Product Block 05 · Sellable today

AI Visibility / Citability Scanner

Show a business how it surfaces in ChatGPT, Claude, Perplexity, Gemini. Sell the fix.

Deploy: Existing — sellable today $99 audit setup $499/mo monitoring
Live demo Try the demo → https://scan.cafecito-ai.com/
Existing live product → GitHub mirror → Source in monorepo →

Best fit: Any service business discovered via search where being missing from AI answers = missing from the funnel. SEO firms, law firms, healthcare, recruiting, real estate, accounting.

⚡ Self-bootstrap · paste into Claude Code or Codex

AI Visibility / Citability Scanner — build it without writing code

Drop the prompt below into Claude Code or Codex. Block 05 already exists as a deployed worker (cafecito-ai.com/boostai/) — the agent's job is SALES not BUILD: pick a prospect + 2 competitors, run all three scans via the live worker, build a side-by-side comparison report, draft the cold pitch with the comparison as the lead artifact.

You provide

You provide: (1) a service-business prospect URL (law firm / accountant / recruiter / etc.), (2) URLs of 2 of their direct competitors, (3) the prospect owner email.

You get back

You get: 3 BoostAI scan reports, a hosted side-by-side comparison page, a 90-second Loom showing ChatGPT citing the competitor and not the prospect, and a draft cold pitch with the comparison URL.

Runtime & cost

Roughly 60 minutes wall-clock (mostly waiting on the 3 scans). ~$5 in API tokens.

📋 Copy the entire block below into Claude Code (`/plan`) or Codex 
You are running the AI Visibility / Citability Scanner sales loop (Block 05 in the Cafecito AI new-hire playbook). Full reference at https://cafecito-ai.com/new-hire/blocks/05-ai-visibility-citability-scanner. The scanner ALREADY EXISTS at cafecito-ai.com/boostai/. Your job is sales, not build. Use plan mode. Stop at every [GATE].

INPUTS YOU NEED FROM THE HUMAN (ask before doing anything else):
1. Prospect domain (a service business — must have 20+ Google reviews and a real website with 5+ indexable pages, otherwise abort)
2. URLs of 2 of their direct competitors
3. Prospect decision-maker email
4. Industry + city (so the comparison report uses the right benchmark queries)

ENVIRONMENT (verify, don't change):
- BoostAI worker is live at cafecito-ai.com/boostai/ — submit domains there
- Working dir: /home/eratner/cafecito-ai
- Cloudflare account: f7a9b24f679e1d3952921ee5e72e677e

THE PLAN:

STEP 1 — PRE-QUALIFY (5 min)
- Fetch prospect domain. Confirm: ≥20 Google reviews + ≥5 indexable pages. If not, abort and tell the human to pick a different prospect.
- Same check for both competitors.
[GATE 1 — show pre-qual results, ask "proceed with these 3?"]

STEP 2 — RUN 3 SCANS (15-20 min wall-clock, mostly waiting)
- Submit prospect to cafecito-ai.com/boostai/ with industry + city. Save the JSON output + email report URL.
- Same for competitor 1.
- Same for competitor 2.
- These run in parallel; total wait is ~10-15 min.
[GATE 2 — show 3 score summaries side by side, ask "proceed to comparison report?"]

STEP 3 — CUSTOMIZE THE OUTPUT FOR THIS PROSPECT (10 min)
- Use Prompt #1 from the block page ("Customize the scan output for the prospect's industry") with the prospect's raw scan JSON.
- Output: industry-specific stakes, 3 highest-leverage missing queries, 5 fixes (ranked), industry comparison, the cold-email lead sentence.
[GATE 3 — show the customized output, ask "this is the prospect-specific story?"]

STEP 4 — BUILD THE COMPARISON REPORT (10 min)
- Create /home/eratner/cafecito-ai/boostai-demo-<prospect-slug>/ with one HTML file.
- Hosted page: 3-column comparison (prospect, comp 1, comp 2) × rows (citability score per platform, sample queries, citation gaps, top 3 fixes).
- Use editorial palette + accent color matched to prospect's brand if findable; default to cream/ink/terra otherwise.
- Add as a route on the cafecito-ai worker: /boostai-demo/<prospect-slug> serves this static HTML.
- Deploy.
[GATE 4 — show live URL + screenshot, ask "ready for the cold pitch?"]

STEP 5 — RECORD THE LOOM (5 min, human action)
- Tell human: "Open ChatGPT (or Perplexity) and ask the exact query from Step 3 (e.g. 'best estate planning attorney in Coral Gables'). Screen-record the result for 30 seconds — show that the competitor surfaces and the prospect doesn't. Save the URL."

STEP 6 — DRAFT THE COLD PITCH (5 min)
- Use Prompt #2 from the block page ("Cold pitch using the scan results") with: prospect score, competitor score, exact query tested, comparison report URL, demo Loom URL.
- Output: subject line + 4-sentence email + 2-sentence WhatsApp.
- Subject must reference the specific competitor by name (Hook: FOMO not fear).
[GATE 5 — show the draft, ask the human to approve / send manually]

STEP 7 — ENABLE THE 14-DAY FOLLOWUP SEQUENCE (5 min)
- The BoostAI worker already has a sequence runner. Enable it for this prospect via POST /admin/sequence/enable with prospect_email + scan_id.
- Confirm Day 3 / 7 / 14 emails will fire automatically (use Prompt #3 templates).

STEP 8 — SHIP THE SUMMARY
- Single-line: "[BUSINESS] · score [N] vs competitor [N] · comparison: [URL] · pitch sent to [EMAIL] · sequence enabled."
- Append to /home/eratner/cafecito-ai/boostai-shipped.md.

DONE.

GUARDRAILS: never run scans on a domain without verifying it's a real business. Pre-qualify hard (>=20 reviews, >=5 pages) — empty scans erode the brand. Cost ceiling: $10 in API.
01Stack▾
  • Existing boost-ai worker
  • D1 boost-ai-db
  • Claude API
  • Resend
  • Cron Triggers
02Live references▾

Working production examples. Read the code. Steal the pattern.

  • BoostAI live Production scanner · cron every 6h · email sequences fire on Day 0 / 3 / 7 / 14
  • BoostAI dedicated Standalone brand domain pointing to same worker
03Day-1 plan▾

A real prospect. A real demo. A real outbound message — all before 5pm.

  1. 09:00–09:30 Pick a service business with a known competitor.

    Solo or small law firm, accountant, recruiting agency, real estate brokerage. Pick one where you can name 2-3 of their direct competitors.

  2. 09:30–10:00 Run their scan via the live BoostAI worker.

    Submit their domain to cafecito-ai.com/boostai/. Wait for the scan (5-10 min). Save the JSON output AND the rendered email report.

  3. 10:00–10:30 Run their top 2 competitors' scans.

    Same flow. Now you have THREE reports — your prospect + two competitors. The comparison is the demo.

  4. 10:30–11:30 Build a side-by-side comparison report.

    Custom HTML page. Three columns: prospect, competitor A, competitor B. Rows: scores per platform, sample queries where competitors show up and prospect doesn't, citability gaps.

  5. 11:30–12:30 Write the prospect-specific cold email.

    Hook: "Searched '<their service> in <their city>' on ChatGPT. <Competitor> showed up. You didn't. Here's the report." Don't lead with their score — lead with the competitor who beat them.

  6. 12:30–13:30 Send the email + a personalized Loom.

    Loom (under 90 sec): screen-record yourself running the same query. Show ChatGPT confidently citing the competitor. Show their company missing. Don't over-explain.

  7. 13:30–14:30 Set up the 14-day follow-up sequence.

    BoostAI already has the sequence. Enable it for this prospect. They'll get Day 3 (one specific gap), Day 7 (industry comparison), Day 14 (final offer + monitoring trial).

  8. 14:30–15:30 Pick prospect #2 and repeat.

    The whole flow takes ~5 hours per prospect on Day 1. By the end of Week 1 you should have 5-7 prospects in the funnel.

  9. 15:30–17:00 Build the follow-up call playbook.

    Day 0-2: cold pitch. Day 3-7: reply to email engagement. Day 14: ask for a 15-min call. Document responses by objection type.

04Best practices & gotchas▾

  • Never lead with "you're invisible in AI" — lead with a competitor who shows up.

    Why: Defensive framing puts the prospect in argue-mode. Competitive framing puts them in fix-it mode. FOMO closes; fear sells worse.

  • The audit IS the demo. Don't promise it — send it.

    Why: Service-business owners get 50 cold emails a week promising audits "if you reply." Sending the audit upfront moves the reply rate to 30-40% because the prospect has already received value.

  • $99 audit is a loss-leader for $499/mo monitoring. Treat it that way.

    Why: You're not making margin on the $99 (it covers Claude API + your time at break-even). Quote them together: "$99 to see where you stand, $499/mo if you want to actually fix it."

  • Monthly comparison reports drive renewal. Send unprompted.

    Why: Monitoring without visible progress = churn at month 3. The monthly report shows: "Last month you ranked X. This month Y. Here's why." Even when there's no change, the report justifies the spend.

  • Never sell to businesses that aren't already on Google. Disqualify hard.

    Why: AI engines pull from web sources. A business with no website, no GBP, no third-party citations has nothing to scan. Pre-qualify by checking they have at least 20 Google reviews and ~5 indexable pages.

  • For the cold email, name the exact query you tested.

    Why: "You're missing from ChatGPT" is hand-wavy. "I asked ChatGPT 'best estate planning attorney in Coral Gables' on May 10th — here's the response — you're not in it" is undeniable.

  • Re-scan monthly, not weekly. AI engine results are stable on the month-scale.

    Why: Weekly scans burn API budget and produce noise. Monthly captures real movement and matches the billing cycle, making value visible.

05Prompts (copy-paste)▾

Drop these into Claude Code. Replace the [BRACKETED] fields with the prospect's details.

Prompt 1 Customize the scan output for the prospect's industry

Adapts the generic scan output into industry-specific language + recommendations.

I have a BoostAI scan output for [BUSINESS NAME], a [industry] in [CITY].

Raw scan JSON:
[paste raw JSON output]

Customize for the prospect:
1. **Translate platform scores** into industry-specific stakes. (For lawyers: "ChatGPT cites Smith & Co for estate planning in Coral Gables before it cites you" — not "Your ChatGPT score is 31/100.")
2. **Identify the 3 highest-leverage queries** they're missing from. Should be queries their actual customers would type.
3. **Recommend 5 fixes**, ordered by impact: 2 high-impact, 3 quick wins (under 1 hour each).
4. **Industry comparison**: where does their score land vs typical [industry] firms in [city]?
5. **One sentence the cold email should lead with** — the most painful, specific finding.

Output as five labeled sections. Under 600 words. Conversational tone — written for the owner.
Prompt 2 Cold pitch using the scan results

Generates the cold email + Loom script using the customized scan output.

Write a 4-sentence cold email + 2-sentence WhatsApp follow-up to [OWNER NAME] at [BUSINESS NAME].

Setup: I ran an AI visibility scan on their domain + 2 competitors.
- Most painful single finding: [one sentence]
- Competitor who beats them: [name]
- Exact query I tested: [query]
- Comparison report URL: [URL]
- Their citability score: [score]
- Competitor's score: [score]
- Industry average: [score]

Email:
- Open with the specific query I tested + the result
- Name the competitor who beat them
- Link the comparison report
- Plain-text the audit + monitoring price ($99 + $499/mo)
- Ask one yes/no: "want to see what's missing from your site that's helping [competitor] get cited?"

WhatsApp:
- Acknowledges email
- 30-second Loom URL of the actual ChatGPT query showing the competitor and not them

No emojis. No "circle back."
Prompt 3 Day 3 / 7 / 14 followup templates

The three follow-up emails BoostAI sends if the prospect doesn't respond.

Write the three followup emails for the BoostAI sequence for [BUSINESS NAME].

Context:
- Their citability score per platform
- The competitor who beats them
- Their industry + city
- Single biggest gap (e.g. "no schema markup," "no FAQ page")
- Day 3 fix: [specific fix]
- Day 7 industry comparison: [where they rank vs avg]
- Day 14 offer: 30-day trial at $99 (instead of $499/mo) — risk-free.

DAY 3 EMAIL: One specific fix the prospect could implement themselves in under an hour to move their score 5-10 points. Show what good looks like. End with "Or I can set this up + monitor monthly: $499/mo. Reply YES."

DAY 7 EMAIL: The industry comparison. One paragraph: where they rank vs the [industry] average + the top performer in [city]. Observational. End with "Ready to talk?"

DAY 14 EMAIL: The trial offer. "I'll do the first 30 days of monitoring + a re-scan + the action plan for $99 instead of $499. If it doesn't move your score 10 points or more, I refund." Final-call energy, no pressure.

Each under 150 words. Subject line + body. Sign as [SENDER NAME] from BoostAI.
Prompt 4 Monthly retention email

The monthly progress report sent to active monitoring customers — the renewal-driver.

Write the monthly progress report email sent to active BoostAI monitoring customers for [BUSINESS NAME].

Inputs:
- Last month's scores per platform
- This month's scores per platform
- Score delta + trend
- New citations found
- New competitor movements
- Action items completed since last report
- Action items recommended this month

Structure:
- Subject: "[BUSINESS NAME] · AI visibility · [MONTH] update"
- Header: "Hi [first name]. Here's what changed in your AI presence in the last 30 days."
- Score table: 4 platforms × (last/this/delta) with trend arrow.
- Wins: bulleted list of new citations + score gains.
- Risks: bulleted list of competitor moves + score drops + cause.
- This month's recommendation: 1-2 specific actions with effort estimate.
- Close: "Questions? Reply directly. Or schedule a 15-min review: [Cal.com]."

Tone: factual, calm, observational. Not selling — reporting.

Output: complete HTML email template with [PLACEHOLDERS].
06Selling script▾

Discovery question (ask this first)

"When your customers ask ChatGPT or Perplexity 'best [your service] in [your city],' have you ever checked what comes up?"

The frame

AI search isn't replacing Google — it's adding a new layer above it. Customers who used to call after Google ranking are now asking ChatGPT, getting one answer, and calling that one. Whoever ChatGPT cites by default for your category in your city wins the next decade of leads. You either become that default or you're shopping at a disadvantage forever.

The demo play

Send the comparison report URL DURING the discovery call. They click. They see their competitor cited and themselves missing. The pitch effectively ends there — wait for them to ask "how do I fix it."

Objections

  • "We already do SEO."

    "SEO and AI visibility overlap about 40%. The other 60% is structured data, FAQ schema, citation networks, and content depth that AI engines weight differently. Your SEO firm is probably moving the SEO needle and ignoring the AI signal entirely. The audit shows you both."

  • "AI search isn't real volume yet."

    "It's 8% of total search volume in mid-2026 and growing 200% year-over-year. The cost of being early is $499/mo. The cost of being two years late is rebuilding citations from scratch while your competitor compounds."

  • "$499/mo seems steep for monitoring."

    "It's monitoring + monthly action plan + monthly progress report + quarterly deep audit. Closest agency offer is $1.5k/mo and they outsource the audit."

  • "Can I just do this myself?"

    "Yes. The audit is $99 to see where you stand. If you act on it yourself, that's the whole engagement. The $499/mo is for the version where you don't want to think about it monthly."

The close

"$99 today, you get the audit and the action plan. If you act on it yourself, that's the engagement — no obligation. If you want monthly monitoring + new actions + the progress report, $499/mo from then on. If your composite score doesn't move 10+ points in 90 days, I refund the monitoring."

07Pricing notes▾

Anchor on alternative cost — agency-built AI visibility programs run $1.5-3k/mo. Per-scan API cost: ~$1.50. Per-prospect monthly cost at scale: under $20 in API + email. Setup ($99) is loss-leader; margin is in monitoring. Bundle: combine with Block 06 (document AI search) for service businesses with internal corpus → $1.2k/mo combined. Audit also doubles as lead-gen for OTHER blocks — a law firm seeing they're missing from AI search often realizes they need Block 02 too.