Who it is for
A market, its filings, and what the rule actually says
Written from primary sources and checked against them. As of 2026-09-06. How that is done.
Short answer
Could a state run its whole cannabis market on Bloom?
Not yet, and this is the furthest out of the four builds. The idea is a tenancy tier above the operator: an agency sees every licensee in its jurisdiction, what they have filed, where their declared inventory diverges from what they hold, which licensees received a recalled package, and licence status and renewal. What makes it plausible is that Bloom already holds the rule alongside the record, so a filing can be checked against the law rather than only against last month's filing.
The hard part is not the dashboard, it is the ingest
No state can require its entire market to change software, so any agency system that only works when everyone uses it does not work. Bloom is therefore being built around a first class ingest layer rather than an importer: format detection, a saved column mapping per source so a shop's monthly export keeps working, unit and timezone normalisation, validation that produces a rejection report the sender can act on, and idempotent reprocessing so the same file arriving twice does not double count.
Metrc exports, BioTrack and BT Government exports, the Washington CCRS file format, and the ordinary spreadsheet a point of sale produces all have to land in the same package, sale and transfer records.
Why the corpus matters more here than anywhere else
An agency asking whether a licensee is compliant is asking a question about the rule, not only about the record. Because every fact Bloom publishes carries its source and the date that source was last read, a variance can be explained in terms of the rule that was in force on the day, and the rule as it stood is still there after it changes, because a published row is superseded rather than edited.
What we would need from an agency before building it
The authority boundary, in writing: what the state may see, what it may change, and what remains the licensee's. The filing formats it actually accepts. And a real example of a dispute it has had to resolve, because that is the case the design has to serve. If you work for an agency and this is interesting, the useful next step is telling us where it is wrong.
Questions people ask next
- Does the regulator build exist today?
- No. It is the furthest out of the four builds. This page describes the shape of it so an agency can tell us where it is wrong before it is built rather than afterwards.
- Would licensees have to switch to Bloom for a state to use this?
- No, and that is the point of the ingest layer. A state cannot require its whole market to change software, so Bloom is being built to accept Metrc and BioTrack exports, the Washington CCRS format, and the arbitrary spreadsheet exports of whatever point of sale each shop already runs, and normalise them into the same records.
- What would a regulator be able to see?
- Filings by licensee, variances between what a licensee holds and what it has declared, recalls in progress and which licensees received affected packages, licence status and renewal dates, and inspection history. Read only against licensee data except where a state has statutory authority to act.
- How is this different from what Metrc already does?
- Metrc is the system of record for what moved. It does not hold what the rule says. Bloom's corpus records the rule, its source and the date that source was last read, which is what lets a filing be checked against the law rather than only against the previous filing.
- Who would own the data?
- The licensee owns their operating data and the state owns its regulatory record. That boundary has to be written into the contract before anything is built, not after.