Atlas Minds — Martell Framework Dashboard

Revenue-Generating

2 Clients — $1K collected, $3.5K/mo pending

Lexington Capital — Call Intelligence

AI call scoring + Notos follow-up extraction • 38 reps • Kixie → AssemblyAI → GPT-4o → Salesforce

$3,500/mo projected — NOT YET COLLECTING. Waiting on Kixie/SF creds to go live.

1

2

3

4

5

6

Live

▼

Stack

Next.js 16 • Supabase • AssemblyAI • GPT-4o • n8n (2 workflows) • Resend

Deployment

audio-analysis-webapp.vercel.app (Vercel + Netlify failover)
35+ API routes • 6 phases complete

Usage

~600 calls/day analyzed • 38 reps
Contact: Grizzly (VP Sales)
14 executions since Feb 23 launch (early testing)

Gate 6 — Scale with AI Agents (Current)

✓ Full async pipeline: Kixie → transcription → AI scoring → Salesforce delivery

✓ Daily digest emails (M-F 9 AM ET via n8n + Resend)

✓ Token-based usage metering (400 hrs included, $5/hr overage)

○ Waiting on Kixie webhook secret + Salesforce Connected App creds from Alex (CTO)

○ Strategic decision: Keep Notos or drop? (Grizzly lukewarm — has competing solutions)

○ Package as Atlas Minds case study for client #2

Next Action: Get Kixie + Salesforce creds from Alex → activate live pipeline → then PACKAGE THIS AS THE CASE STUDY. This is Atlas Minds' proof of concept.

GloboTech NYC — AI Website Manager

Natural language site editing • 3 sites (GloboTech, FIXR, BrightX) • Claude Haiku • JSON patch updates

$1,000 paid (one-time?) — Need to formalize recurring pricing

1

2

3

4

5

6

Live

▼

Stack

Next.js 14 • Supabase SSR • Claude Haiku 4.5 • Resend • 4 Vercel projects

4 Live Sites

manage.atlasmindspreview.com (dashboard)
globotechnyc.com • fixr.repair • brightx.repair

Usage & Revenue

Matt Leeb (matt@globotechnyc.com) — sole user
25/day, 500/month usage limits enforced
API cost: ~$1/mo (Haiku)
$1,000 paid. Need recurring pricing structure.

Gate 6 — Scale with AI Agents (Current)

✓ AI chat editing live (Claude Haiku, patch mode)

✓ Multi-site management with preview + deploy

✓ Usage limits + activity logging enforced

✓ Matt paid $1,000 — engagement is real

○ Convert to recurring pricing (monthly retainer for ongoing management)

○ Run E2E test post-Supabase migration (flagged 10 days ago, still pending)

○ Can this be productized for other multi-site businesses?

Next Action: $1K collected — now convert to recurring ($99-299/mo retainer). Run the E2E test pending since migration. Second case study after Lexington.

Built, Not Shipped

3 Projects — Closest to Money

Board of Directors

Multi-AI deliberation • 5 providers (Claude, GPT, Gemini, Grok, DeepSeek) • 6 debate rounds • Convergence synthesis

$0 — Stripe built but not activated. 1 signed user on Pro ($19.99/mo)

1

2

3

4

5

6

Deployed

▼

Stack

FastAPI • Vanilla JS SPA (63.5KB, 9 modules) • Supabase Auth (email + Google OAuth) • Stripe (4-tier pricing) • AES-256 key encryption • Resend email export

Deployment

board.rishonlgcy.tech (Render)
API: board-of-directors-api.onrender.com
Supabase: xkxmlayykdepvdixlfpb

Pricing (Built)

Free: $0 (5/mo, mid-tier only)
BYOK: $9.99 (unlimited, user keys)
Pro: $19.99 (30/mo, platform keys)
Unlimited: $39.99 (unlimited, platform)

GATES SKIPPED: Gates 2, 3, 4 — No customer validation interviews. No manual solve. No prototype testing. Built on spec. Zero evidence of product-market fit beyond 1 test user.

To Ship (Operational Tasks Only)

○ Create 3 Stripe price IDs (BYOK, Pro, Unlimited)

○ Set platform API keys as env vars on Render

○ Run final Supabase migration (subscription_status CHECK)

○ Push code to GitHub

○ Test checkout → subscription → webhook flow end-to-end

Next Action: Activate Stripe (1 day of work) → convert sbalogun7@gmail.com to first paying customer → THEN backfill Gate 2: get 5-10 target users to try it and give feedback before marketing.

Atlas Metis

RAG-as-a-Service • Multi-tenant • Hybrid search (pgvector + BM25) • Cohere reranking • 62 API routes • 3-tier dashboard

$0 — Production-ready code, no deployment, no customers

1

2

3

4

5

6

Local Only

▼

Stack

FastAPI • Next.js 14 • Supabase pgvector • Cohere rerank v3.5 • OpenAI + Gemini embeddings • Celery task queue
62 API routes, all E2E tested

Deployment

API: localhost:8001 • Dashboard: localhost:3001
Railway config exists (untested)
metis.atlasmindspreview.com (shell only)
Supabase: mvayqtfuucnqpkmilwza

Architecture

3-tier: Master Admin / Org Admin / End User Portal
Multi-tenant RLS isolation
Document ingestion: PDF, DOCX, TXT, CSV, audio, URLs
Cost tracking: $0.0061/query verified

GATES SKIPPED: Gates 2, 3, 4 — Zero customer conversations. Never solved RAG manually for a client. No prototype tested with real users. This is the most technically impressive project with the least market validation.

Blocking Issues

○ OAuth token refresh not implemented (Google Drive/Dropbox expire after 1hr)

○ 8 of 11 connector types not built (only Drive, Dropbox, Webhook)

○ Docker/Railway deployment untested

○ Zero paying customers or even beta users

Next Action: STOP building features. Deploy to Railway → get 3 orgs to try it free → learn what they actually need. Backfill Gates 2-4 with real user feedback before adding more code.

Harmonia CRM

GoHighLevel competitor • 26 modules • 198 components • 87 server actions • 34 DB migrations • White-label multi-tenant

$0 — Massive build. Zero users. Zero validation.

1

2

3

4

5

6

Deployed

▼

The Build (Massive)

Next.js 15 • Supabase • Drizzle ORM • 26 modules (Contacts, Pipelines, Inbox, Calendars, Campaigns, Funnels, Workflows, Invoicing, AI Assistant, White-Label...)
198 React components • 87 server actions • 34 migrations (8,076 lines SQL)

Deployment

harmonia.atlasmindspreview.com (Vercel)
Supabase: yinickcahkfxleblpcjc
Est. monthly cost: $125-325/mo to run

Reality Check

Architecture: 5/5 (world-class)
Feature breadth: 5/5 (26 modules)
Code wired up: 2/5 (schemas done, UI 40%, interactivity 20%)
Customer traction: 0/5
Validation: 0/5

GATES SKIPPED: Gates 2, 3, 4 — Built top-down from GoHighLevel feature checklist. Zero customer conversations. Zero manual workflows. Zero prototype testing. The textbook Martell anti-pattern: "fell in love with the tech, not the problem."

Martell Assessment

✗ No evidence anyone will pay for a GHL alternative

✗ No customer interviews (not even 1)

✗ No manual CRM management done for anyone first

✗ No prototype shown to potential customers

✗ Partner Brief says "Production Ready" — aspirational, not actual

✓ Sandbox mode works for self-testing

Next Action: FREEZE development. This needs Gates 2-4 before another line of code. Find 10 agencies currently paying $300+/mo for GHL. Ask: "What do you hate about it?" If nobody cares, archive this.

In Progress

1 Project

Atlas Argos

Lead Gen SaaS • Reddit intelligence + Google Maps scraping + AI website generation (Forge) • 114K lines backend

$0 — 1 client onboarding (Matt Leeb / GloboTech). No payment.

1

2

3

4

5

6

Building

▼

Two Products (Evolved)

Original: Google Maps scraping → Apollo enrichment → AI website generation (15 styles) → video walkthroughs → Cloudflare R2
Pivot: Reddit RSS intent scoring → CSV export → warm prospect lists
Pricing designed: Free / Trial / $29 / $99

Code

Backend: 95% (114K lines, 12 Python tools, 33+ endpoints)
Forge (AI sites): 80% (862-line design_ai.py, 15 styles)
Frontend: 40% (18 Next.js pages, missing signup/dashboard)
DB: 14 migrations (11 applied)

Validation

1 conversation (Matt Leeb) — not enough
Case study page built but no metrics
Reddit pivot: PRAW → RSS (Feb 2026)
Smoke test passes but no user testing

GATES SKIPPED: Gates 2, 3, 4 — Only 1 customer conversation (need 10+). Never sold lead gen manually. No prototype tested with real businesses. 114K lines of code before market validation.

Next Action: STOP coding. Talk to 9 more businesses about lead gen pain. Do you even know if people will pay $29/mo for Reddit leads? Validate before finishing the frontend.

Client Deliverables

3 Projects

AJM FIT

Personal training website for Anthony • Next.js 14 • Beehiiv newsletter • Live

1

2

3

4

5

6

Delivered

▼

Stack & Status

Next.js 14 + Tailwind + Framer Motion • Beehiiv newsletter (2,500 sub cap)
Email: anthony@ajmfit.com (Hostinger)
Live: ajmfit.com

Remaining

Pending: Welcome email automation in Beehiiv dashboard
Feature dev stalled since Mar 2, 2026

Next Action: Confirm Anthony is happy. Collect testimonial for Atlas Minds portfolio. Close engagement.

Spark Team Event

Lyric Sync Creator • Browser-based MP3 + lyrics sync • AI backgrounds (Hugging Face) • Standalone HTML export

1

2

3

4

5

6

Ready

▼

What's Built

2 versions: AI-enhanced (Stable Diffusion backgrounds) + basic (gradient)
Upload MP3, paste lyrics, sync with spacebar, export standalone HTML
Cost: $0.01-0.06/song for AI backgrounds

Decision Needed

MVP complete. Not deployed anywhere.
Is this a product or a one-off client tool?
ClickUp: list 901711321603

Next Action: Decide: product or one-off? If one-off, deploy to static hosting and close. If product, go back to Gate 2 — who else needs this?

Roshay Entertainment

Entertainment website • Static HTML + Tailwind CDN • Not deployed

1

2

3

4

5

6

Staging

▼

Status

Static HTML complete. Dev server works (node serve.mjs).
Brand assets organized. Walkthrough documented.
ClickUp: list 901711321608

Blocking

No live URL. Ready for Vercel/Netlify deploy.
Needs: client approval on design + domain setup

Next Action: Is this client engagement still active? If yes, deploy and close. If stalled, park it.

Dormant / Needs Triage

2 Projects

Atlas Hermes

AI email assistant for LIS • Gmail triage, drafting, auto-archive • Telegram notifications • GPT-4o

1

2

3

4

5

6

Dormant

▼

What Exists

Config file (hermes.config.json) with GPT-4o, Telegram API, polling strategy (30-min, 8-18 ET)
Configured for operations@legacyinsuranceco.com
Auto-archive domains for marketing/newsletters
Node.js framework, dependencies installed

Assessment

Internal LIS tool. Problem is real (email overload).
But: config only, no substantial source code visible.
Overlaps with Claudia V2 dashboard capabilities.
Decision needed: does this solve a problem Claudia doesn't?

Next Action: If Claudia handles email triage → archive Hermes. If there's a gap → start at Gate 2 (document the manual email workflow first).

mneme

Personal knowledge base • "Better than getrecall.ai" • Next.js scaffold only

1

2

3

4

5

6

Dormant

▼

What Exists

Next.js scaffold on localhost:3003
Credentials: shanmarric@gmail.com / Mneme2026
Created Mar 2, 2026 — no development since

Assessment

Overlaps with Obsidian (Obadiah) which is already the knowledge layer.
No defined problem. No target customer. Gate 0.

Next Action: Archive. Obsidian already fills this role. Don't start new knowledge tools without Gate 1-2 first.

Strategic Insights

💰 Revenue Reality

$1,000 total collected (GloboTech one-time). Lexington is $3,500/mo but not yet collecting — blocked on Kixie/Salesforce creds from Alex. Board of Directors has 1 signed user but Stripe isn't active.

Fastest path to revenue:
1. Unblock Lexington — get creds from Alex → go live → $3,500/mo starts flowing
2. Activate Stripe on Board of Directors (1 day → $20/mo minimum)
3. Convert GloboTech $1K one-time → $99-299/mo recurring

🚨 The Martell Pattern

7 of 11 projects skipped validation gates. This is the defining pattern. Massive builds (Harmonia: 198 components, Argos: 114K lines, Metis: 62 routes) with zero proof anyone will pay.

Dan Martell: "The biggest mistake AI founders make is they fall in love with the tech, not the problem."

The fix: Freeze all new feature development on Harmonia, Argos, and Metis until each has 10+ customer conversations documented.

📋 Case Study Gap

Atlas Minds' #2 goal is "sign first paying client." But you already HAVE paying clients — Lexington ($3,500/mo) and possibly GloboTech. The case studies aren't written.

Lexington: 38 reps, 600 calls/day, AI-powered scoring, Salesforce automation. That's a compelling story.

Package it. One case study page + one demo video = Atlas Minds has proof of value.

🧹 Portfolio Cleanup

Archive candidates:
• mneme — Obsidian already fills this role
• Atlas Hermes — If Claudia handles email triage, this is redundant

Decision needed:
• Spark Team Event — Product or one-off?
• Roshay — Client still engaged?

Reducing from 11 to 7-8 active projects = less context switching = more finishing.

⚡ Martell Priority Order — What to Do This Week

1. Unblock Lexington Capital (1 call to Alex) — Get Kixie webhook secret + Salesforce Connected App creds. This unlocks $3,500/mo. HIGHEST PRIORITY.
2. Activate Board of Directors Stripe (1 day) — Create 3 price IDs, set env vars, test checkout. First product revenue.
3. Convert GloboTech to recurring (1 call to Matt) — $1K collected, now set up $99-299/mo retainer.
4. Archive mneme + decide on Hermes (30 min) — Reduce portfolio noise.
5. FREEZE Harmonia, Argos, Metis feature work — No more code until each has 10 customer conversations. This is the hardest but most important one.