Loyalty & Members · Offline receipts
Turn a paper receipt into verified loyalty points.
Reward the in-store spend you'd otherwise never see. AI vision reads the receipt, rules award the clear ones automatically or route the rest to a person or an AI teammate, and a fraud agent holds anything suspicious for review. Real spend earns real points; fraud and duplicates don't.
Selling on Amazon or TikTok Shop? Online orders have their own path — see Marketplace orders.
The problem
Most of your customers' spend happens where you can't see it.
In-store spend is invisible
A paper receipt is proof of a purchase your loyalty program never recorded — and it walks out the door.
Self-reported points invite fraud
AI-generated fake receipts spiked in 2025. Honor-system points reward whoever games the form fastest.
Manual review doesn't scale
Checking every slip by hand is slow and expensive — but auto-crediting everything hands fraudsters the keys. You need a middle path.
SoClaw · AI review teammate
Early accessLet an AI teammate clear the review queue — not a person, one item at a time.
When a receipt isn't clear-cut, hardcoded rules route it to review — and that review no longer has to be a person. SoClaw, Flash's autonomous AI teammate, can staff the queue — approving the clear ones, rejecting the bad ones, and escalating anything it's unsure about to a person.
It never touches the fast path: receipts the AI already reads confidently are still awarded automatically. SoClaw only works the items that would otherwise wait for a human — so the queue clears around the clock instead of piling up until someone gets to it.
Decides the way a reviewer would
For each queued item, SoClaw confirms the amount and awards the points, or rejects it — using the same guarded actions a person clicks. It can't invent a new way to grant points or bypass a check.
Escalates when it's unsure
A genuinely ambiguous receipt or a borderline order is handed to a person, not guessed at. Doing nothing and asking for help is treated as the right call, never a failure.
Writes down every decision
Each approval or rejection lands on the record with SoClaw named as the reviewer and the reasoning attached, so your team always sees what happened and why.
Off by default, instantly pausable
SoClaw does nothing until you switch it on, works within the limits you set, and stops the moment you pause it. You stay in control of the queue.
How it works
A photo becomes verified points in four guarded stages.
Flash reads the receipt with AI vision, automatically fixes garbled results, and allows extra time for large phone photos. Every stage is fraud-checked before a single point is awarded.
Duplicate-photo blocking
Every receipt photo is fingerprinted; the same photo can't be claimed twice.
Auto-approval only when it's clearly valid
Points are awarded automatically only when there are zero errors and the AI is confident enough. Everything else goes to review.
Daily limit that fails safe
If the daily-limit check can't run, the points are held for review — never quietly granted.
No double-processing
Built-in safeguards mean one receipt can never be processed or awarded twice.
When something goes wrong, the system tags it with a reason so it knows whether to retry — and automatically recovers any receipt that gets stuck.
Where it fits
The fastest way to start digitizing offline spend.
A receipt needs no POS integration — a member just snaps a photo — so it's the lowest-friction way to capture in-store purchases and prove the value before any deeper integration.
Shopping malls & multi-tenant retail
Reward purchases from any tenant under one roof. Members snap a receipt from any store inside the mall and earn the mall's points — without integrating a POS with dozens of independent tenants.
Validate offline digitization, fast
Start capturing real in-store purchase data in days, not quarters. Receipts let you prove the loyalty case on actual offline spend before committing to a heavier POS or ERP integration.
Franchises & distributed retail
When POS systems differ store to store, a receipt photo is the common denominator — one capture path that works across every location, brand and operator.
The third layer · risk control
Early accessAnd a fraud agent watches the points after they're awarded.
Reading and approval catch bad receipts at the door. The last layer catches abuse that only shows up across many receipts and members — receipt farming, household double-dips, refunds that should have pulled points back. A separate SoClaw fraud agent runs hardcoded rules and, the instant one fires, freezes the suspect points (unspendable, and paused from expiring) and opens a review: confirm and remove them, clear a false alarm, or escalate to a person.
Three layers, one offline pipeline
- 1
AI reads the receipt
Vision extracts the store, totals and date with a confidence score.
- 2
Rules route it — auto-award, a person, or SoClaw
Clear-and-confident receipts award instantly; the rest go to a human or the SoClaw approvals agent.
- 3
A fraud agent guards the points
Suspicious points are frozen for review — fraud can't cash out before you've looked.
Why it's different
Verified spend you own — not data harvested for someone else.
Receipt-aggregator apps capture purchase data for themselves; the member belongs to them, not your brand. Flash's receipt scanning is yours by design — the verified purchase and the member feed your loyalty program, with checks built to resist the 2025 wave of fake receipts.
Typical approach
Receipt-aggregator apps
Capture purchase data for themselves; the member is theirs, not yours.
Flash, by design
Receipt scanning that's yours — the verified purchase and member belong to your brand.
Typical approach
Honor-system points
Trust whatever the form says; easy to game with fabricated receipts.
Flash, by design
Duplicate-photo blocking, confidence checks and a fraud agent on the points confirm the spend.
Typical approach
On-site-only loyalty
Only counts actions on one storefront.
Flash, by design
Earns from the in-store receipts your program never saw — real offline spend, not just on-store actions.
AI & innovation
Verification as a feature, AI vision under the hood.
Where the category rewards self-reported activity, Flash rewards verified purchases — and uses AI where it earns its place: reading a crumpled receipt no simple software could.
AI vision reading
Reads the store, totals, tax and line items from a phone photo — with a confidence score that decides auto-approval.
Not locked to one AI
A different AI model could be swapped in without rebuilding the process.
Fixes its own mistakes
If the AI returns garbled output, Flash re-asks it to fix the result before giving up — so messy receipts still go through.
Limitations & how we handle them
AI reads most receipts. A person catches the rest.
AI vision is strong, but it isn't infallible — a torn or faded slip, a handwritten total, an unusual format, or a genuinely low-confidence read shouldn't be auto-credited. Flash is built to be honest about that: anything ambiguous is routed to a person before any points are awarded.
You set the confidence bar. Above it, points are awarded automatically; below it, the receipt lands in a review queue where your team confirms the store and amount in seconds. Real spend still earns — you just never auto-credit a misread.
And that queue no longer has to be staffed by hand: SoClaw, your autonomous AI teammate, can work it for you — making the same approve-or-reject calls a reviewer would, and handing the genuinely ambiguous ones back to a person.
Poor or unusual receipts
Blurry photos, faded thermal paper, handwritten or non-standard formats are flagged low-confidence and sent to review — not silently rejected.
Low-confidence reads
Auto-approval fires only when the AI is confident and there are zero validation errors; everything else waits for a human.
Fail-safe by default
If a safety check (like the daily limit) can't run, the points are held for review — never quietly granted.
Stuck items recover themselves
A typed failure reason decides whether to retry, and an automatic recovery sweep brings back any receipt that gets stuck mid-process.
What changes for the business
Spend you couldn't see becomes verified, owned membership — and your most-rewarded customers are genuinely your best.
AI vision
reads receipts in production, fixes garbled output
Duplicate check
blocks the same photo being claimed twice
Fraud-guarded
suspicious points frozen for review, not quietly spent
In-store receipts read, reviewed and fraud-checked into points.
Related reading
Keep exploring the pages most related to this one.
Behavior Capture
One live, unified profile — served through a governed semantic layer, the single interface the AI reads the business through.
Read more CapabilityPOS, ERP & OMS via API
Sync point-of-sale, ERP, and order-management data through governed API integration into the same customer record.
Read more CapabilityProduct Catalog & Industry Models
An AI-ready PIM that flexes to every vertical — serialized luxury items, apparel size matrices, F&B combos — synced from ERP/Shopify and wired to order line items for category-level personalization.
Read more CapabilityShopify
Auto-syncs paid Shopify orders into verified points and a member record, email-matched to in-store receipts, idempotent by order GID.
Read more CapabilityAmazon & TikTok Shop
Bring marketplace buyers into the member graph through verified Amazon and TikTok Shop order capture.
Read more CapabilityZero-Copy Data Sources
Query Snowflake, Databricks, Salesforce DC and Delta Sharing live — no ETL, no permanent copy, governed by scope and a per-query cost budget.
Read more