing for last week" or "look at the payments service instead".

Find and share investigations#

• Investigations in the sidebar lists your chats, newest first.
• Share copies a link to the chat. Anyone in your workspace can open it.
• New chat starts a fresh investigation.

Investigate from a ticket#

On a ticket in your help desk, the Pluno sidebar has a Troubleshooting Agent box.

1. Select Investigate ticket.
2. Wait for the Summary.

Pluno gets the ticket's messages and details, so you do not have to describe the problem. To steer it, type in Chat with Troubleshooting Agent... and select Send. Each message starts a new investigation of the ticket, so include everything Pluno should take into account.

The result appears in the sidebar. When the fix is something the customer can do themselves, Pluno also posts it on the ticket as an internal note, ready for you to pass on. Nothing is sent to the customer.

When Pluno confirms a bug, it goes to engineering with the recommended fix attached. See Escalation Copilot for how escalations and their updates work.

Pull requests#

When Pluno is confident about a code fix and has access to the repository, it may end its answer by asking whether you want a pull request.

If you say yes in your reply, Pluno creates a branch whose name starts with pluno/, commits the change, and opens a pull request for your team to review. Nothing is merged.

Your reply in the chat is the approval, so answer deliberately. The GitHub connection needs write access for this to work.

Start an investigation from your own tools#

Use the API to trigger investigations from an alert, a script or an internal tool.

1. Create an API key under Settings, then API. See Settings, plans and credits.
2. Send the key in the X-Pluno-Api-Key header.
3. Start a query, then ask for its result until it is finished.

The status is pending, running, completed or failed. A failed query has an error_code: limit_exceeded when you are out of credits, query_timed_out, or query_failed.

Each query is a separate investigation. The API settings page shows ready-to-copy examples.