Ventura · Summer 2025, later in Y Combinator (Winter 2026)
When guessing costs a customer
Reading napkin orders into the ordering system, with a human in the loop
I joined Ventura as its second person. By summer’s end, 15 industrial distributors were paying for our agent.
- Role
- Co-founder, product and go-to-market
- Team
- Two founders, then a chief technology officer I brought on and a small engineering team
- Timeline
- June to August 2025
- What changed
- From retyping orders all day to checking what the agent flags
In short
- A rep types every order, from faxed purchase orders to the odd napkin, line by line into the distributor’s ordering system (its ERP).
- Our agent matches each line to the catalog and flags the ones it isn’t sure of.
- The sales calls decided what we built: the one job distributors described.
My part
- Decidedthat nothing would reach the ERP until a person had resolved every line the agent flagged.
- Ranevery sales call, the pricing and the business plan as the de facto product manager.
- Builtthe agent, from reading the order to syncing it to the ERP.
- Builtthe public demo of the review workflow, so anyone could see it without a sales call.Clean-Data
Try the review step
Two lines the machine isn’t sure of wait for you.
In
- Emailneed 20 of the 3/4 ball valves brass + 10x 1/2 npt tees
- Faxed purchase orderPO 58812 · 50 HEX BOLT 3/8-16 x 2 ZN
- Counter slip6 x 2" pvc cplg, 2 x pipe dope
Matched to the catalog
- Ball valve, 3/4", brass, full port× 2097% sure
- Tee, 1/2" NPT, black iron× 1088% sure
- Hex bolt, 3/8"-16 × 2", zinc× 5099% sure
For a person to check
Two lines need a person: pick one to check it.
Made-up orders and parts. The reader is a perforating machine from the Smithsonian’s National Postal Museum.
In motion
The problem
An inside-sales rep spends most of the day re-keying orders, and every line is a chance to get a part number or quantity wrong. The orders are chaos: an email that says “blk iron niples 3/4”, or a purchase order scanned at an angle. The rep knows the catalog well enough to decode all of it, which is why the job is so hard to hand to software.
From an idea to 15 paying distributors
We had one summer, and I owned the outreach, the calls and the prototypes. Plenty of ideas sounded good from the engineering side, my co-founder’s among them, and distributors could tell in one conversation that they wouldn’t help. I set those ideas aside. We kept the job they described: getting the order into the ERP the way their best rep would.
At the end, the distributors thanked us, and so did the reps who had spent their days entering orders and got that time back.
Decision log
to test every sentence of our cold outreach, the subject line and the channel.
Rejectedone good pitch for everyone.
Whywe didn’t yet know the buyer.
Result10 percent of the people we reached answered, against the usual 1 to 2 percent.
to lock “Send to ERP” until a person has resolved every flagged line.
Rejectedsending orders on their own.
Whya wrong part that ships is a returned pallet and, often enough, a lost account. A flagged line costs a rep about ten seconds.
Resultthe review step became the demo.
to show the receipt: hover any line to see the words it came from.
Rejecteda clean table.
Whyreps needed to check the agent’s work.
Resulttrust in the first month came from reps finding the receipt right.
to sell the review workflow, not the model.
Rejectedleading with accuracy numbers.
Whyevery buyer we met had heard accuracy claims. None had seen a review step designed around their own people.
Resultabout $600,000 closed in three months.
to leave pricing rules and live inventory checks out.
Rejecteda complete order-to-cash flow.
Whymatching lines correctly, with a person’s sign-off on the uncertain ones, was the whole promise.
Resultthe first version did one thing well enough to be paid for.
What I would do differently
Go in person: spend a day shadowing a distributor’s order desk instead of relying on how people described their process. And plan the summer better: the testing paid off but pushed the building into the last few weeks, all-nighter after all-nighter. I should have asked for help sooner.
What I would do next
Measure the two numbers a distributor cares about, before and after: minutes from an order arriving to its entry in the ERP, and wrong part numbers per hundred lines. We had testimonials, not those numbers.
A postcard
From San Francisco.
Postmarked 1951, photograph by Hope Moore (public domain). Click or tap the card to read the note on the back.
Try it: run the demo, resolve what it flags, and send the order to a mock ERP.
Go deeper
- Clean-Data source on GitHubMy public re-implementation of the review workflow.
