Founders Q&A: why we built Pragmon AI

My family has been in manufacturing for more than 150 years. I have seen how fast an outside shock reaches a factory's profit. Gas prices jump. A new tariff lands. A buyer switches to imports.

I spent a decade on strategy work for CEOs, most recently as a Principal at Boston Consulting Group in New York. Large companies pay for that work. A company with 50 or 500 million dollars of revenue makes the same calls with a finance spreadsheet and the trade press. Pragmon AI gives that company a strategy function of its own.

My family runs a small manufacturing business, so I know the problem from the inside. I build data systems. I did that at SEGA and other technology firms. Running strategy methods every month for many companies is a data problem as much as a strategy one.

Most mid‑sized companies have no strategy team to build it. Today they rely on gut feel, the finance team's spreadsheet, the trade press, a general AI chat and an occasional consulting project. Each covers part of the job. A general AI chat does not know the company. A consulting project ends.

Pragmon AI reads more than 1,000 outside sources and dates every fact. Financial models tie each change to profit. Keeping that running every month is a full engineering job. A client pays a fraction of one consulting project's cost for it.

Financial models built as formulas do the arithmetic. Each figure is checked against its source before a client sees it. In the app, any number opens to its source and the date behind it.

Pragmon AI labels its own estimates "Estimate", so a reader knows what was measured and what was worked out. Pragmon AI says it is an AI. Every recommendation goes to a named person, who decides.

The first analysis, and it is free. Pragmon AI reads the company and its markets and shows what the latest changes mean for profit, in margin points and cash profit. It arrives within ten working days.

We ask for last year's accounts and a short call on how the company makes money.

After that, the Operating Environment Review arrives each month. The team reads it in the app, opens any figure to its source, and asks Pragmon AI follow-up questions between meetings. Uploads or an ERP link through Pylon add more. The Business Performance Review is an add-on.

Each company is walled off. Every request is checked against the user's membership of that company before any data comes back. Traffic is encrypted with TLS 1.2 or higher. Every sign-in and every opening or export of a report is logged and kept for 13 months.

We never train AI on client data, ours or a vendor's. Anthropic and OpenAI do not train on our API traffic under their commercial terms. Uploaded ledgers are deleted when the analysis that reads them ends. Single sign-on and SOC 2 are planned.

Pragmon AI is the AI operating system for mid‑sized companies. We start with manufacturers in metals, chemicals, plastics, packaging and food. Each step adds a piece of the strategy function.

• Q4 2026Company File: one sourced, dated record of the company
• Q1 2027Alerts and decision playbooks
• Q2 2027Weekly check-in and Projects
• 2028Scenarios and a record of decisions

The rules stay the same as it grows. Every figure has a source and a date. A named person makes every call.