Why every consulting firm should run a private knowledge base
Email threads, scattered Slack channels, shared drives. Works at three; falls apart at thirty.
In short
A private, per-workspace knowledge base keeps each client's reading isolated and citation-grounded — the opposite of an org-wide wiki that nobody trusts or maintains.
The team behind Pith Lab
The case
Most consulting firms still rely on email threads, scattered Slack channels, and shared drives. That works at three people; it falls apart at thirty. The case for a single retrieval surface is now clearer than ever — especially with cheap embeddings and decent retrieval models.
What breaks first
Three failure modes appear, in order: (1) the same article gets sent in DM by two partners on the same day; (2) a junior asks for the regulatory landscape and gets four different shared-drive folders; (3) a deck for a client cites a piece of research nobody can re-find later. Each is small. Together they're the whole problem.
The knowledge a consultancy compounds is its real product. Email and Slack are not knowledge bases — they're sieves.
Three minimum tables
A private knowledge base doesn't need a CMS, a headless service, or a vector-db daemon to start. Three tables get you most of the value: bookmarks (URL, title, content, tags), highlights (passages saved from articles), and concepts (auto-emerged topics with citations back to bookmarks). Everything else — briefings, audio, related links — is derived.
Why not Notion
Notion is a write-tool. It rewards the act of authoring. A knowledge base for consultants needs to reward the act of reading. The interface should be a stream of inputs (bookmark, highlight, dismiss) that compound into outputs (briefing, concept page, search). Authoring is the wrong primitive.
Cost analysis
A one-person consultancy reads ~15 articles a day. At ~5 minutes each, that's 75 minutes daily on input. At ~10 seconds to bookmark + 30 seconds to highlight a passage, the cost of capture is ~10% of reading time. The return — re-find anything ever read, generate weekly briefings, surface concepts across an engagement — is the rest of the consulting career.
FAQ
How is this different from Notion or Confluence?
Notion and Confluence are write-tools. Pith is a read-tool. The output is briefings and concepts, not pages someone has to author and maintain.
Where does the data live?
Frankfurt, Hetzner. EU-only data residency. Per-workspace Postgres database. Export anytime as SQL or markdown.
What about confidentiality?
Workspace-scoped. Tag bookmarks per client; the briefing for that client only sees those bookmarks. Auditable.