How Chalk Ideas works
Four promises the pipeline is built to keep, and what each one costs.
We read the market, not opinions about it
Every night the scanner works through 160 public sources across 9 sectors — funding filings, regulatory dockets, hiring threads, job boards, code repositories, research indexes and the trade press. Each one is an official feed or a public API. Nothing is scraped, nothing is behind a login, and nothing arrives as somebody's opinion about what is happening: the raw material is the record itself.
- 160
- 9
- Scanned
- paused
- Last scan
- 27 Aug 2026, 17:52 UTC · 9 of 9 sectors
One finding becomes one idea
A cluster of evidence is written up once, under the single lens it actually fits: a business to start, a project to build, or a job to pursue. The model proposes which, and then has to get past checks it does not get a vote on — a job needs hiring evidence in the source material, not an inference from news, and a business needs a named payer, because without someone to sell to it is a project. Findings that fail their own lens are dropped rather than filed under a different one.
- Per finding
- one idea
- Lenses
- Business · Project · Job
- Chosen by
- evidence
Checked against the live web
Each new idea is put to a model with live web access, which searches and cites as it answers. It comes back with named competitors, what each of them already covers, and the gap they leave — plus a read on how crowded the space is. The citations travel with the idea onto its page, so every judgement can be followed back to the page it came from. An idea already in the bank keeps its recent verdict rather than being re-checked nightly; markets move on the timescale of funding rounds, not overnight.
- Grounding
- live web
- Citations
- shown on every idea
- Competitors
- named, with gaps
Scored, and held back if it isn't good enough
What survives gets a 0–100 composite from four axes — demand, openness, simplicity and timing — and a plain reading of how contested the space is: Wide open, Emerging, Competitive, Saturated. An idea whose market the check calls covered, and whose score agrees, is not published: it is held for review or retired outright, because no amount of rewriting turns a crowded space into an open one. Ideas that keep recurring accumulate weight and their score is blended rather than replaced, and anything not seen for a fortnight is archived. Nothing is deleted.
- 0–100
- 4
- 14