Fractional VP of Sales for AI and computer vision companies
Every enterprise wants AI. Very few will sign for yours.
AI founders don’t have a demand problem. Buyers take the meeting, love the demo and agree to a pilot. Then the pilot runs for nine months, a privacy review stalls it, and nobody can say who owns the budget for year two. I have sold AI and computer vision into enterprise retail, most recently as VP of Sales for a computer-vision company selling store analytics, and spent 30+ years before that selling emerging technology to large, careful buyers. This page is how the AI sale actually closes, and what I do inside it.
Based in Scottsdale, Arizona. I work in person across Greater Phoenix and remotely with founders across the US.
The pilot was a success. The deal is still not signed.
AI deals rarely die in the demo. They die in the space between a promising pilot and a production contract, because the pilot proved the model and nobody proved the business case.
- No one owns the outcome. Innovation teams sponsor pilots. Operations, marketing or finance own production budgets. If those are different people, the pilot has no buyer.
- The metric was never agreed. “Better accuracy” is not a reason to sign. A number the business already reports — shrink, conversion, labor hours, dwell time — is.
- The review arrives late. Privacy, legal and InfoSec meet the product after the pilot, and they have questions about data, video, models and retention that nobody prepared for.
- The category is unfamiliar. The buyer has no budget line for what you do yet. Someone has to create one, and it is not going to be the person who liked the demo.
None of this is a technology problem. It is a sales problem, and it is fixable before the next pilot starts.
Position against a metric the buyer already reports.
The AI companies that scale stop selling capability and start selling a measured outcome. At a computer-vision company I built the positioning around a single engagement metric that brought digital-style analytics into physical stores, instead of selling cameras or people-counting.
What that positioning work looks like
- Pick the one number your product moves that a VP already answers for in a quarterly review.
- Build the business case in their units, at their store count or user count, before the pilot.
- Score yourself and the competitors against an independent standard, so the buyer’s analysts can check your claim instead of trusting it.
- Say which use case funds the first contract and which ones expand it. Lead with the one that has a budget today.
- Never position as a point solution if you are a platform. Never position as a platform if the buyer needs a point solution this year.
When I scored that company’s whole vendor field against the industry’s published attention-measurement guidelines, it changed the conversation with every analyst on the buying side. That is the kind of proof AI buyers need.
Start the review before they ask for it.
For computer vision, voice, biometrics or anything trained on customer data, the review is the sale. Founders who treat it as paperwork lose quarters.
- A data packet on day one. What you capture, where it lives, how long you keep it, who can see it, and what the model learns from it.
- Privacy by design you can show. Anonymization, edge processing, opt-outs and signage, in plain language a legal team can approve.
- Model governance. How you test for bias and drift, who signs off on changes, and how a customer audits it.
- Standard security answers. SOC 2 status, pen-test results, the questionnaire answered before procurement sends it.
A buyer who sees all of that before the pilot decides you are a safe vendor. A buyer who has to ask for it decides you are a small one.
Hardware, integrators and consultancies open the doors.
Most AI companies need someone else’s hardware, someone else’s platform, or someone else’s credibility to get through an enterprise door. The partner motion is not optional, and it rarely gets built on purpose.
In the computer-vision engagement that started this practice, partnerships weren’t in my charter. I built them anyway: an exclusive hardware partnership, the operating motion behind a Big Four consulting alliance, and a co-sell network of integrators and solution providers. I also qualified partners out, which matters as much as signing them.
For a founder that means a short list of partners who can actually carry you into named accounts, a joint plan for each, and a weekly cadence that turns logos into pipeline.
Fractional VP of Sales for AI founders.
I help you run the sales motion and bring you in only where the founder is the difference. For an AI or computer-vision company that usually means:
- An ICP and buyer map that names who owns the production budget, not just who sponsors the pilot.
- Metric-led positioning and a business case in the buyer’s units.
- A pilot design with the success measure, the production owner and the decision date agreed in writing.
- The privacy, security and model-governance packets built once, before the first review.
- A partner motion — hardware, platforms, integrators — with joint account plans and a pipeline you can forecast.
- Signal-based prospecting, so the next ten pilots are with accounts that have a reason to buy this year.
Selling AI to the enterprise, answered.
Our pilots go well but never convert. What is wrong?
Almost always the pilot was designed to prove the model, not the business case. Before the next one starts, agree the metric, the production budget owner and the decision date in writing. A pilot with no decision date is a science project. Late-stage stalls are power-base problems is the longer answer.
Do you only work with computer vision?
No. Computer vision in retail is where I have the most recent hands-on work. The same motion applies to any AI product sold to an enterprise committee: analytics, forecasting, agents and copilots, voice, document AI. The gate is a committee sale, not the model type.
How do we sell a category the buyer has no budget for?
Attach to a budget that already exists. Find the use case that saves or makes money in a line someone already owns, win there, and expand. Category creation is a marketing goal. Closing the first contract is a sales job, and it happens inside an existing budget.
When should an AI startup hire its first salesperson?
After someone other than the founder can describe the buyer, the metric and the reason to buy this year in the same words. Before that, a hire inherits an experiment. The first sales hire: a leader or a rep covers the choice.
Are you a full-time hire in disguise?
No. I am a fractional VP of Sales: senior, part-time, inside your deals for as long as the work needs it, then handed over, handed back or kept running as advisory. Services explains the shapes.
Notes from inside these deals.
Why enterprise deals stall after the demo
If a deal dies after the demo, it is rarely because the demo was bad. It is because you never mapped who can actually say yes.
Your first sales hire: a leader or a rep?
Should your first sales hire be a leader or a rep? Usually neither, yet. The answer depends on whether a motion exists, and most founders hire before it does.
Getting out of founder-led sales without losing the deals
Transitioning away from founder-led sales fails when the founder hands deals to a VP before a motion exists. Here is the sequence that keeps the deals, a 90-day plan, and how to hand off a live deal.
Other industries I work in: Retail technology · B2B companies in any industry · B2B SaaS · eCommerce & order management · Point of sale & store technology · Supply chain & logistics technology
Thirty minutes. Bring the pilot that won’t convert. Leave knowing what I’d change before the next one.
Not ready for a call? Check the fit in two minutes