Find a real need.
Kai and Piper look for needs many people already pay for, and for places where the leading products fall short.
How It Works
Fourteen AI employees.
A human making the calls.
Our AI employees find needs, check every idea, build small products and learn from what real people do. The founder approves anything involving money, launches, prices or videos.
Select anyone on the team to see what they look after.
Who works with whomAn illustrative handoff
See the work
behind a decision.
Imagine a small file-conversion tool for creators. Follow one possible path from a need to a working beta and real use. This is a made-up example, not a live product or customer case.
Illustrative scenario / 04 stepsNo customer results claimed
Kai · AI employee / FindFind the missing tool.
Imagine many creators paying for a clumsy way to batch-convert image files into a required format, and complaining about the leading tool. Kai and Piper would record the requests, what people pay today and where they look for tools.
What moves forward
A specific task, source-linked observations and a possible route to users.
Decision gate
This is a made-up example, not an announced SUNVEX product.
01 / 04
Five steps
Good ideas get
questioned.
Research narrows the question. A small working product answers what research cannot. What real people do decides the next step.
Question it independently.
Every idea is checked by a second, separate AI system. Nora asks: is it true, and is it allowed? Rook asks how it could fail and finds the cheapest test. Sage asks what the results say.
Go, park or stop.
Grant decides, with one line of reasoning. The founder approves anything involving money, launches, prices or videos.
Make the smallest real version.
Isaac and Iris build it. Cole helps people find it, and Eden answers users. Success and stop lines are written before launch and never moved.
Keep the lessons.
Sage reviews the work every week. What we learn is written down and kept, so the next idea starts from it.
Evidence standard
A hunch is a start.
A sale tells us more.
For online products, these levels distinguish early signals from real usage, payment and repeatability.
- L1Observed signal
A real complaint, workaround or expensive friction.
- L2Independent corroboration
The same problem appears in independent sources.
- L3Direct confirmation
Real target users complete the task the product is designed for.
- L4Commercial confirmation
Users choose a priced option or start a checkout.
- L5Commitment
Users pay for the product.
- L6Repeatability
Paying users return or keep paying.
Public research alone cannot exceed L2. A visit, install or sign-up is not useful usage or payment. Every result needs a source and a measurement period.
Human authority
The final call
stays human.
Everyone on the team except the founder is AI. They do real work, but none of them can make a promise on behalf of SUNVEX.
AI employees work within set permissions and spending limits. The founder approves public launches, prices, payment collection, commercial commitments and every video. No product has launched yet; Eden, our AI customer support manager, joins when the first one does.
The thinking behind SUNVEX ↗