| name | outbound-studio |
| description | Outbound Studio — a 7-agent autonomous outbound sales crew that finds SMBs with weak web presence, builds them a better site on spec, and sends targeted outreach. Use when the user asks about running the Outbound Studio pipeline, adding leads, triggering agent runs, or working with the prospect-to-close workflow. The system covers: Prospector (Maps + Exa), Strategist (SEO audit + dossier), Builder (site on Vercel), Auditor (quality gate), Packager (video + email), Persona Tester, and Conductor (notify + approve). |
Outbound Studio
An autonomous outbound sales crew that turns SMBs with weak web presence into paying clients — by building them a better site on spec before asking for anything.
Core Mechanic
Build the best version of their site → show it to them → let them claim it within 7 days → recycle the asset if they pass (rerun for a similar business in a different location).
The 7-Agent Crew
1. Prospector
- Input: vertical + geo (e.g.,
power_washing in Houston, TX)
- Output: Lead pack — businessName, domain, sector, contactEmail, contactPhone, opportunityTag
- opportunityTag values:
no_website | weak_website | outdated | not_ai_citable
- Tools: Google Maps, Exa
- Handoff: scored lead JSONL → Strategist
2. Strategist
- Input: Lead JSON from Prospector
- Output: Target dossier — business info, services, brand voice, top review themes, competitor comparison table, gap inventory
- Tools: Firecrawl (full-site crawl), Exa (competitor research), Browserbase (JS-rendered sites)
- Skills: SEO Auditor
- Handoff:
dossier.md + competitor_comparison.json → Builder
3. Builder
- Input: Target dossier from Strategist
- Output: Live Vercel URL, GitHub repo, asset manifest (schema + llms.txt baked in)
- Deploy pattern:
business-name.vercel.app
- Never free-code — use templates + design systems + components
- Tools: GitHub, Vercel, Cloudinary, Pexels, GenerateImage
- Design reference: awesome-design-md repository (60+ brand DESIGN.md files — match archetype to vertical)
- Handoff: site URL + asset manifest → Auditor
4. Auditor
- Input: Site URL + dossier + competitor list
- Output: Audit report — gaps_closed per item, competitor scores, customer-journey grade (clarity/trust/ease/confidence), responsive check at 375/768/1440, AI-citability checklist
- Quality gate: Pass (all gaps closed + agent swarm grade > competitor set) OR Fail (fix list → Builder)
- V2: Monte Carlo customer-journey simulation — agent swarms of target customer, iterate until site wins
- Tools: Browserbase, Exa
- Handoff: pass-flag + audit report → Packager
5. Packager
- Input: Site URL + audit report + contact info + brand voice
- Output: Email draft (subject, body, video embed, live URL, price, 7-day window), video walkthrough
- Video generation: hyperframes / website-to-hyperframes (NOT screen recording — generated from HTML)
- Handoff: email draft + video URL → Persona Tester
6. Persona Tester
- Input: Email draft + target persona context from dossier
- Output: Persona review — clarity score, trust score, confidence-to-reply score, red flags
- Loop: green-light → Conductor | fix-list → Packager
- Tools: Gmail (receive), Exa (persona context)
7. Conductor
- Input: Full package from all agents
- Output: Telegram notification to user with approve-to-send CTA
- Handoff: user confirms → email sends | user requests changes → loops to relevant agent
AEO Pitch (lock verbatim)
Right now, when someone asks ChatGPT for the best [service] in [city], your website is structurally invisible to it — dynamic-loading, no schema, no crawler allowlist, no llms.txt. We built this new site so AI agents can read it cleanly. The goal is to get you recommended first. It's a race, it takes time, reviews matter — but right now, you're not even on the track. This puts you in the race.
Quality Differentiator
I didn't stop building this until agent simulations of your typical customer ranked it ahead of every competitor. Here's the link.
Pricing (May 2026, verified from 11 sources)
| Offer | Price |
|---|
| Core build (5 pages, AI-optimized, deployed) | $1,997 / $2,497 |
| Premium build (conversion-focused + AEO baked in) | $3,997 / $4,997 |
| Maintenance retainer | $97/mo |
| SEO retainer | $297/mo |
| AEO Growth retainer | $497 / $597/mo |
| AI Quote Tool one-time install | $797 / $997 |
Framing rules:
- "Launch audit" bundled into month 1 — never billed as a setup fee
- Month-to-month cancellation on retainers
- DIY counter: Wix/Squarespace costs 40-80 hours of owner time = $2K-$6K hidden opportunity cost
Knowledge Libraries (Supabase)
components table: id, archetype, vertical_fit[], source, user_rating, last_used, win_rate
- Builder reads with preference-weighted retrieval: highest-rated + highest-win-rate wins
vertical_packs table: per-vertical research — industry norms, must-have sections, competitor archetypes, proven hooks, pricing benchmarks
Lead Statuses
prospected → strategized → built → audited → packaged → outreached → won | lost | recycled
Opportunity Tags
no_website: Business has no web presence at allweak_website: Site exists but is poor qualityoutdated: Site is old/not AI-citablenot_ai_citable: Site is technically functional but invisible to AI search
AI Search Stats (for pitch)
- ChatGPT: 900M weekly active users (Feb 2026, doubled in 12 months)
- Google searches per US user fell ~20% YoY 2024-2025
- 37% of consumers now start searches with AI tools (not Google)
- Perplexity: 45M MAU early 2026, 800% YoY growth
- 51% of consumers changed research habits due to GenAI
Retainer Positioning
$497-$597/mo AEO Growth undercuts Optimum Web ($490/mo SEO-only, no AEO) and HelloSEO ($850/mo) while bundling more verifiable deliverables: llms.txt, schema, AI-crawler allowlist, AI-visibility monitoring.
References
- See
references/agent-configs.md for per-agent system prompt starters
- See
references/vertical-pack-template.md for the vertical research scaffold
