Tenon Stack

Work / Compliance ยท Knowledge

Decades of regulation, answered with the page number.

A retrieval system over 1,500+ regulatory documents and 500+ amendments, some of them typewritten originals, that cites the exact source of every statement.

The situation

The body of rules had been written over decades. The oldest documents were typewritten originals; later years added laws, rules, regulations and more than five hundred amendments, each changing what an earlier text meant. In all: more than 1,500 documents, 10,000 sections and 24,000 subsections and clauses.

The people who needed answers from it were rarely the people with time to read it. The requirement was simple to state and demanding to meet: answer questions in plain language, account for every amendment, and say only what the text supports.

What we built

We built a retrieval layer over the full corpus. Typewritten pages were converted to clean text and checked. Every document was split along its own structure (parts, sections, subsections and clauses) so that a retrieved passage keeps its legal context, and each amendment was linked to the provision it changes.

Answers are written only from retrieved passages, and each statement carries a reference to the document, section and page it came from. Where the regulations do not address a question, the system says so plainly.

What changed

Questions that used to mean a search across decades of paper are answered in seconds, with the current wording, the amendment history and the source one click away. The same design now underlies our Private Knowledge Engine for internal and confidential documents.

Documents
1,500+, including typewritten originals
Amendments
500+, linked to the provisions they change
Structure indexed
10,000+ sections, 24,000+ subsections and clauses
Every answer
Cites document, section and page
Design
Runs on a cloud model or fully on-premises

The pattern behind it

Private Knowledge Engine: Staff ask in plain words and get the answer with the document and page it came from. Cloud or fully on your own machines.

The workflow

From a question to a cited answer.

  1. Your documentsPDFs, Word files, scans and intranet pages stay where they are.
  2. AI step: IndexEach document is split, labelled and made searchable, with access rules kept.
  3. A questionStaff ask in their own words, in the tool they already use.
  4. AI step: Retrieve and answerThe layer finds the relevant passages and writes an answer from them alone.
  5. AI step: Cited replyThe answer arrives with the file, section and page for each claim.

Amber steps are the AI layer. Everything else is the system you already have.

Next step

Want the same result?

Tell us what your version of this problem looks like, and we will map the route to the same result.