DaemonLayer does not send your ticket to a generic LLM. Before any classification runs, it gathers M365 user context, reads attachment screenshots for error codes, and searches your entire resolved ticket history for similar cases. The AI works with evidence.
The Context Pipeline
Every triage run moves through a structured enrichment pipeline. The AI agent does not receive a raw ticket. It receives a ticket with everything we know about the person, their environment, and the error.
Before any AI analysis, DaemonLayer queries Microsoft Graph for the requester's display name, job title, department, account status, group memberships, and licence status. The AI knows who is asking, not just what they wrote.
Image attachments are analysed by a dedicated vision agent that understands screenshots the way a technician would. Error codes, error states, affected applications, and diagnostic context are extracted automatically, capturing details the user never had to describe.
The Knowledge Service scores your entire resolved ticket history across multiple signals — recency, keywords, category, error codes, and semantic similarity, surfacing only the cases most likely to be relevant. Not the newest. Not the most similar. The most useful.
With all gathered context in hand, the triage agent produces a category, subcategory, urgency level, confidence score, automation recommendation, duplicate analysis, and human-readable reasoning for the decision.
A second agent queries your live PSA for valid field values and maps the triage output to the exact queue IDs, issue types, and sub-issue types your PSA accepts. No manual field mapping per ticket.
Knowledge Service
The Knowledge Service scores your entire resolved ticket history across multiple signals, recency, keywords, category, error codes, and similarity, surfacing only the cases most likely to be relevant. Not the newest. Not the most similar. The most useful.
The Knowledge Service weighs recency, keyword overlap, category fit, error code matches, and similarity together to rank past cases. Only the top matches reach the AI, each with a resolution summary and the reason it was selected.
The requester's last three tickets and the last three tickets from their company are attached to every triage run. Repeat issues, known patterns, and onboarding quirks surface before a technician ever opens the ticket.
Based on the highest-ranked historical matches, the Knowledge Service suggests the most likely PSA field values and the technician with the strongest track record in that category, with a confidence score and human-readable reason.
Continuous Learning
Every resolved ticket is automatically re-indexed with its resolution text and outcome. It joins the pool of reference cases the Knowledge Service scores against for future tickets. No manual curation. No training runs. It compounds quietly in the background.
If the Knowledge Service is unavailable, triage continues without it. M365 context and vision analysis still run. No ticket is ever blocked by a downstream service failure.
DaemonLayer extracts structured signals from ticket text and attachments at triage time. These are stored on the ticket, surfaced to technicians, and indexed for future searches.
Related
A 30 minute walkthrough on your PSA setup. We show exactly what DaemonLayer would handle for your team, on your data, with no obligation.
No sales deck. No obligation. Live walkthrough on your environment.
