One system for the whole life of a business: AI agents, led by a CEO, find opportunities, test ideas against evidence, take projects through gates and follow the businesses already running. People decide.
Early access is by invitation.
One year at Halden Mutual, a fictional insurer: 412 ideas researched, 371 discarded with their reason and 4 new business lines under way.
Early access is by invitation.
In the panel
Three screens from the online panel, with illustrative data: the agent room, today’s steps with the CEO and a report requested in plain words.
Illustrative data — Fictional company, ideas and figures, shown for illustration only.
Agent room in Desk view: a shelf of four robots above an agent-by-project grid of solid and hollow cubes; the Analyst has a spotlight on project P-0112. Header figures: 1 agent active now, 9 conversations in 24 hours, 4 steps pending today.
Illustrative data — Fictional company, ideas and figures, shown for illustration only.
Today panel docked beside project P-0107: step 1 of 4, labeled “Overdue experiment”, says experiment E-0131 closed three days ago without a result, with the buttons “Yes, I have the result”, “Let the CEO prepare it” and “I need more time”.
Illustrative data — Fictional company, ideas and figures, shown for illustration only.
Custom report request: on the left, the conversation with a one-page request for the risk committee and a question back about the quarter; on the right, the real one-page PDF with key figures, stages, risks and discards.
Illustrative portfolio
For anyone who builds new businesses: founders, venture studios, consultancies, accelerators, corporate innovation teams, growing companies, holding companies, venture capital funds, private equity firms, search funds, family offices and asset managers. FounderFold keeps every new business idea on record with its evidence, takes each project through stage gates with kill criteria, and produces the report your leadership reads. Your people make every decision.
This is Halden Mutual, a fictional insurer, after a year of searching for new lines: 412 ideas researched, grouped by theme and niche. Discarded ideas fade but keep their reason. Ideas that pass a gate light up. Select any idea to see its record.
In FounderFold, every idea stays on the map: those that become projects, those still being tested, those on hold with a review date, and those discarded with their reason. In the panel, the “Universe” view draws your own portfolio this way.
Hover or tap an idea to see its record. Drag to move.
How to read it
Each one with its evidence and sources; those evaluated against Halden Mutual’s priorities carry a score per criterion.
Competition ended 128, small market 74, compliance 52. The reason stays on the record, so the same idea does not come back next year.
Paused with a written condition — a missing flood map, two unverified claims — and a review date.
Each one is running an experiment with a numeric success criterion agreed before it started.
Working-dog insurance already earns revenue. The other three are in Validation and Pilot, with their gates and kill criteria.
| Theme | Ideas | Chosen | In test | On hold | Discarded | Main discard reason |
|---|---|---|---|---|---|---|
| MobilityExploring | 74 | 0 | 4 | 5 | 65 | Competition |
| Small businesses and professionsProjects running | 82 | 2 | 3 | 4 | 73 | Competition |
| Housing and rentalProjects running | 46 | 1 | 2 | 3 | 40 | Competition |
| Pets and working animalsProjects running | 38 | 1 | 1 | 2 | 34 | Competition |
| Property and climate riskExploring | 52 | 0 | 2 | 4 | 46 | Competition |
| Farming and parametric insuranceExploring | 36 | 0 | 1 | 2 | 33 | Market |
| Devices and embedded insuranceExploring | 48 | 0 | 1 | 3 | 44 | Competition |
| Health and later lifeExhausted: nothing left alive | 24 | 0 | 0 | 0 | 24 | Competition |
| Digital assetsExcluded by the board | 12 | 0 | 0 | 0 | 12 | Compliance |
| Total | 412 | 4 | 14 | 23 | 371 |
FounderFold covers the full lifecycle of a business, from the first opportunity to a company that has to keep growing. Every idea keeps its sources, every decision its reason, and every project the figures that say whether it should go on.
FounderFold is the agentic OS for building new businesses: AI agents, led by a CEO, help you discover opportunities, validate ideas, build new businesses and grow the ones already running. Every idea stays on record with its sources and, when it is discarded, its reason; every project passes stage gates with kill criteria; and the reports leadership reads are built from those records. People decide. FounderFold applies the same method to founders, venture studios, consultancies, accelerators, corporate innovation teams, growing companies, holding companies, venture capital funds, private equity firms, search funds, family offices and asset managers, at the stage each one is in: many ideas on record, most discarded with a written reason, and the survivors taken through gates to results leadership can read. Agents propose; people decide.
Overview
Illustrative data — Fictional company, ideas and figures, shown for illustration only.Ideas
| Id | Idea | Status | Score | Discard |
|---|---|---|---|---|
| I-0217 | Usage-based insurance for e-cargo bikes— | In test | 7 | — |
| I-0198 | Embedded travel insurance for coach operators— | On hold | 6 | — |
| I-0176 | Flood alerts and micro-insurance for ground-floor shops— | On hold | 5 | — |
| I-0242 | Index insurance for vineyard frostcompliance | Discarded | — | compliance |
| I-0251 | Peer-to-peer gadget insurancecompetition | Discarded | 4 | competition |
Each stage has its own work and its own decisions. The record carries over, so what you learn in one stage is still there in the next.
Discover
The Explorer searches open sources for problems and markets that match what you are looking for, and records each idea with its links and the date it checked them. Add your own ideas with a short form, or bring a whole list in from a spreadsheet.
Validate
Every idea is scored against the same rubric, from 0 to 2 per dimension, with the evidence behind each score saved in its record. Experiments test what evidence cannot settle, with a numeric success criterion and a deadline agreed before they start. Then a person decides.
Build
A chosen idea becomes a project on a lifecycle for its type of business. Each stage ends in a gate a person decides, with the date and the reason, and kill criteria written in advance are measured from the figures you record. The CEO lines up the next step.
Grow
Bring in a business that is already running at the stage it has reached. A guided diagnosis conversation becomes an improvement plan, with experiments, ideas for new lines and child projects. Leadership follows it in the Board view and in reports it can take to the board.
Five reasons to keep your ideas, your decisions and the reasons behind them in one record your board can read, instead of slides, spreadsheets and memory.
I-0217 Usage-based insurance for e-cargo bikes
Illustrative data — Fictional company, ideas and figures, shown for illustration only.Added to the portfolio
new, undecided
New, undecided → under study
decided by the editor
Under study → in test
decided by the editor “Pilot partners confirmed”
Every idea in a portfolio is scored with the same rubric and the same version of it, so an idea from March and one from October can sit side by side. A score without evidence counts as zero.
A CEO coordinates and hands each question to the specialist that fits: market, finance, product, compliance and more. You can add agents of your own, with versions and a scope. Every quote they use is checked against your records.
Ideas, projects, experiments, lessons and the decision on each, with permanent IDs such as I-0398 and a history of who decided what, when and why. “All portfolios” shows every portfolio side by side.
Every plan ends in steps with an owner: an experiment with a deadline, a milestone with its evidence, a kill criterion to watch. Today shows the next one, and the figures you record show whether it worked.
The evidence that justified an idea follows it into the project, and the lessons from a closed project inform the next search. Nothing is retyped between stages, and nothing is lost when a team changes.
Four examples, one per kind of team. The organizations, names and figures are fictional.
A solo founder looking for a software business for independent physiotherapy clinics runs the Explorer, keeps 3 of 14 opportunities and tests the strongest with a paid-pilot experiment before writing any code.
Studio Larkspan scores twenty opportunities on one rubric, discards sixteen with their reason, runs experiments on four and chooses two to build. The partners see why each one survived.
Quillmere Advisory keeps one portfolio per client, each with its own criteria, and each client sees only its own. Research, comparison and the board report follow the same method for all of them.
Ferrand Packaging, a mid-sized manufacturer, starts with a diagnosis conversation, turns it into an improvement plan and tests two new service lines with experiments before it commits a team.
FounderFold turns a portfolio into a PDF in two layouts: a portrait report to read and file, and a landscape board deck with one idea per page.
Weighted value per year rose to €3.1M; two decisions need the board.
Decisions requested
Cutoff Sep 30, 2026 · Portfolio New lines 2027 · Page 3 of 24
The method is the same at every stage: evidence, gates, kill criteria and a report. What changes is who decides, what has to be proven and where the report goes.
Compare many opportunities on one rubric, stop most of them early and put the studio’s people on the few worth building.
Run each client’s opportunities, evaluations and recommendations on the same method, in a portfolio only that client can see.
Evaluate every startup in a cohort the same way, see its weak points and follow its milestones until it leaves the program.
Build new lines inside a company through gates leadership agreed in advance, with a report the board can act on.
Find where a running business can grow next and turn it into a plan of experiments, new lines and figures to follow.
Find out which business is worth building before you spend your savings or a year on it.
FounderFold is built so that nothing important happens without a person and a reason on record.
Agents create proposals. Choosing or discarding an idea, passing a gate or closing a project is always a person’s decision.
Ideas, projects, experiments and lessons have permanent IDs, such as I-0398 or P-0112. Each change records the date, the person and the reason.
Each organization sees only its own portfolios. Access is enforced in the database, not just in the interface.
FounderFold’s agents work inside your organization’s limits. You don’t need to bring, connect or manage an AI of your own.
Early access is by invitation. Tell us what you are discovering, building or growing, and we will set up your organization.