The 7-Product Founder Playbook: How I Run an Agentic Software Group Without a 200-Person Team

Every time I tell another founder Autothink Group ships seven agentic products, the next question is some variation of "how big is the team". The answer is always smaller than they expect, and the surprise is always the same. The default mental model for "running a software group" is the SaaS-era one: more products means more product managers, more engineering pods, more customer success heads, more office space. None of that is true any more. This is the anti-headcount manifesto.

The stack — deliberately narrow

Tooling sprawl is the silent operating tax inside most founder-led software companies. Every additional tool adds context-switching cost, integration cost, billing cost and security surface. The Autothink stack is intentionally narrow.

Lovable for product surface area and shipping velocity. The same platform powers every Autothink product and the Creative Marketing Group site. One mental model, one shipping pattern, one design system.

Supabase for backend, auth, storage, database and edge functions. The default backend assumption for every product. Replaces what would otherwise be five vendors.

Lovable AI Gateway for inference across multiple models. One billing relationship, multiple foundation models, no API-key-per-product sprawl.

Cursor and Claude for engineering velocity. The pair-programming bar is now an LLM, not another engineer.

SendGrid for transactional email. Boring, reliable, not the place to innovate.

Why no VC

The cycle-one default for a founder shipping seven products would have been a Series A inside the first three. The cycle-two answer is no, and the reasoning is structural. The model layer is being commoditised faster than any prior infrastructure layer in software history — see /insights/the-death-of-saas — so VC's traditional value-add (capital to win a model-layer arms race) is structurally less valuable than it was in the prior cycle.

Meanwhile, the Creative Agent agency funds the Autothink product portfolio. Real customers, real cash, real product feedback, no board pressure to ship a worse v1 faster. The compounding loop between agency and product is documented in full at /insights/owning-the-full-stack.

The operating cadence

Seven products is a cadence problem, not a capacity problem. The week is structured around it. Mondays are product reviews — every product, every week, no exceptions, capped sharply on time. Daily ship cadence runs across the engineering surface. Fridays are GEO content review for both Creative Marketing Group and the Autothink product surfaces. The full week-by-week breakdown is at /insights/speed-over-perfection-weekly-cadence.

The org-chart heresy

Headcount is no longer the proof of seriousness. Agent count is. The future agency blueprint at /insights/future-agency-12-people-200-agents applies the same logic to services — twelve people, two hundred agents. The same heresy applies inside Autothink. Every role is defined as the exception handler for an autonomous loop. If a role cannot be defined that way, the role is not yet AI-first.

What this means for other founders

The category-label window is roughly eighteen months. Inside that window, founders who ship a working v1 of an agentic product will hold the category label for the rest of the cycle. Outside that window, the labels are taken and the operator is competing for the number-two slot in someone else's category.

The market map — where the unclaimed labels still are — lives at /insights/agentic-stack-100bn-opportunity-map. The operating doctrine that sits behind every shipping decision is at /insights/kj-doctrine-10-rules.

If you are inside that window and want to compare notes, book a call.