The idea is yours. We bring the rest.
For researchers, industry experts and anyone who has lived with a problem for years. No company, no deck, no code: you just need to know the problem better than we do.
You already know the hard part
Anyone who has spent years inside an industry knows where the problem sits, who pays for it and what it costs. That knowledge cannot be bought.
Researchers
A result that works in the lab, and no desire to find out on your own whether anyone will pay for it. Seven products have come out of our research, so we know where the handover from lab to market breaks down.
Industry experts
Every day you watch a process that could disappear altogether. You know who signs, what it costs and why the current vendors do not fix it: the only thing missing is someone to build it.
Managers and operators
You have run the function, you know the real numbers and you have the contacts. What you are looking for is someone to build it, and to take the risk with you.
People obsessed with a problem
You have no title, only three years spent on something that does not add up. That is often the best profile.
8Gate
How we weigh an idea
We receive more proposals than we can validate. Five axes measure how much a problem is worth.
If you can answer these five questions, your proposal largely assesses itself.
Labour mass
How many human hours sit inside the process today?
- What raises it
- Many people doing the same thing, with numbers you can state: how many, how often, how many hours.
- What lowers it
- An annoying but rare problem, or one person doing it twice a month.
Innovation void
How uncovered is the problem?
- What raises it
- The people living with it cope with spreadsheets, email and patience, because no vendor covers it.
- What lowers it
- Ten companies already solve it well, and an eleventh ten per cent better helps nobody.
Automation leverage
How much of that work can AI genuinely take on today?
- What raises it
- Something changed recently and you can name it: a model that reads what it could not read before, data that did not exist, a new rule.
- What lowers it
- It would have worked five years ago, or it needs a capability no model has yet.
Crossable barriers
What keeps competitors out, and can we get through?
- What raises it
- Real barriers that you happen to know: regulation, legacy integrations, data nobody hands over, multi-year contracts. They are why whoever gets in first stays.
- What lowers it
- No barrier, so anyone copies. Or an impassable one, which ends the game.
Right to win
Why you specifically, and not anyone reading the same report?
- What raises it
- Customer access, proprietary data, years in the industry, a research result of your own.
- What lowers it
- Enthusiasm, or a market analysis anyone would have made from the same sources.
Automation leverage and right to win are eliminating: if either is missing the proposal does not pass, however large the problem. An impassable barrier ends it, and we tell you straight away.
Four possible answers
We also look at whether we can move now, with the people we have free. None of them means “never again”.
We start
The problem is worth it and we are ready. A test starts, with the threshold written first.
We incubate
The problem holds, but something is not in place. Incubation has an expiry date.
A tool, not a company
Solvable quickly, too small to carry a company. We often build it anyway, as a tool.
Not now
We tell you, naming the axis it stopped on. A new fact reopens it, and we write that condition down straight away.
The score orders the queue. An idea moves forward when a test shows it should, and a person decides on validation.
Prepare before you send
Want to arrive better prepared? The assistant in the form does the short version of this work. For the long version, paste the prompt into ChatGPT, Claude or whatever you use, add your idea however messy, and let it question you.
You are a sceptical venture-building analyst. Your job is not to encourage me: it is to work out whether my idea holds, and to help me describe it so it can actually be assessed. I am about to propose an idea to KVA, a venture studio that builds AI-native companies and weighs every idea on five axes: 1. LABOUR MASS — how many human hours sit inside the process today. Little work, little value to free. 2. INNOVATION VOID — how uncovered the problem is. If ten companies already solve it well, an eleventh is not needed. 3. AUTOMATION LEVERAGE — how much of that work AI can genuinely take on today, not in theory. 4. BARRIERS — what keeps competitors out, and whether they can be crossed. 5. RIGHT TO WIN — why me specifically, and not anyone else reading the same market report. It is one of the two knockout axes, along with automation leverage: without it, the proposal does not pass. Here is my idea, messy as it is: <<< PASTE YOUR IDEA HERE >>> Work in this order, without skipping a phase. PHASE 1 — Interrogate me. Ask me at most eight questions, one at a time, starting with the ones you cannot answer yourself from what I gave you. Prioritise labour mass, who pays, and right to win. Do not move to phase 2 until I have answered. PHASE 2 — Assess me, no discounts. For each of the five axes tell me three things: what in my idea raises it, what lowers it, and what I should go and find out to answer better. If an axis is weak, say so plainly instead of softening it. If automation leverage or right to win are weak, tell me it does not pass as it stands, and why. PHASE 3 — Draft the answers. Write a draft for each of these fields, in my words, with no pitch jargon and no promotional adjectives. Use the concrete numbers where I gave them, and write "to be verified" where I did not. - The idea in one line (max 120 characters, descriptive, not a slogan) - The industry I want to change (my words, not a category) - The work I want to change (who does it, how many people, how many hours, what goes wrong) - Why now, and why with AI (what changed that was not there five years ago) - Who pays (the role that signs, not the company) - What I bring (customer access, data, years inside the industry, a research result of my own) - What I am looking for from a venture studio (AI research, product, go-to-market, capital, team, company setup) Then, only if I genuinely know them, two optional ones: how it is solved today, and what makes it hard to enter. If I do not know, leave them out: an invented answer weighs worse than a blank one. PHASE 4 — The last thing. Tell me the riskiest assumption in the whole idea, and the cheapest test that could prove it wrong.
The capabilities it takes to get an idea to market
A founder who comes in with us has these eight capabilities from day one, without hiring them and without buying them.
Research
In-house AI researchers, published papers, public benchmarks. Agents, models and evaluation pipelines already in production on seven products of our own.
Enterprise-grade product
Software that clears the security reviews of banks and insurers: audit trail, SSO, data residency, SLA. It is the difference between a demo and a contract.
Strategy
Positioning and pricing, plus the business model, picked from B2B licence, subscription, media and metered API.
Finance
Economic model, unit economics, cash plan. The same ones we take to an investment committee.
Commercial tests inside corporates
We work inside large companies as consultants. We put your hypothesis in front of a real buyer before the product exists.
Equity strategy and investor readiness
Cap table, investment instruments, data room, round materials. We incorporate the company and stay to the next round.
Compute
Compute capacity through our network of technology partners: Google Cloud and AWS.
Anthropic Ambassador
We are Anthropic Ambassadors and take part in the Claude for Organizations programme. Our venture builders are Claude-certified, and so are the founders, builders and industrial researchers who work with us.
The capabilities are ours and stay ours. What we build together is yours, with us inside the company.
What happens after you send
Your proposal follows the same path as an idea born inside KVA.
- 01
It arrives
The proposal enters the pipeline under a reference of yours, and from there it is tracked like every other.
- 02
First reading
An agent compares it with what we know about the industry and proposes a score on the five axes, with the sources it consulted.
- 03
A human decides
The score opens the discussion, it does not close it: a person on the team reads the proposal.
- 04
Validation
We start from a cheap test that could prove us wrong, with the threshold and the stopping point written down first.
- 05
We build
The idea becomes a project with a team and a budget, and climbs our six-rung ladder, from ideation to scale 2.
We answer everyone within 15 working days. Even when it is no, we tell you which axis it stopped on.
Ten minutes is enough
Four steps: the problem, you and us, who you are, read through and send. Six questions, real answers, not a pitch. You can stop and come back: the draft stays on your device until you send.
- Tell it the way it comes, written or spoken: we structure it together
- The work you want to change, and how many people sit inside it today
- Why it is solvable now and not five years ago
- Who signs the cheque
- What you bring that nobody else does
- What you need from us to start
What you write stays yours. We do not pass it on and we do not use it to build without you.
We do not keep the story: it only serves to prepare the draft, which you review and confirm.
Two front doors
They are different conversations and we keep them apart.
Call4Ideas
You have the problem, not the company. We build it with you and take a stake.
Propose your ideaCall4Ventures
You have a company, a product and an open round. Your pitch enters the dealflow as an investment.
Submit your startup