This page is password protected.
Autonomous SEO content at scale. AI agents that publish, optimize, and rank — while you sleep.
Scribe Walker — our autonomous publishing agent — has been running in production since January 2026. No human editing. No manual publishing. Every article gets full SEO metadata, internal linking, YouTube embeds, and schema markup.
This is the engine Scribe is built on.
| Tier | Articles | Price | Target Customer |
|---|---|---|---|
| Free | 10 on signup + 5/month | $0 | Everyone — the hook |
| Pro | 30/month (1/day) | $99/mo | Solo entrepreneurs, bloggers, local businesses |
| Business | 150/month (5/day) | $299/mo | Growing businesses, niche sites |
| Agency / Enterprise | Custom volume | Custom | Agencies, publishers, multi-domain |
🎯 The Hook: 10 articles generated on signup in real-time. User is hooked before they even think about pricing.
💰 Unit economics: AI generation cost per article is under $0.50. Even at Pro ($99/mo for 30 articles), margins are 80%+.
📈 Revenue target: 100 Pro users = $9,900/mo. 50 Business users = $14,950/mo. Mix of tiers + enterprise = $30K-$50K/mo.
🔒 Grandfather clause: Early adopters locked in at launch pricing.
Refer 1 friend → +5 free articles
Refer 3 friends → +20 free articles
Refer 5 friends → 1 month of Pro FREE
Anyone who signs up counts — free or paid. Every user becomes a growth channel.
Cost per referral reward: ~$2.50 (5 articles × $0.50)
Cost of 1 month Pro reward: ~$15 (30 articles × $0.50)
Value of each new signup: Potential $99-$299/mo customer
Every free user is a sales channel. Dropbox grew from 100K to 4M users in 15 months with this exact playbook.
Business name, type, city, website URL (optional). No CMS, no API keys, no tech jargon. A plumber can do this.
We instantly spin up joes-plumbing.tryscribe.co — their own live site. Articles start generating in real-time.
By the time they explore the dashboard, 10 fully optimized articles are already live. Real content. Real SEO. Their business.
Already have a site? We connect to their domain/CMS. Don't have one? Keep using the Scribe subdomain.
Scribe runs daily on schedule. Dashboard shows publishing activity, keyword rankings, traffic growth.
WOW FACTOR: The onboarding IS the demo. Zero technical knowledge required. Host first, integrate later.
People building affiliate/content sites who need volume. Currently paying $50-200/article to freelancers.
Dentists, lawyers, plumbers who know they need SEO but can't afford agencies ($3-5K/mo).
White-label content engine. They charge clients $5K/mo for SEO, we power the content for $299.
Product-related content, buying guides, comparison articles. Drives organic traffic without paid ads.
vs. Jasper/Copy.ai: They generate text. We publish and optimize entire sites autonomously.
vs. SEO agencies: $3-5K/mo for 4-8 articles. We do 20/day for $299/mo.
vs. Freelancers: $50-200/article, 1-2 week turnaround. We do it in minutes.
1,285+ articles published. Engine proven. SEO pipeline validated.
Multi-tenant architecture, user signup, subdomain provisioning, publishing engine, Stripe billing. Ship fast.
Real users, real content, real feedback. Iterate daily. Fix what breaks.
Product Hunt, X campaign, YouTube content. Target: first paying customers before month ends.
Agency white-label tier, performance analytics, 100+ users. April 1st surprise incoming 😏
Claude (Anthropic) for writing + research. Modular — can swap models per task for cost optimization.
WordPress.com API (proven), WordPress.org REST API, Ghost API. Extensible adapter pattern.
Next.js + Vercel (dashboard), Node.js (agent orchestration), MongoDB Atlas (data), Stripe (billing).
Keyword research, competitor analysis, schema markup, internal link mapping, meta optimization.
| Milestone | Clients | MRR | ARR |
|---|---|---|---|
| Beta (Month 1-2) | 50 free + 10 paid | $1.5K | $18K |
| Launch (Month 3-4) | 500 free + 75 paid | $12K | $144K |
| Growth (Month 6) | 2K free + 300 paid | $45K | $540K |
| Scale (Month 12) | 10K free + 1K paid | $150K | $1.8M |
Latest test: Nancy's Bakery — March 10, 2026 (EC2 worker, Claude.ai Max Pro + Extra Usage billing)
Previous estimates from March 4 were significantly understated. Updated with real billing data below.
| Component | Cost | Notes |
|---|---|---|
| Claude Opus 4.6 (Extra Usage billing) | $2.54 | 3 articles, ~1,592 words each. Research + writing + MongoDB ops. |
| DALL-E 3 standard featured images (1792×1024) | $0.24 | $0.08/image × 3 |
| Total per job | $2.78 | $0.93/article (all-in observed cost) |
Previous estimate (Mar 4): $0.14/job ($0.02 Claude + $0.12 DALL-E). This was based on incomplete billing data and significantly understated the Claude cost. The $2.54 figure above is from direct observation of Claude.ai Extra Usage billing during a clean test run.
| Metric | Measured (Mar 10) | Previous (Mar 4) |
|---|---|---|
| Articles per job | 3 (configurable per tier) | 3 |
| Avg words per article | ~1,592 | ~1,516 |
| Generation time | ~6 minutes per job | 4-6 minutes |
| Max concurrent jobs | 3 (per worker instance) | 3 |
| Cost per article | $0.93 | ~$0.05 (understated) |
| Scale | Customers | Jobs/Day | Articles/Day | AI Cost/Month (Extra Usage) | Infrastructure |
|---|---|---|---|---|---|
| Current (MVP) | 3 test | 3 | 9 | ~$250 | 1 EC2 t3.small |
| Beta | 50 | ~50 | ~150 | ~$4,170 | 1-2 VMs |
| Launch | 100 | ~100 | ~300 | ~$8,340 | 2-3 VMs |
| Growth | 500 | ~500 | ~1,500 | ~$41,700 | 3-5 VMs |
| Scale | 1,000 | ~1,000 | ~3,000 | ~$83,400 | 5+ VMs |
⚠️ These projections use Extra Usage pricing ($0.93/article). Direct API pricing or flat-rate subscriptions would significantly reduce costs at scale. See comparison table below.
| Tier | Price/Mo | AI Cost/Mo* | Gross Margin | Previous Estimate |
|---|---|---|---|---|
| Free (5 articles/mo) | $0 | ~$4.65 | Loss leader | Was: $0.23 |
| Pro (1 gen/day = 30 articles) | $99 | ~$27.90 | 72% | Was: 96% |
| Business (2 gen/day = 60 articles) | $299 | ~$55.80 | 81% | Was: 97% |
| Agency (5 gen/day = 150 articles) | $999+ | ~$139.50 | 86% | Was: 98% |
*Based on $0.93/article (Claude Extra Usage + DALL-E). Direct API pricing expected to improve margins. Testing needed.
[Mar 4, 2026] Key insight: DALL-E images are 85% of per-job cost. Biggest optimization lever if we need to reduce costs at scale.
[Mar 4, 2026] Architecture already supports horizontal scaling — MongoDB job queue + multiple workers. No code changes needed, just deploy more worker instances.
IMPORTANT: The $0.02/job estimate above was measured during early testing and appears significantly understated. Real-world billing data from Claude.ai Max subscription shows substantially higher costs. Below are three pricing models to compare.
| Metric | Measured |
|---|---|
| Articles generated | 3 |
| Total words | ~4,775 |
| DALL-E images | 3 (generated + uploaded to MongoDB) |
| Claude Extra Usage increase | $2.54 |
| OpenAI cost (DALL-E 3 standard, 1792×1024) | $0.24 ($0.08/image × 3) |
| True total cost | $2.78 |
| Cost per article (observed) | ~$0.93 |
| Generation time | ~6 minutes |
| Model | Cost / 3-Article Job | Cost / Article | Notes |
|---|---|---|---|
| Anthropic API (Direct) | TBD — needs testing | TBD | Claude Opus 4.6: $5/MTok input, $25/MTok output. Likely cheapest. Need to measure actual token counts. |
| Max Pro Subscription | $200/mo flat | Depends on volume | Unlimited usage within rate limits. Best if running high volume. Rate limits may bottleneck concurrency. |
| Max Pro + Extra Usage | ~$2.54 Claude + $0.24 OpenAI | ~$0.93 | Current setup. $2.54 Claude Extra Usage + $0.24 DALL-E ($0.08/image × 3). March 10 test data. May include failed retry overhead. |
| Tier | Price/Mo | AI Cost/Mo* | Gross Margin |
|---|---|---|---|
| Free (5 articles/mo) | $0 | ~$4.65 | Loss leader |
| Pro (1 gen/day = 30 articles) | $99 | ~$27.90 | 72% |
| Business (2 gen/day = 60 articles) | $299 | ~$55.80 | 81% |
| Agency (5 gen/day = 150 articles) | $999+ | ~$139.50 | 86% |
*Based on $0.93/article ($2.54 Claude Extra Usage + $0.24 OpenAI per 3-article job). Claude and OpenAI costs are billed separately. Direct API pricing expected to be lower — testing needed.
| Configuration | Cost/Heartbeat | Cost/Day (1h interval) | Cost/Month |
|---|---|---|---|
| Opus heartbeat (broken config) | ~$0.24 | $5.76 | $173 |
| Haiku heartbeat + lightContext (fixed) | ~$0.007 | $0.17 | $5.10 |
| Haiku as primary (forced test) | ~$0.002 | $0.05 | $1.50 |
Verified March 11: 4-hour idle monitoring confirmed $0.05 total cost (~$0.007/heartbeat). Production config: Opus primary for article generation + Haiku for heartbeat. 95% cost reduction vs original misconfiguration.
| Component | Cost/Month | Notes |
|---|---|---|
| EC2 t3.small | ~$15 | On-demand pricing, us-east-1 |
| Elastic IP | $3.65 | Free while attached to running instance |
| Heartbeat (Haiku, 1h interval) | $5.10 | 24 heartbeats/day × $0.007 × 30 days |
| OpenClaw gateway (idle overhead) | $0 | No cost beyond heartbeat when no jobs running |
| Total idle infrastructure | ~$20/mo | Fixed cost regardless of article volume |
| Activity | Cost | Notes |
|---|---|---|
| Each Opus conversation message | ~$0.35-$0.67 | Depends on context/history length |
| Email hooks (GPT-4o-mini) | Minimal per email | Adds up with high inbox volume |
[Mar 10, 2026] CRITICAL: The $0.02/job original estimate was wrong. Real observed cost is $2.78/job ($0.93/article) — $2.54 Claude Extra Usage + $0.24 OpenAI (DALL-E + API). These are billed separately (Anthropic vs OpenAI). May include overhead from failed retries. Direct API pricing needs separate testing to establish true floor cost.
[Mar 11, 2026] Idle cost ROOT CAUSE found: (1) wrong model name "claude-haiku-3.5" (doesn't exist, correct: "claude-haiku-4-5"), (2) openclaw config CLI double-quoted the allowlist key, so haiku was never actually allowed, (3) heartbeat silently fell back to Opus. Fix verified with 4-hour monitoring: $0.05 total ($0.007/heartbeat). 95% cost reduction ($173/mo → $5/mo).
[Mar 10, 2026] Three pricing approaches to evaluate: (1) Direct API — likely cheapest, full control; (2) Max Pro flat rate — best for high volume if rate limits allow; (3) Extra Usage — current setup, most expensive per-unit but simplest.
[Mar 10, 2026] OpenAI cost for 3 articles = $0.24 (DALL-E 3 standard at 1792×1024 = $0.08/image × 3). DALL-E is NOT 85% of cost as originally estimated. At $0.93/article total, Claude (~$0.85) is 91% and DALL-E (~$0.08) is 9%. Claude is the dominant cost driver.
💡 The Play: Instead of charging upfront, make Scribe invite-only and free. Create scarcity, demand, and virality. Raise investment on traction metrics, then turn on revenue levers as it scales.
Invite-only. Each user gets 3-5 invite codes. Can't buy in — you have to know someone. Creates FOMO, word-of-mouth, and perceived value.
Precedent: Gmail (2004), Clubhouse (2020), Superhuman. Scarcity = desire.
Users become the pitch deck. 500 active users publishing thousands of articles is more compelling than 12 paying customers at $299/mo.
Metrics that matter: DAU, articles published, domains served, retention rate, invite conversion.
Once traction is proven and investment secured, introduce premium tiers. Free users already see the value — conversion is natural, not forced.
Trigger: 1,000+ active users OR seed round closed — whichever comes first.
Cost to acquire each user: $0 (invite-driven, zero CAC)
Cost per free user/mo: ~$4.65 (5 articles × $0.93)
500 free users burn: ~$2,325/mo — manageable with seed investment
Upside: If even 10% convert to paid at $99+, that's $5K+/mo MRR from zero spend on marketing.
[Mar 13, 2026] Proposed by Taha. Hybrid approach: invite-only gate (exclusivity) + generous free tier (virality) + visible paid tiers (anchors value). When revenue switch flips, it's an upgrade users have been eyeing — not a surprise.
[Mar 11, 2026] Agency partnership channel: Mary Pedler / INPUT Global. Taha has existing relationship from Ferrum ($8K+/mo engagement). Two plays — reseller (white-label Scribe under their PR packages) or case study/endorsement. Warm intro advantage. Screenshots + links saved.
[Mar 8, 2026] AWS Startup Showcase = free distribution channel once Activate approved. Also unlocks Offers Hub partner discounts + potential $100K additional credits.
[Mar 8, 2026] Pro article count needs revenue discussion — 30/month currently listed, decide before launch.
[Mar 3, 2026] Product named "Scribe" — clean, battle-tested, says exactly what it does.
[Mar 3, 2026] Onboarding can't take hours. First articles must appear in minutes. The onboarding IS the demo. WOW factor is non-negotiable.
[Mar 3, 2026] Host first, integrate later. Subdomain becomes both a demo AND a working SEO site.
[Mar 3, 2026] Freemium model: 10 articles free on signup → viral referral loop → convert to paid.
