IPE

Candace AI

Candace AI is building the operating system for small-business M&A. A single secure, auditable workspace that connects brokers (advisors), sellers, and buyers from first contact to close. The US is entering a two-decade wave of business-ownership transfers as Baby Boomer owners retire (the „Silver Tsunami“), while much of the industry still runs on email, spreadsheets, and PDFs. Candace AI closes that gap with AI-assisted, privacy-first tooling built around a „walled garden“ and traceable data provenance.

Our project

Over six weeks in Mountain View, our two-person team works directly on the deployed Candace AI product, with three focus areas:

  • Mapping the application: documenting every screen, workflow, and user role (advisor, seller, buyer, brokerage) into one complete product map.
  • Improving AI document generation: refining the prompts and workflows that turn deal data into business documents such as the Broker Opinion of Value and blind profiles.
  • Improving AI document parsing: testing and improving how uploaded documents are read and converted into structured, source-traceable deal facts.

The platform already used AI, but the prompts were simple and static, so the output was inconsistent. Rather than judging results by feel, we defined golden input and output pairs, built a scoring rubric, and measured every prompt change against it.

Project log

Week 1
Kickoff with the Candace team and a full walk-through of the deployed product. Agreed the six-week scope with our client and started mapping the application.

Week 2
Finished mapping the advisor workspace and turned it into a functional tree diagram. Started reading the current generation prompt and workflow in the codebase.

Week 3
Got the application running locally and reported our first confirmed bugs to the client. Began building the scoring rubric for generated blind profiles.

Week 4
Ran baseline and improved prompts against the golden set and measured the difference. Found and fixed two flaws in our own measurement logic along the way.

Beyond the desk

Our team with Paul Jon Kelley at the CABB Silicon Valley Chapter Lunch in Sunnyvale
Silicon Valley CABB Chapter Lunch, 23 July 2026, Plug and Play Tech Center, Sunnyvale. Our Project Provider Paul Jon Kelley presented „AI Revolution in Brokerage“ to the California Association of Business Brokers and brought us along, which put us in a room full of the brokers our work is built for.

The team

Vejhan Dervisoski
Dionis Koci

Client: Candace AI, Mountain View, California. Technical lead Chris Oliver, domain lead Paul Jon Kelley. Academic supervision: Prof. Christian Schucan, HSLU.