Operations
GAPIVO admin
Corpus
as of last run · no run recorded yetWhat is actually stored, by source. Read from the committed run artefact — the admin origin never reaches the database.
| source | live rows | tombstoned | oldest | newest | last collected |
|---|---|---|---|---|---|
| No per-source totals recorded yet — they are written by the next Collect run and appear here from the committed artefact. | |||||
0 of these rows are fixture data, not real problems.
demo_forum has no live API — it exists to exercise the mapper. Exclude it before quoting a corpus size to anyone.
1
Collect problems
Fetches from public sources into the corpus. Free, repeatable, safe to run twice — unchanged rows are just touched.
Sources
Period
1 source selected
Queued — waiting for a runner
dispatch
queued
fetching
storing
publishing
| source | status | window | fetched | new | updated | unchanged |
|---|---|---|---|---|---|---|
| CFPB | OK | — | — | — | — | — |
2
Analyse & generate ideas
Clusters the corpus, reasons over it, writes business ideas. This is the only step that spends money.
| New since last analysis | 0 problems — nothing to re-analyse |
| Estimated spend | $0.00 · month to date $11.40 of $50.00 |
Collect something new first
Only changed inputs are re-analysed. Ideas whose source problems are unchanged keep their existing workup — that is why the price is per new problem, not per run.
3
Review & publish
Rebuilds the public site and pushes it to the CDN. Free and fast — but it is the step that makes things public.
Nothing new to publish
Separate from step 2 on purpose. Merging them would put every generated idea in front of the public unreviewed — and the legal guard (derived insights only, never verbatim source text) is checked here.