Module 01 · Close plan and up
Eight agents that draft the night shift.
You approve the morning.
Invoice capture, catalog enrichment, loyalty-gap recovery, promo ROI, vendor credits, price intelligence, POS migration and the briefing that ties them together. Every draft is tied to a Metrc tag, gated by your state's rule pack, and waits in the approval queue.
- Invoice capture · MVP
- Morning briefing · MVP
- Vendor credits · MVP · MI
- Catalog enrichment · V2
- Promo ROI · V2
- Loyalty gap recovery · V2
- Price intelligence · V2 · licensed data
- POS migration · V3
01 · Invoice capture
Invoice to draft bill in minutes, checked tag by tag.
Reads the invoice inbox, extracts lines, checks every package tag against the accepted Metrc manifest, matches SKUs with confidence, proposes new SKUs under your naming and pricing rules, and drafts the bill. A tag not on the manifest holds the line and drafts a corrected-invoice request. Wholesale tax booked in cost.
- Captured
- 2m 14s
- Reviewed
- 1m 50s
- Lines · held
- 6 · 1
- Executes as
- Draft bill → AP rules
02 · Morning briefing
One page at 06:30. What changed, what's waiting, what looks wrong.
Ties every overnight proposal, anomaly and compliance clock together on your phone. Each item cites its sources and as-of times. Superseded items can’t be approved. The footer always reads “Drafted N · executed 0”, and there is no approve-all. Push and email carry titles and counts only; details stay in the app.
- Waiting for you
- 5
- Anomalies
- 3
- Clocks
- NY 7-day · 9h 40m
- Footer
- Drafted 5 · executed 0
03 · Vendor credits
Your deals sheet, cross-checked against Metrc-proven units.
Reads the sheet you already keep, merges renamed and overlapping deals, flags zero-redemption promos, excludes out-of-window sales, and drafts per-brand report packs with the email trail hashed. The state gate runs first: MI allowed on uniform terms, NY blocked, CA routed. If the brand is on BudAlly, the pack becomes a promo claim and the credit nets in settlement.
- Deals read
- 14
- Reports drafted
- 7 · $3,044
- NY / CA
- Blocked / routed
04 · Catalog enrichment
Descriptions and images with a license on file and a linter in front.
Drafts copy only from brand content kits and factual fields; images only with an active source license (no web search, no stock). Terpenes and potency per package from Metrc lab results. Health claims, kid appeal and “free” are blocked before a human sees the draft. Effects vocabulary off by default. SEO measured from your own Search Console against unchanged pages.
- Drafted
- 214
- Pass linter
- 171
- Flagged · no license
- 29 · 14
05 · Promo ROI
The counterfactual, with a confidence interval, before the budget meeting.
Full promo cost, rule leakage (discounts that fired outside their weekday or category), matched-cohort repeat rates with a 95% CI and an “observational, not causal” label, retention curves, and a recommendation that proposes a holdout test rather than a budget change. Medical tickets excluded; k ≥ 8 per cohort.
- Leakage
- 21.7% · $18,740
- Repeat, discounted − not
- −3.4 pts (−6.3, −0.5)
- Proposal
- Hold +$120k; test
06 · Loyalty gap recovery
Find the buyers you can't reach — then only reach the ones you may.
Cross-references POS buyers with members, dedupes across stores, scrubs placeholder emails, and splits the rest by consent proof: marketing consent with proof gets an email invite with a holdout; transactional-only gets a receipt line; no record gets a counter prompt. No SMS without your own 10DLC. No auto-enrollment, no retro points, medical profiles excluded.
- Non-members
- 14,280
- Invitable with proof
- 1,210
- Top 5% of non-members
- 31.9% of their revenue
07 · Price intelligence
Out-the-door, matched by brand, size and form, floored at landed cost.
Sources: licensed market data, your own in-store checks, and a k ≥ 8 network median from Metrc receipts. No scraped menus — uploads with scraper provenance are rejected. Fuzzy SKU matching with a review band, category index, a gap dot-plot, and repricing drafts that never go below landed cost plus your floor. Executes as a POS draft where a write surface exists, else a StoreHome task.
- Matched SKUs
- 138 of 410
- Basket index
- 112
- Drafts
- 3 price changes
08 · POS migration
Any POS to any POS, with consent proofs that survive the move.
Legacy import, customer crosswalk, consent carry-over (an opt-in without proof is imported as unknown), loyalty relink, dry runs, pre/post benchmark and a cutover that needs two approvers. Points never lived in the POS, so they don’t move. Metrc history already in the ledger is the continuity backbone.
- Customers mapped
- 8,450 · 180 review
- Opt-ins with proof
- 1,780 of 2,140
- Cutover
- 2 approvers
What every agent has in common
Sources with as-of times and a taint label (Metrc, POS, licensed, own check). Rules applied, by ID. Confidence with a method. A precondition hash. Executed count: always 0. Learned rules only after approval. Aggregates never below 8.
© 2026 BudAlly, Inc. · Figures are demo-tenant fixtures.