The Forge AI

The platform I operate on: quote from a phone inside the client's shop, follow the client from first contact through to payment, and stand up the agents that get delivered.

Why it exists

I sell in the field, walking into businesses. In that conversation you can't say "I'll send you the quote tomorrow" — by tomorrow the owner has cooled off. The price has to come out right there, in front of them, and it has to be right.

Everything here follows from that constraint.

The pieces

• Quoting engine with a price book. Prices live in one place, in the database, and the quoting engine, the manual and the proposals all read from it. Editing a price means editing it once, not chasing five copies that already contradict each other.
• Cost guard. Before a price is committed, the engine calculates what that client's infrastructure will cost and warns if the package eats the margin or pushes past the providers' free tiers.
• Visit route. Orders who to visit by what makes that visit worth making today — not by a stale score that no longer reflects anything.
• Field manual. The sales method written down and published: how to walk into a business, the eight owner profiles, the question that decides the visit, how payment works, and the lines that don't get crossed.
• Agent engine. What actually gets delivered to the client: WhatsApp agents wired into their inventory, their CRM or their spreadsheets.

Cost decisions

The business operates in Venezuela, and a client paying $90 a month can't carry a $40 infrastructure bill. The architecture came out of that:

Real operating cost: between $0 and $15 per client per month. That number is what leaves margin in the prices above.

Scale

82 commits on the platform itself, plus the product and client repositories that hang off it. Quoting engine, panel, route, manual and agent engine in one system.