DenserAI Logo
City of San José seal
Local GovernmentUnited States

How the City of San José turns its code and contracts into cited answers

The civic heart of Silicon Valley deployed Denser so staff and residents can ask plain-language questions about procurement, the municipal code, and engineering standards — and get answers cited straight to the source document.

~50 min → minutesResearch time per inquiry
6 core sourcesDocument sets indexed
100%Answers with source citations
No codeLive in days, not months

The challenge

San José's procurement and public works knowledge lives across the City Charter, the full Municipal Code, and hundreds of pages of contracts and pre-qualification documents. Finding a single answer — the bonding rule for a contract, a clause in the code, a required engineering standard — meant digging through PDFs and pages with keyword-only search.

The team needed a way for staff (and eventually residents) to ask a question in plain language and get a precise, trustworthy answer — without losing the ability to verify it against the official source. They evaluated a generic ChatGPT workflow first, but on the same set of uploaded documents it frequently returned no usable answer.

What Denser indexed

The Procurement Division's chatbot was trained on six core document sets — a mix of crawled web pages and uploaded PDFs — so every answer is grounded in official city records.

City Charter

Foundational governing document

Entire Municipal Code

Full code, searchable by question

Major Public Works Contracts

Contract terms and references

Minor Public Works Contracts

Smaller-scope agreements

Contractor Pre-Qualifications

Eligibility and requirements

Consultant Contracts

Professional services agreements

How they deployed it

Setup required no code. The team pasted in source URLs, uploaded the documents that aren't publicly crawlable, and the chatbot was answering questions in days. After more than a month of hands-on evaluation across divisions, the City moved to an annual Business plan.

Because the chatbot is configured entirely from the dashboard, the same approach now extends to other departments — each builds a focused bot on its own documents without engineering support.

Top questions it answers

The impact

Research that used to take staff around 50 minutes per inquiry now takes a few — a direct productivity gain that frees the team to handle higher-value work instead of document hunts.

In a side-by-side evaluation against a generic ChatGPT setup using the same documents, Denser returned a valid, correct answer where the alternative often returned none — with no ads and no login prompts standing between staff and the information they needed.

Accuracy you can verify

Every answer links back to the exact source — for PDFs, Denser highlights the passage on the page so reviewers can confirm it in one click. Below is a representative example of a cited answer.

ES-CMS Procurement Chatbot
What are the conduit depth requirements in the engineering standards?

Per the City's engineering standard details, the minimum cover (conduit depth) depends on the conduit type and location — the referenced detail specifies the required depth below finished grade for each case.

Source:Engineering Standard Details (PDF) — passage highlighted

Illustrative example based on the City's deployment; wording and UI may differ.

Denser cut our research time from about 50 minutes to just a few — a huge boost to our team's productivity.
City of San José · Digital Services

Build the same for your agency

Turn your code, permits, and contracts into a cited, 24/7 chatbot. See how Denser works for the public sector.