The Barrett Agentic Story: Building 21 AI Companies on One Platform

In late 2025, I was building RATS.io, a stock analysis platform with 70+ API endpoints and an 84-page trading dashboard. The stack was FastAPI, Next.js, Supabase, and a lot of Python. I was deep in the weeds of financial data, and I started noticing something interesting.

Every time I talked to small business owners, friends who ran accounting firms, a buddy with a trucking company, my barber, they all described the same problem: they were drowning in operational overhead. Not the work they started their business to do, but the emails, the scheduling, the invoicing, the compliance paperwork, the customer follow-ups.

And when I looked at what software existed for these industries, it was either generic horizontal tools that did not fit, or expensive enterprise solutions designed for companies with 500+ employees. The small business with 5-50 people? Largely ignored.

That was the lightbulb moment. AI had gotten good enough to handle these operational tasks autonomously. And the infrastructure to build AI products had been commoditized. The only thing missing was someone willing to build purpose-built AI for industries that big tech ignores.

Why 21 Verticals Instead of One

The conventional wisdom in startups is to pick one market, dominate it, then expand. There is good logic behind this. Focus matters. But I had a different thesis.

When I mapped out what each vertical needed, a pattern emerged. Every single one needed the same core capabilities:

• Multi-agent AI with industry-specific personas
• SMS, email, and voice communication
• Compliance rules specific to the industry
• Billing and subscription management
• Knowledge base for business-specific documents
• Client relationship tracking

The 80% was shared. The 20% was industry-specific. If I built a shared platform for the 80%, I could launch a new vertical by writing a YAML configuration file that defined the 20%: agent personas, compliance rules, workflows, and brand identity.

This meant instead of building one company and spending years expanding, I could build the platform once and deploy it across 21 industries in months. Each vertical would be a real product with its own domain, landing page, agent team, and go-to-market strategy. But underneath, they would all run on the same engine.

The Build Timeline

What We Learned

1. The YAML-Driven Architecture Works

Our bet that verticals could be defined mostly in configuration paid off. Each vertical's YAML file defines its agent personas (names, expertise, communication style), compliance rules (what regulations apply, what checks agents must perform), workflows (intake, scheduling, billing sequences), and pricing (Stripe product IDs, tier limits). The platform reads the config and instantiates the right agents, rules, and workflows automatically.

2. Industry Compliance Is the Real Moat

Any developer can build a chatbot. But building an AI agent that correctly enforces DOT Hours of Service regulations, or knows when a CPA firm needs to send an engagement letter, or tracks IRS deadlines across 200 clients? That requires deep domain knowledge that cannot be prompt-engineered in an afternoon. Our compliance engine is the hardest part of the platform to replicate, and the most valuable part to customers.

3. Go-to-Market Is Per-Vertical

Even though the platform is shared, each vertical has its own go-to-market strategy. TruckFlow sells through cold outreach to small carriers. FirmFlow sells through CPA professional networks. CutsBot will sell through walk-in conversations. The product is a platform, but the sales motion is vertical-specific. This is the right trade-off: shared engineering, specialized sales.

4. SMS Is the Killer Channel

We expected email to be the primary communication channel. We were wrong. For industries like trucking, salons, and veterinary, SMS is king. Drivers, pet owners, and salon clients do not check email regularly. But everyone reads their texts. Our SMS agent interactions have 10x the response rate of email.

5. Small Businesses Buy Different

Enterprise SaaS sells through demos, pilots, and procurement processes. Small businesses buy through word of mouth, Google searches, and "can I try it right now?" experiences. Our self-serve signup with a free trial converts better than any sales call. The landing page to checkout pipeline matters more than the sales team.

Where We Are Going

Right now, TruckFlow and FirmFlow are our most developed verticals. The platform is live and processing real transactions. Our sales engine is sending outbound emails and handling responses automatically.

Over the next 12 months, we plan to:

• Activate 5 more verticals with full agent teams and live customers (BriefFlow, TaxBolt, ClosingBot, AuditBolt, HardHatBot)
• Build voice capabilities so agents can answer phone calls, not just SMS and email
• Launch the partner program for industry experts who want to operate a vertical with our technology
• Hit $10K MRR across all active verticals

The long-term vision is 21 profitable vertical AI businesses running on one shared platform, each serving an underserved industry, each operated by a small team (or a partner), and each generating real revenue from real customers who could not afford this capability before AI made it possible.

The Thesis

The biggest opportunity in AI is not building better tools for tech companies. It is bringing AI to the millions of small businesses in industries that Silicon Valley has never cared about.

That is what Barrett Agentic is building. One platform. Twenty-one industries. AI that actually runs your business.