AI agent for technical support specialists
Support Attachment Interpretation Agent
A clear, evidence-based next diagnostic step for every ticket with attachments.
What it does
Tickets often arrive with screenshots, logs and descriptions that do not tell the same story. Agents then guess and ask generic questions. This agent reads every attachment, uses text recognition on screenshots, and places all events on one timeline. It forms a working hypothesis and lists where the sources disagree. It then runs one permitted diagnostic query that would settle the biggest disagreement and checks whether the result supports the hypothesis. If not, it revises the hypothesis, looks at the relevant timestamp again, and repeats. It ends with an evidence map and a specific next step on the ticket. The support agent approves any message to the customer. Edge case: when screenshot times and log times are in different time zones, it aligns them before comparing.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Ticket with attachments
- Read screenshots (OCR) and logs into one timeline
- Form a hypothesis and list disagreements
- Run a permitted diagnostic query
- Does the evidence support the hypothesis?If not: revise the hypothesis and inspect the relevant timestamp. Back to step 3.
- Write the evidence map and next step
- Agent approves any customer communicationThe agent waits here for your OK.
- Targeted investigation note on the ticket
How it decides
It chooses queries to resolve disagreements between attachments; when a log contradicts its reading of a screenshot, it revises the hypothesis and inspects that timestamp.
- Which query to run: the permitted one that resolves the biggest disagreement in the timeline.
- Revise the hypothesis: when the query result or logs contradict the screenshot reading.
- Time alignment: convert all sources to one time zone before judging order of events.
- Stop and hand over: after three hypotheses without support, write the evidence map and ask a senior agent.
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Diagnostic queries the agent may run (default read-only account and session queries)
- Maximum hypothesis rounds before handing over (default 3)
- Attachment types read (default images, log files, HAR files)
- Default time zone for the timeline (default UTC)
- Who approves customer messages (default the assigned support agent)
What keeps you in control
It always asks you first
- Customer communications
- Production changes
Hard limits
- Read-only.
It stops when
- Done: hypothesis supported by evidence.
- Needs a human: evidence insufficient.
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide
