What we run / Capital advisory
Advisory firms on one research desk: public facts collected every day, checked, and signed off by a person.
Where it started
Advisory teams read filings, news and lender activity by hand. Each firm kept its own notes, so the same research was repeated and some of it could not be traced back to a source. The rule from day one: a missing number is better than a wrong one.
M&A, debt and real-estate advisory firms on one intelligence layer. A company 16x9 runs. Its name is kept off this site.
- skills, AI helpers and tables in the firm kit
- 6 · 3 · 4
- sources every published note must cite, at minimum
- 2
What we built
What runs today, and what is still being built.
- Running
Research engine
Securities filings and news are collected every day. Each source is scored for credibility, a second pass tries to disprove each claim, and forecasts are later scored against what happened.
- Running
Public risk desk
Footnoted notes on jurisdiction changes, licence deadlines and companies whose ownership is hard to see.
- Running
Firm operating kit
Six skills, three AI helpers and four tables a firm installs in one step. Shipped August 2026.
- Running
Sector guides and a data-backed public site
Plain guides by sector, and a public site whose figures are read from the research tables, so every number traces to a row.
- Building
Finance runners and a pre-call check
Background research run before every client call.
How it runs
From the source of the work to the person who signs off.
- 1Fixed scripts collect public filings, news and records on a schedule. No AI in this step.
- 2Each fact lands with its source link, a confidence score and a key that stops duplicates.
- 3AI writes the note, but may only copy the facts it was given. It cannot compute a new number.
- 4A sealed check reviews every note: length, sources, banned wording, no price talk, no near-copies of recent notes. If nothing new passes, nothing publishes that day.
- 5A named person approves anything that leaves the firm.
What people still decide
- Which sources the collectors read.
- Whether any client-facing work is released.
- What counts as a good enough source for a claim.
- Whether a published note stays up.
What comes next
- Finish the pre-call check.
- Add more checks that block invented figures before anything is drafted.
We run this ourselves. We can run it for you.
Tell us what your company does and where the hours go. The first conversation is about your work, not our software.
info@16x9.ai