An AI that knows the company as well as the people who built it.
Years of institutional knowledge sat across four disconnected tools, accessible to no one in any useful way. I built an internal AI assistant that surfaces exactly what employees need, when they need it, without requiring anyone to know where to look.
“We always said our biggest asset was our people's knowledge. The problem was it stayed with our people. Now anyone can ask a question and get a proper answer in seconds. Onboarding used to take weeks just to get someone up to speed. It is faster now than we ever expected.”
A fast-growing UK scale-up with years of institutional knowledge locked in documents, Slack threads and Notion pages, and no reliable way for employees to find it.
As the team grew, the gap between what the organisation knew and what new employees could access was widening. Onboarding was slow because knowledge was scattered. Experienced team members were spending increasing time answering questions the organisation had already answered somewhere else. The information existed. Finding it consistently did not.
Connect the sources, train the model, and put the answers where people already work.
I designed and built a custom AI assistant that connects to the organisation's existing knowledge sources, indexing content from Notion, Slack, Google Drive and an internal documentation wiki into a unified retrieval layer. Employees interact with the assistant through a familiar interface embedded in their existing workflow, asking questions in plain language and receiving cited, accurate answers drawn from the organisation's own material. The platform includes an admin layer for managing source connections, reviewing query logs and identifying gaps in the knowledge base, giving the team the ability to continuously improve coverage.
Institutional knowledge that used to live in people's heads is now accessible to everyone in seconds.
Onboarding time fell by 65% as new hires could answer their own questions without depending on colleagues. The assistant now handles over a thousand internal queries per month, with answers delivered in under 30 seconds on average. Senior team members report a significant reduction in the interruptions that were previously consuming meaningful parts of their working day, and the query logs have surfaced recurring knowledge gaps that the team has since addressed.
Notion
Slack
Google DriveCited · NotionNotionSlackGoogle DriveCited · NotionLet's start building, together.
A 30-minute chat, no pitch deck. Directly with Colin, no middlemen. Honest advice on fit, scope and cost.
Select a date