telamin.

ONE CONNECTED ENGINEERING LOOP

Fix the problem
with the facts in view.

The Analyser ties the system together: the application’s structure, the observations from a run and the investigation that explains them.

The compiled application structure and its recorded audit facts meet in the Analyser. A connected local AI assistant investigates and proposes a change; people review, test and replay it.
From a recorded problem to an evidence-backed change. The application structure gives the audit facts context. A connected AI coding assistant can investigate the recorded behaviour, propose a change and help check the result through tests and replay.

MORE THAN A LOG VIEWER

Give the assistant
evidence to work with.

Load the audit log with its application graph and configured local source. Connect an AI coding assistant to the Analyser’s tools through Model Context Protocol (MCP). It can investigate recorded values, examine the code context and propose a targeted change.

Locate the problem.

Bring an event, its recorded values and the relevant application structure into the same investigation.

Propose a change.

Use that context to identify a cause and suggest a fix, with the supporting observations visible to the reviewer.

Check the result.

Run tests and, where captured inputs support it, replay and compare the behaviour. Keep people responsible for accepting the change.

EVIDENCE BUNDLES

Share the investigation.
Keep its context.

An evidence bundle is a portable .fexp file: the captured event history, its available graph and a saved investigation. Charts, reports and saved step-by-step investigations let the recipient follow the relevant events and inspect the supporting records.

How evidence bundles work
Audit records, graph context, saved charts, reports and walks are packaged into an fexp file that another Analyser opens as a fresh investigation.
The facts and the investigation travel together. The recipient opens the same captured material and saved investigation. Supported bundles can also include paired replay inputs for comparison against the recipient’s own matching build.

ONE INVESTIGATION. DIFFERENT AUDIENCES.

Send the context.
Let others examine it.

Engineering teams.

Hand over the event history and saved investigation so the next developer can investigate, propose a fix and check the changed behaviour.

Clients and partners.

Share an investigation report with its supporting captured records and context, rather than relying on a narrative alone.

Assurance reviewers.

Support an answer to an audit or regulatory question with a reviewable investigation and the observations behind it.

WHAT TRAVELS

The evidence
and the explanation.

Whole-log or excerpt captures make a support case, a review or a proposed change easier to hand to someone else.

  • The captured log, with any excerpt identified.
  • The graph, when one was open.
  • Saved charts, reports and step-by-step investigations.
  • Notes and a manifest of member hashes and sizes.
  • Paired replay records, when explicitly captured in a supported replay bundle.

Machine settings, source-root configuration, processor binaries, credentials and assistant settings stay outside the bundle. The recipient brings their own build and source configuration where those are needed.

What verification and replay establish

EXPLORE THE NEXT STEP

Walk through a problem with the evidence.

See an event, its graph context and an investigation together. Then explore how a tested change and a portable bundle can support review.

Request a walkthrough