The receipts. What brands actually said, indexed and searchable.

The unique challenge

• Brand-perception dashboards are $50k/year, invite-only, and updated monthly.
• Wanted the same signal — "what is AI actually saying about my brand right now?" — for anyone who can spell their own name.
• Needed to feel like a receipt: date-stamped, exportable, unarguable.

The approach

• Query engine sends the same tracked prompts to multiple LLMs on a schedule.
• Renders responses as branded, screenshotable "receipts" you can drop into a Slack thread.
• Storage-friendly: heavy assets only for on-demand exports, everything else stays cheap.

The outcome

• Live at brandreceipts.com.
• A working paid product built on Lovable, no external SaaS spine.

The stack

Highlighted = the pieces beyond the standard Lovable stack.

Prompts used

The actual seed prompts.

Copy them. Adapt them. Ship faster.

You are a brand-communications analyst. Given this raw statement, extract: (1) brand, (2) topic tags, (3) concrete commitments (each with a verb + object + timeframe), (4) statement date, (5) 1-sentence neutral summary. Return strict JSON.

Given this new statement from brand B and the last 10 statements from brand B on overlapping topics, flag any commitments that contradict earlier ones. For each, cite both statement IDs and describe the contradiction in one line.