Meridian gives support agents evidence-grounded answers with full provenance tracing — and gets smarter from every resolved ticket.
Watch the system encounter a novel problem, detect the knowledge gap, generate a new article, and then retrieve it — all in real time.
Add 6 novel "Report Export Failure" tickets to the system.
Run gap detection — compare each ticket against the KB corpus.
Cluster the gaps by category/module to detect a new pattern.
Use the LLM to draft a new knowledge article from the first ticket.
Approve the draft and add it to the retrieval index.
Query the copilot — the new article should now appear in results.