What Claude said when it finished building my software — and why that sentence is the best argument I’ve ever heard for the irreplaceable salesperson.
By Brad Hawley, Head of Client Engagement | 6 min read
Last night, I finished building a piece of software with Claude, an AI assistant. After hours of iteration — debugging, refining, polishing — Claude delivered the final version with a simple sign-off:
“All done. Go close some deals.”
That sentence is the entire argument of this paper.
Claude knew exactly what it had built. It knew exactly what it had not. It built the asset. It could not carry the asset across the finish line. That work — the closing — was mine.
This is not a limitation of the current generation of AI. It is a structural feature of what selling actually is. And once you understand why, you stop worrying about being replaced and start figuring out how to use AI to become two or three times more effective than the seller next to you who isn’t.
Walk into any banking conference, leadership offsite, or LinkedIn feed and you’ll hear some version of the same anxiety: AI is going to replace sellers. Producers, lenders, treasury officers, relationship managers — all of it, automated away.
The narrative is loud. It is also wrong. Or more precisely: it is wrong about the people who actually generate revenue, and roughly right about the people who do not.
AI is replacing tasks. It is not replacing trust. And in any business where the buyer is making a high-stakes, high-consideration decision — choosing a primary banking relationship, moving treasury operations, taking out a $20M commercial loan — trust is the entire transaction. Everything else is paperwork.
Let us be honest about where AI is genuinely powerful, because the case for the human seller only matters if we are not in denial about the case for the machine.
Pulling a prospect’s financials, recent news, leadership changes, and competitive landscape in seconds.
Email sequences, follow-ups, proposals, one-pagers, and meeting recaps.
Taking a sprawling discovery call and producing a clean summary with action items.
Spotting which deals in a pipeline resemble deals that closed, and which resemble deals that died.
Decks, scripts, microsites, calculators, and portals — the things a seller used to wait two weeks on a marketing team to produce.
If your job consists mostly of any of the above, you should be nervous. Not because AI is coming for you, but because someone using AI is.
But notice what is not on that list.
AI cannot walk into a community bank’s lobby with a box of pastries and have the receptionist remember its name three months later.
AI cannot read the pause on the other end of the phone when a CFO says “we’ll think about it” and know — know — that the real objection is something the CFO has not yet said.
AI cannot sit across from a market president whose biggest producer just walked to a competitor and convey, with the silence between sentences, that it understands what that loss actually means.
AI cannot be present in the room.
That last point is the whole game. In revenue-generating roles — the kind of roles we recruit for at Talnted — the buyer is not buying a product. The buyer is buying a person whom they believe will not let them down. And no amount of large language model fluency replaces the simple, irreducible fact that one human is choosing to trust another.
Back to the moment that started this paper. Claude built the thing. Then it told me to go sell. Implicit in that handoff is a complete and accurate model of who does what:
Claude was not being modest. It was being accurate. It built faster and better than I could have alone. And it correctly identified that the next step — the part where money changes hands — was not a step it could take.
This is the pattern. Not just for software. For everything that matters in selling.
Community and regional banking is, in many ways, the purest test case for this argument.
The product is a commodity. Money is money. Rates compete within basis points. Treasury platforms are functionally similar. The technology stack is converging. If selling were really about explaining the product, AI would have already won.
It has not. And it will not. Because nobody picks a community bank for the product. They pick it for the banker.
The CFO who moves $40M in operating accounts to your bank is making that decision because she trusts your treasury officer to answer the phone at 4:47 PM on a Friday when something has gone sideways. The business owner who refinances a $12M facility is doing it because your lender showed up to his daughter’s softball tournament. The board chair who consolidates personal wealth at your bank is doing it because your market president has been dropping by for two years asking nothing in return.
You cannot automate any of that. And if you try, you do not just fail — you actively erode the thing that made the relationship work in the first place.
The actual competitive threat to a seller today is not AI. It is another seller who is using AI well.
That seller is walking into discovery calls with research the unaided seller did not do. They are sending follow-ups within 20 minutes that used to take two days. They are producing custom proposals overnight that used to take a week. They are remembering every detail of every conversation because their AI is keeping notes that they actually review.
They are not being replaced. They are being amplified. And they are eating the lunch of every seller who is still doing it the old way out of pride or fear.
This is the lesson of every previous technology wave in sales. The CRM did not replace the salesperson. It replaced the salesperson who refused to use a CRM. Email did not replace the salesperson. It replaced the salesperson who insisted on faxes. AI is the same pattern, just faster and steeper.
If you run a community or regional bank, the implication is straightforward and a little uncomfortable:
You are not hiring sellers who will be replaced by AI. You are hiring sellers who will either use AI to multiply their output — or lose to the ones who do.
The profile of the seller you want has shifted. Raw activity matters less. Judgment matters more. The willingness to pick up the phone matters more, not less, because everyone else is hiding behind automated sequences. The ability to be present in the room matters more, because the room is where the deal closes.
We screen for this at Talnted. The producers, lenders, and treasury officers we place are not the ones who are scared of AI. They are the ones who already use it and who understand — at a gut level — that the tool does not do the job. The person does.
Claude told me to go close some deals because it knew exactly where its job ended and mine began. That self-awareness is, ironically, more sophisticated than the human commentators who insist sellers are about to be replaced.
The future of sales is not human or AI. It is human plus AI, working the way I worked last night — the machine building the asset, the human carrying it across the line.
If you sell revenue for a living, that should be the most reassuring sentence you read this year.
Now go close some deals.
Technology makes sales teams more efficient. The foundation is still people. Talnted helps you find the humans who will actually move the needle.
Get a Hiring Assessment✓ AI builds the asset; the human carries it across the finish line. Both roles are essential — and not interchangeable.
✓ Trust is the entire transaction in revenue-generating roles. Trust does not transfer to a chatbot.
✓ The competitive threat to sellers is not AI — it is another seller who uses AI well.
✓ In community banking, the product is a commodity. The banker is not. That is why human sellers win.
✓ Hire for judgment, presence, and the willingness to be present in the room. Those are the traits AI cannot replicate or replace.