Stephie’s World · late editionNotes · by Stephie · October 2026
Every AI deployment needs a fundraiser
The engineer gets the system working. Someone else has to get people to care about it. When I ran Little Bun, that someone was me.
When I ran Little Bun, I was the fundraiser.
I raised a little venture money, but that wasn't most of it. Most of it happened in smaller rooms. A retail buyer who had a hundred other things to put on that shelf. A Michelin-star chef with no reason to spend an afternoon on a dumpling startup. Nobody owed me their attention. Every yes had to be earned with a story they could believe.
I keep thinking about those rooms as AI companies work out who they need sitting with a customer.
Two jobs, one hire
The forward deployed engineer has become the hero of AI deployments, and they should be. Someone has to sit with the customer's data, wire the model into their systems, and make the thing actually work. That's most of what makes a deployment real.
But most engineers I know aren't thinking commercially while they do it. They shouldn't have to. They're trying to get the plumbing right and the evals to pass, and that's a full job. Asking the same person to also read the politics of the company they're deployed into is asking for two jobs in one hire. Some people can do both. Not many.
Not a babysitter
Not the role most companies reach for. That seat used to be built for software that got sold once and then kept alive: watch the dashboard, send the reminder, run the quarterly review. AI doesn't need much babysitting. It needs someone to fight for it.
What the engineer needs is a partner who understands how organizations actually behave. Who really decides. Who's quietly against this. Which VP needs a win before the end of the year, and which one got burned by the last AI pilot. None of that is in the data. All of it decides whether the deployment lives.
Raising attention, raising money
That's the closest word I have for the job.
Inside the customer, you're raising attention. Every team has a roadmap, and your project is competing with everything else on it. Someone has to get people to care, over and over, to different people, in their language.
Outside, you're raising money. Someone has to pay for this, and the person holding the budget doesn't care how elegant the retrieval is. They care what changes for their business, how soon, and what waiting costs them.
A founder does both every day. So I think this seat should be hired like a founder, not like support.
Eight more yeses
Once it's signed, usage looks fine, the meetings are polite, and it will probably renew. But the company still has to change how it works. That's the longer raise.
And every step of it has a different person who has to say yes:
- A curious engineer tries it.
- A teammate starts using it too.
- A team lead changes the workflow.
- Whoever owns the metric sees it move.
- A manager carries that into a room.
- Another team with the same pain hears about it.
- Someone who owns process makes it standard.
- An executive who already saw the proof expands it.
Step three is where I'd expect it to get stuck. Say one engineer has had a coding tool open for a few weeks. In a demo it looks like a different job. In standup nobody notices. The team lead isn't against it. They just have a sprint to close, and code review still has to mean something if half the changes come in looking different.
More enthusiasm won't move that person. A smaller ask might. One repository, one kind of change, a review rule the lead can defend, until there's a good week they can point to. That week is the proof. A manager can carry proof into a room, and the room is how the next team hears about it.
The team lead who agrees to that rule is the champion for that step. They're already inside, people trust them, and they have their own reasons. Give them evidence and an opening and they'll do what no outside pitch can. That's how companies change. From the inside, one person at a time.
A story worth their time
I spent two and a half years at Amazon working with some of its biggest sellers. The days that mattered were never the ones where I followed a playbook. They were the ones where I understood a seller's business well enough to see a move that helped both sides, and then sold it to both sides. Their team, and mine.
Selling it to mine was the part nobody warns you about. Amazon is teams stacked on teams, all of them swamped, none of them with a reason to answer you. It overwhelms a lot of new people. It didn't overwhelm me, because I'd fundraised. Getting a reply from a busy Amazonian is easier than getting one from an investor who's ghosting you, and you get it the same way: with a story worth their time.
That's the job I'm describing. It doesn't have a good name yet. But it's the part of the work I've always liked best.