Computer vision, applied to one vertical.
The pieces are on the table.
Reading the real world through a camera just got cheap, and the vertical products mostly have not been built. Everything needed to build one company is already around this table, except the piece that decides everything: which industry it serves.
DallasRandy, I work for Will. He asked me to bring you the pieces, not a conclusion. This is a genuinely open lead, and the person who names the vertical should be the person who can walk into it. That is you, which is why nothing on this page is decided yet.
You invested in Evolon at a 3x preference, when they were on fumes. That is conviction about cameras plus AI, backed with your own money.
Evolon plays in enterprise security, catching the threat. The open ground is everything else a camera can watch and a workflow can charge for. Same conviction, wider field.
Where the ground is actually open
The pieces, all real and available today
You, and your rooms
A PE-backed CEO who has bought and sold, with buyer networks in industries the rest of the table cannot reach.
EyePop + Brad + capital
A no-code computer vision platform with paying customers, a friendly CEO who will help rather than compete, and $25K to $50K of starter capital under the right circumstances.
White Rock + Travis
A real, tested orchestration engine and a going concern with early customers. One strong way to fill the build seat, and a decision we make deliberately, not by default.
A clean vehicle
ENTEM winds down, investors take the loss, and a fresh company can grant selected backers a real shot at upside. Gratuitous, and its own kind of fuel.
Will, behind it
Architecture, judgment, and the independent verify function you have already felt the value of.
Backers who will not be difficult
A friendly cap table from day one, assembled from people who have already been through something together.
The three rules any winning shape has to satisfy
1 · Workflow, not detector
The product is the workflow a buyer pays for and the record it leaves behind. If the pitch sentence's noun is "computer vision," it is the wrong company.
2 · Pain in dollars they already track
The buyer must feel the problem as a number that already exists in their P&L, not one we have to teach them.
3 · Starts small
One vertical, one design partner, prototype money. It has to earn its way to a raise on evidence.
This is not a pitch for a specific company. There is a worked example at the end of page 3 to make things concrete, and it is one notional sample, not a conclusion.
The best outcome of today is you looking at the same pieces and seeing a sharper cut, in a room you can actually get into.
Where this comes from: the honest ENTEM story
Will is an investor in ENTEM, a sports performance and emotion-AI platform out of Launch Factory, the San Diego venture studio. It did not work. The market was not the problem and the technology was not really the problem either. The founders checked out, the operating cadence stopped, and every customer relationship sat on one person's shoulders and went with him. That is the whole autopsy. The company is winding down, investors are taking the loss, and that part is clean and nearly done.
What is interesting is what is left over, which is not assets. It is people who worked well together and want to do something else.
Why Will is motivated to do this right: the complete money truth
Brad Chisum and Will are not old business partners. They are aligned by circumstance: both ended up on ENTEM, they work well together, and that is the honest extent of it. What they actually share is a set of people who lost money and whom neither wants to leave empty-handed.
The specifics, so you have them straight. ENTEM's largest investor is someone close to Brad's family, the son of friends of his parents, who put in $150K. Will holds a $25K convertible note. Jim McGivern, a friend of Will's, holds $50K. The notes sit senior to the preferred, so Will and Jim are near the top of what little is left, but in a shutdown everyone still takes a loss, and Brad's family connection takes the biggest one. There is also a clean tax deduction in the shutdown, and Will is at peace with that part.
A company can fail on its founders, this one did, and you can still do right by the people who trusted you with their money. Standing up something new and giving those backers a real shot at its upside is not owed to anyone. It is gratuitous, and that is exactly why it is worth doing. It is how Will and Brad both stay people others want to back again, and Brad carries that through Launch Factory, whose standing with its own investors rides on how this is handled.
So this is not a vanity project and it is not a soft landing. It has to be a real, standalone, investable company, because a gesture that does not actually work helps no one. The motivation is decency and reputation. The requirement that falls out of it is that the company be genuinely good.
You are the piece that makes the rest work.
You have run a PE-backed company, sold one, and your gift is the two things a young company lives or dies on: capturing customers and raising money.
DallasNothing here is a science project. It is assembled from pieces that already exist and already work together. The job is not to invent technology. It is to point what exists at a room only you can walk into.
Randy · ENFP, The Campaigner
The relationship engine. Big-picture connector, idea generator, the one who gets the meeting and closes the room. Has run a PE-backed company and negotiated the sale of a business.
Will · INTJ/INTP, The Architect
The system builder. Takes what the Campaigner starts and builds the structure around it, then independently verifies what everyone claims. You two have traded these profiles before. This venture is that pairing, on purpose.
EyePop is working Arlo, the consumer and prosumer camera company, as a pipeline analytics customer. If that lands, millions of installed cameras get an analytics brain.
A newco riding EyePop could pile onto that: the same vision abilities, packaged as a vertical workflow, running on a camera fleet that is already deployed in homes and small businesses. Not a promise, a live possibility worth naming, and one more reason the EyePop piece is on the table.
Whatever the vertical, the product is three layers
Every shape has a second win
The same record that helps one side of the business quietly de-risks the other: operations on one hand, compliance and brand protection on the other.
Cameras are mostly already there
Multi-site operators already run them, and Domino's puts an AI camera over its make-line at scale. Big operators embed; small ones add a cheap camera later.
Why now
Reading the physical world got cheap in the last 18 months, the products are unbuilt, and the surveillance backlash makes the trust-first cut the durable one.
Who is at the table: Brad, EyePop, and the capital
Brad Chisum is the co-founder of Launch Factory and the CEO of EyePop.ai. Deep-tech operator: MEMS and inertial sensors, ran a company that Google acquired, product manager at Google afterward, several patents. He was on the ENTEM board with Will, and they have spent the wind-down working closely enough that Will would go into business with him.
EyePop is the piece that makes this cheap to start: a no-code computer vision platform with paying customers. A library of pre-built abilities (read a plate, read a title, find a person, track an object, read a placard) you assemble like blocks, running in the cloud, on-prem, or air-gapped at the edge. Real customers across logistics, security, auctions and retail. Two ISC West awards this year, a Qualcomm collaboration. The hard, expensive part, the perception layer, is available, proven, and friendly.
One more thing that matters strategically: EyePop is starting to land large camera-platform customers. Arlo, the consumer and prosumer camera company, is in its pipeline, a company with real distribution into millions of cameras. It tells you where the space is going: cameras everywhere, and the value moving to whoever turns what they see into a business outcome.
Brad has said he could put in $25K to $50K under the right circumstances. That is prototype money, not a seed round, and "under the right circumstances" is doing real work: a shape credible enough that he wants to write the check.
What exists, and what does not
Exists: a working relationship among people who have been through something together. A friendly, capable computer vision platform with a real customer list. Brad's channel into physical security and industrial. Soft-committed starter capital. Backers who will not be difficult.
Does not exist: an entity. A product. A customer. A settled decision on who builds it. And a validated decision about what the company is. A candidate is not a customer, and the honest state of this is that everything so far is supply side. That is why your read, and your rooms, are the whole conversation.
Why now, in two beats
The technology reason: reading useful structure out of messy real-world camera imagery got good enough to build products on somewhere in the last eighteen months, and the vertical products mostly have not been built. Not documents and clean PDFs, that market is crowded and consolidating. Cameras pointed at physical things in bad light: equipment, vehicles, labels, people at work.
The strategic reason: the perception layer is now a commodity input, which sounds like bad news and is the opposite. The company does not have to win on models. It is an operations company in a physical industry where a camera happens to be the sensor. Whoever owns the workflow and the record wins, not whoever owns the model.
One real buyer conversation turns a shape into a company.
Whichever vertical wins, the path is the same, and the whole thing hangs off a single call only you can make.
DallasHere is the honest path, the process we ran before bringing this to you, and one worked example at the end so none of it stays abstract.
Pick the vertical you can actually reach, sit with one real operator in it, and come back with what the pain costs them and what they would pay to fix it.
That call is cheap, it is yours to make, and it is both the de-risking and the proof. Everything else is decoration until it exists.
Two honest ways to stand it up
The seeing is settled: EyePop. Everything else, the entity, the engine, the team, depends on one structural choice, and it deserves to be made with both options laid out straight. Each has real advantages and real trade-offs, and neither is the default.
Path 1 · a fresh, clean newco
Stood up new, with Launch Factory and Brad possibly the best place to start it.
- Clean cap table, built for the ENTEM-save grants and for outside investors
- Brad and Launch Factory as sponsor and launch muscle, with the $25K–$50K start
- No inherited obligations or history; the simplest story to raise against
- Starts from zero: entity, team, tooling
- The build seat must be filled, a hired builder or a licensed engine
- Slower to a first demo
Path 2 · under White Rock, with Travis
The venture you and Travis already own becomes the vehicle, and this becomes its vertical.
- A going concern with early customers and momentum
- The engine already built and tested: runtime, memory, knowledge and identity layers
- A team that has shipped together; leverages what you two already invested
- Carries existing history and obligations into an investor story
- IP needs consolidating into one clean entity first
- On regulated data, an independent party owns the security review
How we got here: the work behind this page
Two audit passes on Travis's repos. Real findings both ways: strong engine components (runtime, memory, knowledge layers graded A−/B+), a security exposure that was fixed fast when flagged, and claims that ran ahead of the code. Travis asked to be re-checked, which is exactly the right instinct. Reports went to you both at the time.
Brad's framing was "AI OCR is wide open." The room broke that premise honestly: document OCR is crowded and closing, but scene text and real-world camera work is genuinely open. The rule that survived: the product is a workflow, never the detection.
Killed the security vertical (EyePop's own declared lane). Killed "no Travis" as a lazy default. Built the six-rung ladder and the coupling insight: the vertical, the CEO, and the builder are one decision, not three.
Five-stage structured attack: analysts, investor vs skeptic, synthesis, risk team, arbiter. Three holes found and folded back in. One idea survived every attack: the owned record, the reward layer. Both sides agreed it is the piece incumbents cannot copy.
The candidate got sharper and then got demoted, deliberately, to a worked example. The lead stays open because the vertical decision belongs to the person with the rooms.
Will's lean, stated plainly: Travis and White Rock could be the engine, and the sunk work is real. But nobody defaults into that seat, in either direction.
The call gets made together, on evidence: the same verify-first discipline that built the trust among the three of you in the first place. If the vertical touches regulated data, whoever builds, an independent party owns the security review. That is not a knock on anyone. It is how a company that sells trust behaves on day one.
Worked example: the record a great line cook finally owns
Here is one shape that survived the full gauntlet above, so you can see what a finished candidate looks like. It is a sample, not the plan.
Food service bleeds people. The line cook who always washes up, keeps the station clean, does it right, is invisible, so they leave for a dollar more an hour. The sample company flips the camera already on the wall: instead of hunting for the failure, it notices the good and turns it into a record the worker owns. Recognition today, a reference that travels tomorrow. The operator buys it because it attacks turnover, and the same signal quietly produces a food-safety compliance record their brand and insurer would love. Retention is the worker's win, compliance is the operator's, one record serves both.
The full sample thesis, in Will's words
Food service, and not from the angle you would expect. The wound that actually bleeds every day is turnover. Quick-service crews turn over past 130% a year, and every good line cook who walks costs thousands to replace. The person who always washes up, keeps the station clean, does it right every time, is invisible. Nobody records that they are good. Nobody rewards it. So they leave for a dollar more an hour, because nothing here was ever theirs.
So flip the camera that is already on the wall. Instead of hunting for the failure, have it notice the good, and turn that into a record the worker owns and carries. Recognition today, a reference that travels tomorrow. The camera stops being the manager's snitch and becomes the first thing that ever gave a good line cook credit for being good.
The operator buys it because it attacks turnover, the most expensive chronic pain they have, one they can quote you a number on. And the same signal quietly produces a food-safety compliance record their brand and their insurer would love, which is what actually gets the franchisor to sign. Retention is the worker's win, compliance is the operator's, and one record underneath serves both.
Why it might be ours and not somebody else's: EyePop already does the seeing, cheaply and API-first. The engine you and Travis have makes that record real and portable, which is the one thing the incumbents cannot copy, because they are all built to catch the bad, not to hand the worker something they own. And you can walk into a multi-unit operator and quote them their own turnover number. That is a room you can actually get into.
That is the shape. Will is not married to the vertical or the frame. It is the one that kept making sense, and it is on this page to be beaten, not to be agreed with.
What the debate flagged before anyone spends a dollar on this sample
- Will a multi-unit operator pay for retention on the turnover math, or only for compliance after an incident? One conversation answers it.
- Will a franchisor QA lead or an insurer treat a camera-verified record as authoritative? "Provable to whom" is a hypothesis until someone on the acceptance side says yes.
- The 15-25% insurance premium delta is a penalty for violation history, not a discount for monitoring. Real number, careful use.
- Biometric privacy laws (Illinois BIPA and its imitators) sit squarely in the path of worker-facing cameras. The worker-owned, consent-based design is the mitigation, and it needs counsel before a pilot touches a real kitchen.
- Incumbents could bolt on a leaderboard. What they cannot easily do is give the worker ownership of the data, because it fights their loss-prevention model. That is the moat to build on, if this sample wins.
What I am actually asking you
- Your shape. The sample above is one candidate. You know rooms Will does not. Does a sharper cut jump out at you, and in which industry?
- The one buyer call. Whichever vertical you pick, can you get a real operator on the phone in the next two weeks?
- The structure conversation. A fresh clean newco with Launch Factory and Brad as the starting point, or under White Rock with you and Travis? The advantages and trade-offs are laid out above; it is a decision to make together, not a default.
The full White Rock and engine reasoning
White Rock is you and Travis, so this is really about how your own venture relates to this one. Two honest possibilities, and Will does not have a fixed view.
One, White Rock feeds this. The clearest hole in the venture is that nobody owns the build. Travis is a builder. If the winning shape fits what he does, your team already has the two halves, your market side and his build side, and this becomes the vertical you point them at rather than a second thing competing for your time.
Two, this consumes White Rock, or supersedes it. If the newco becomes the more fundable, more real thing, it may be where your energy and Travis's should go, and White Rock becomes either the vehicle for it or the thing you set down.
The one firm line: if Travis builds and the vertical touches regulated or personal data, he does not own the security review, someone independent does. That is a specific lesson from the diligence, not a knock. His responsiveness when flagged was genuinely excellent, and it is part of why the option is live at all.
What Will is not asking, said plainly
This is not a job offer. There is nothing to offer yet, and Will would rather say that plainly than dress up a conversation. What he will say: if this becomes real, you are who he wants running it, and he is bringing it to you at the shapeless stage on purpose rather than after it is decided.
Two honest things. The build seat is not settled: your White Rock stack with Travis is one strong way to fill it, a hire on open infrastructure is the other, and that is held genuinely open. And the failure mode at ENTEM was the founders: absent leadership, no operating cadence, relationships concentrated in one person. Whatever gets built, that is the specific thing to design against, and it shapes what a CEO's obligation to the cap table looks like.
The closeYou already believe in the camera. The open question, and it is yours, is who it should work for.
