Proof, not promises.

The problem

BeanieBrew was growing, but its customer data lived in a sprawl of spreadsheets, notes, and memory. Orders, preferences, and follow-ups were easy to miss, and the business had no reliable view of who its customers were.

What we built

A custom CRM — designed around how the business actually works — that became the single source of truth for customers and orders. It automates the follow-ups that used to fall through the cracks and gives the owner a clear, up-to-date picture of the business at a glance.

The outcome

• One source of truth for customers and orders
• Automated follow-ups that no longer depend on memory
• A system the owner can actually run, without a full-time IT department

The idea

LearningPath applies the core principle behind The Manuel Group: guide, don't guess. Instead of dumping answers on a learner, the system walks them through understanding, one step at a time, checking comprehension as it goes.

How it works

The system uses agentic methods to plan a learning path, present the next right step, and adapt to how the learner responds — the same plan / execute / verify loop we build into every AI system for a client.

Why it matters

It is a working demonstration of responsible, agentic AI: a system that builds a person's understanding rather than bypassing it — exactly the posture we recommend when an organization adopts AI.

The challenge

Aviation repair is high-stakes, specialized, and hard to standardize. A generic chatbot is useless — and dangerous — in a domain where a wrong answer has real consequences.

What we built

A domain-specific AI system for RC Fontana's aviation repair work, built to operate inside the guardrails the work demands: the right context, the right sources, and human verification before anything is acted on.

Why it matters

This is the opposite of vibecoding. It is a specialized, high-stakes application that only works because of governance — the Four Protects in practice. It is the proof that responsible AI is not a constraint; it is what makes AI usable where it matters.