A portfolio of what’s next.
SUNVEX Ventures is our first internal project — a portfolio of AI-driven opportunities, from early exploration through to real-world deployment. Nothing here is advertised as finished before it is.
Four stages, and a bar between each.
A venture only moves when evidence says it should. Most do not move, and publishing the stage rather than the pitch is how that stays honest.
A problem has been sourced and written down. Nothing is proven, and most things stop here.
The case survived a red team. Assumptions are being tried against real work.
Evidence cleared the bar. A minimum credible product is being built.
In production with real users, real support and real failure modes.
Small Business Operations
Taking what already runs my own admin — email triage, documents in and structured data out, follow-ups that chase themselves — and building it for other small operators. It is the venture with the least mystery attached: I know the problem is real because I had it, and I know what the fix costs because I paid for it.
- Stage
- Exploring — problem statement written, evidence still being gathered.
- Open question
- Whether what fits one operator’s week generalises to another’s.
- Next gate
- Operators outside my own network confirming the loss is real and quantified in their own numbers.
Vertical Research Tools
Domain-specific research workflows for people whose job is to read faster than they can. Built narrow on purpose: a general research assistant competes with everyone, a workflow that knows one domain’s sources and conventions competes with almost nobody.
- Stage
- Testing — assumptions survived the red team; being tried against real work.
- Open question
- Whether the time saved is large enough that anyone changes habit.
- Next gate
- A user running their real workload through it for a fortnight, unprompted.
SME Growth Platform
The layer above the first two: taking what the operations and research work proves and turning it into something a small business runs on rather than tries out. Scope is deliberately unfinished — it is shaped by what V-001 and V-002 learn, not decided in advance.
- Stage
- Building — minimum credible product in progress.
- Open question
- Where the boundary sits between a tool and an operating layer.
- Next gate
- A paying customer running a real process end to end.
Hermes & Eden
The in-house pair. Hermes takes enquiry intake and follow-up, and is what answers when you use the contact form. Eden, its chat counterpart, ran on this site too and is off the front door for now while the shape of first contact gets another think. Between them they are the only things here that have met real traffic — and the only ones whose failure modes were paid for by us.
- Stage
- Operating — Hermes is live on sunvex.me; Eden is built and currently off the front door.
- Open question
- How much of this generalises beyond our own front door.
- Next gate
- Running the same stack for someone else’s business.
Real commercial evidence earns the next step.
Interest is not evidence. A warm conversation, a signed letter of intent, a waitlist — none of these move a venture forward on their own, because none of them cost the other party anything.
What counts is behaviour. Someone paying, someone switching off the thing they used before, someone putting their real work through it when nobody is watching.
The consequence is that this portfolio will look slow. Stages are published as they are, including the ones that stall, because a portfolio that only ever moves forward is not being honest about its own bar.