Agents that close the loop: lessons from 60% auto-resolution
What it took to move an AI support suite from demo to resolving the majority of tickets on its own.
Read articleTrust is the product
An answer without a source is a rumour. In enterprise search, people need to verify before they act — so every answer we generate links back to the passage it came from.
Chunking is a design decision
How you split documents decides what the model can find. We chunk by structure — headings, tables, tickets — not by arbitrary token counts.
Permissions all the way down
Retrieval respects the same access rules as the source systems. If you can’t open the document, the assistant can’t quote it.
Have a question about this, or a project where it applies? Write to us — Slava reads every reply.
