There is a version of Arseny who does not wait for perfect clarity.
There is a version of Arseny who does not wait for perfect clarity.
He is not calmer than the current Arseny. He is not more naturally confident. He still wakes up with too many browser tabs open, too many half-built projects, too many possible directions: iBrain, Aftercall, EchoThread, Relay CRM, agency intelligence, fashion retail, artist platforms. His mind still tries to turn every idea into a courtroom. Prosecution. Defense. Cross-examination. Appeal.
But in this version of the story, he notices one thing early enough: the question "Is this the right idea?" is usually unanswerable from the desk.
So he stops trying to solve it in his head.
That is the first difference.
He has been studying Michia Rohrssen and Prodigy, and the part that stays with him is not the $110 million sale. That part is too cinematic, too clean, too dangerous. Anyone can mythologize an exit. What stays with him is the mess before the exit. Prodigy did not begin as the perfect idea. It began as an intelligent guess inside a huge, ugly market.
Car dealerships were ancient. Consumers were changing. Software was eating everything. Somewhere inside that collision there had to be a company.
Michia and his co-founder did not know exactly where.
Their first idea was not "let people buy cars online." Their first idea was closer to advertising optimization for dealerships. Dealers spent huge money on ads. The founders assumed dealers would want better software to make that spend smarter. It was logical. It sounded venture-backable. It fit the market.
And then they tried to sell it.
Nobody cared enough.
That is where most founders quietly die. Not because the market says no, but because the founder takes the no personally. He retreats into redesign. He opens Figma. He rewrites the positioning. He asks another AI for a sharper pitch. He tells himself the idea needs one more insight before he can show it again.
But the Prodigy lesson is harsher and more useful: the wrong idea was not a failure. It was the entrance fee.
The founders had earned access to the real conversation. Once they were already speaking to dealers, they could ask a better question: "If you could wave a magic wand and change anything in your business, what would it be?"
That question changed the company.
The dealers did not say, "Please optimize my ads." They said, in effect, "I wish customers could just buy the car online."
There it was. Not an abstract SaaS pain. A transaction pain. A money pain. A labor pain. A customer-experience pain. A workflow pain.
And in this alternative life, Arseny sees the parallel.
He stops asking, "Should I build iBrain? Should I build Aftercall? Should I build EchoThread?" Those are founder-side names. Internal names. Museum names. They describe the artifact, not the wound.
He asks instead: "Where is the painful transaction?"
For iHousedesign, the painful transaction is not "AI knowledge management." Nobody wakes up wanting AI knowledge management. The pain is the client call that creates ten promises, five hidden risks, three follow-ups, one estimate, and zero reliable memory. The pain is the expensive creative agency that loses money because commitments leak between Telegram, email, Asana, Teams, voice notes, and someone's tired brain. The pain is not information. The pain is operational amnesia.
That is Aftercall.
Not a meeting recorder. A post-call control layer.
In this version, Arseny does not build the full product first. He does what Michia did. He sells the idea before it exists.
He makes a list of twenty people who already feel the pain: small agency owners, creative directors, boutique operators, production people, maybe fashion retail founders, maybe gallery owners, maybe architecture studios. Not enterprise buyers. Not people with procurement departments. He looks for the slightly chaotic operator who still makes decisions personally and feels the cost of every missed follow-up.
He does not email them, "I built an AI tool."
He writes something more human.
"I'm building a system for agency owners that turns every client call into a clean operating record: decisions, promises, follow-ups, risks, estimate changes, and draft client replies. I'm not selling it yet. I'm trying to understand if this is painful enough to deserve a product. Could I ask you a few questions?"
Some ignore him. Some say yes because the framing is not threatening. He is not pitching them a dashboard. He is asking them to describe the part of their business they already hate.
The first calls are awkward. Arseny overexplains. He talks too much about AI. He mentions local recording, Teams, Telegram, client memory, internal search, maybe even the grand iBrain architecture. He can feel the buyer's attention slipping. Then he remembers the lesson: do not sell the system. Find the wound.
So by the fifth call, he asks better questions.
"What happens after a client call today?"
"What gets lost?"
"Where do you lose money?"
"Which promises are dangerous if forgotten?"
"Who is responsible for turning the conversation into action?"
"When was the last time a client said, 'But I thought we agreed…'?"
Now people start talking.
One agency owner says the worst part is not the call itself. It is the two hours after the call when he is already late to the next thing and has to reconstruct what happened from memory. Another says estimates quietly mutate after calls, and nobody knows what was included. Another says the team hears one thing, the client remembers another, and the owner becomes the human legal archive. Another says he does not need more notes. He needs the system to tell him what changed since the last conversation.
Arseny hears it.
The product is not "AI meeting notes."
The product is "client-call memory that protects margin."
That sentence is not perfect, but it is alive.
Now he needs a first customer.
This is the second place where the alternative Arseny behaves differently. He still doubts. He still wonders if the category is too small, if Apple will build it, if Fireflies or Fathom or Zoom AI or some YC startup will crush it. But he has learned that doubt does not get a vote until after market contact.
So he makes the offer concrete.
"I'm putting together a small customer advisory group of five agency owners. You get founder-level setup, direct influence on the product, and a permanent early price. In exchange, I need honest feedback, permission to use anonymized results, and if it works, a testimonial or reference. If it does not save you time or prevent mistakes, I refund you."
This is not a trick. It is a fair trade.
The first person says no. The second says maybe. The third asks how much. Arseny almost says free, because free feels safer. Then he remembers another lesson: free produces compliments, not truth.
He charges.
Not a huge amount. Enough to make the buyer decide. Maybe $300/month. Maybe $500/month with a setup fee. Maybe a discounted annual pilot. The exact number matters less than the psychology: the customer must cross from "interesting" to "I want this to exist."
One customer says yes.
That is the beginning.
Not the TechCrunch beginning. Not the launch-video beginning. The real beginning. A Stripe notification. A messy onboarding call. A folder full of recordings. A half-manual workflow behind the scenes. Arseny does not yet have the perfect app. He has a promise and a narrow use case.
For the first version, he does not build everything. He creates a black-car product: one workflow, one pain, one output.
After every client call, the system produces five things
A clean summary of decisions.
A list of promises made by the agency.
A list of promises made by the client.
A risk register: unclear scope, budget tension, missing assets, deadline pressure, possible disagreement.
A draft follow-up email the owner can send within ten minutes.
That is it.
No giant dashboard. No knowledge graph. No mood analysis. No "AI brain." No philosophical interface. Just the thing that saves the owner after the call.
At first, much of it is manual. Arseny listens to pieces himself. He edits the outputs. He notices what the AI misses. He sees patterns. Clients rarely say, "This is a scope-change risk." They say, "Maybe we can also add…" or "This should be quick…" or "We already discussed this…" The product has to understand business subtext, not just transcription.
This becomes the moat.
Not the model. Not the summarization. The moat is the accumulated understanding of how creative-service conversations turn into margin leaks.
After two weeks, the first customer says the output is useful, but too long.
After four weeks, the customer says the risk section is the best part.
After six weeks, the customer forwards one of the follow-up emails almost unchanged.
That is signal.
Arseny resists the urge to expand. He wants to add CRM sync, Telegram ingestion, Asana automation, estimate generation, voice memory, project dashboards, relationship intelligence, maybe even predictive client-risk scoring. All of that might come later. But he remembers Prodigy: the first paid version can still be wrong. Twenty-two paying dealerships can still cancel. Early revenue is not truth. Retention is truth.
So he narrows further.
The product becomes: "After every client call, know exactly what changed, who owes what, and what could cost you money."
Now the second customer is easier.
He does not pitch a concept anymore. He tells a story.
"One agency owner was spending two hours after calls reconstructing decisions. Now he sends a clean recap in ten minutes and catches scope drift before it becomes unpaid work. I'm opening two more advisory spots."
That is different from "I built an AI tool."
By the fifth customer, Arseny has enough pain data to see the category more clearly. The buyer is not buying productivity. They are buying control. The emotional purchase is: "I want to stop being the only person who remembers what happened."
The technical product can now evolve.
He hires or assigns a small team, not a big one. One developer. One product/design person. One operator who understands agency delivery. Maybe he uses the existing iHousedesign team, but carefully: no vague delegation, no giant internal initiative, no endless side project. He gives the team a narrow mission.
Build the thinnest software layer around the workflow that is already working manually.
The first real product is ugly but useful. Upload or record a call. Attach it to a client/project. Generate the five-part operating record. Let the owner edit it. Send the follow-up. Track whether action items were completed. Compare this call to the previous call and highlight what changed.
That "what changed" feature becomes the first moment of magic.
Because normal meeting tools summarize each call in isolation. Aftercall remembers the relationship.
The product says:
"Client previously approved Option B. In this call, they appear to reopen Option A."
"Budget was described as fixed last week. Today they asked about additional deliverables."
"Agency promised delivery by Friday. No owner assigned."
"Client has mentioned urgency in three consecutive calls."
This is no longer a transcript tool. It is a memory and risk system for client work.
The company starts to have a point of view.
The point of view is that service businesses do not lose margin only because they estimate badly. They lose margin because reality changes in conversation, and the business fails to capture the change.
That is the wedge.
Now Arseny begins to understand timing.
Five years earlier, this would have been hard. Transcription was worse. LLMs were not good enough. Building a native capture layer was painful. Agencies were less comfortable with AI in operations. But now the timing is strange and perfect. Calls are everywhere. Remote work normalized recorded conversations. AI made summarization cheap. But most tools still stop at notes. They do not close the operational loop.
That is the "why now."
He writes it down:
"Every business conversation is becoming machine-readable, but most companies still operate as if calls disappear when they end."
That is a real thesis.
There is still doubt. Of course there is doubt. Doubt does not vanish. It changes shape.
At first, he doubted the idea. Then he doubted the buyer. Then he doubted the market size. Then he doubted whether this was venture-scale or just a useful service. Then he doubted whether he was too scattered to lead it. But each time, instead of solving the doubt by thinking, he designs a test.
Will people pay? Ask ten.
This is how hesitation becomes a machine.
The business is no longer built by confidence. It is built by tests.
By month three, he has ten advisory customers. Not all are happy. Two barely use it. One cancels. One says the summaries are generic. One says the follow-up email alone is worth the money. One asks for estimate integration. One wants it inside Asana. One wants WhatsApp voice notes. One wants it for internal team meetings.
The old Arseny would see fragmentation and panic.
The alternative Arseny sees segmentation.
He does not build for all of them. He looks for the strongest repeated pattern. The best customers are owner-led creative/service businesses doing high-touch client work where scope, approvals, and follow-up matter. That becomes the ICP.
He says no to everything else for now.
No internal meeting summaries.
No general personal memory.
No dating CRM.
No founder-wisdom library.
No universal iBrain.
The product has a job.
Protect margin after client calls.
The team now has clarity. The product roadmap is not a wish list. It is a sequence of risk removals.
Can we reliably capture the call?
Can we identify decisions and promises?
Can we detect scope drift?
Can we produce a client-ready follow-up?
Can we push tasks into the system the team already uses?
Can we show the owner where money is leaking?
At this point, the company becomes more Prodigy-like. Not in market, but in pattern.
Prodigy did not win because it had a pretty interface for buying cars. It won because buying a car is a brutal workflow: inventory, financing, trade-ins, compliance, dealership systems, paperwork, state rules, sales teams, lenders. The mess was the moat.
Aftercall's version of the mess is different: client memory, approvals, scope, estimates, project management, team accountability, email follow-up, payment justification, relationship history.
If Arseny can own that mess deeply enough, competitors cannot copy it by adding "AI summaries."
The story becomes strategic.
A generic recorder is worth little. A system of record for client commitments is worth more. A margin-protection layer for service businesses is worth even more. And if one day it connects to CRM, invoicing, estimates, and payments, then it sits near the transaction.
That is where the multiple lives.
The company does not need to become huge immediately. It needs to become necessary in one narrow workflow.
The first case study writes itself:
"Before Aftercall, a creative agency owner spent 5–7 hours a week reconstructing client calls and still missed follow-ups. After Aftercall, every call produced a decision record, risk list, task list, and client-ready recap within minutes. In the first month, the agency identified three scope changes before unpaid work began."
Now the product has language buyers understand.
Unpaid work prevention.
That phrase gets attention.
Month six is not glamorous. The product still breaks. Integrations are annoying. Some clients record inconsistently. The AI sometimes invents confidence where there should be uncertainty. The team has to build safeguards. The onboarding needs work. But the founder has something he did not have before: a narrow truth.
He knows who it is for.
He knows what pain it solves.
He knows what output matters.
He knows what customers ignore.
That is product-market fit beginning to form.
Not the romantic version. The ugly version.
The version where the founder's job is not to dream forever, but to keep removing reasons the customer stops using the product.
And maybe years later, if the company works, people will compress the story into something stupidly clean.
"Arseny built Aftercall, an AI operating system for client-service businesses, and sold it for X."
They will skip the uncertain part. They will skip the first awkward calls. They will skip the manually edited summaries. They will skip the canceled pilot. They will skip the week where he nearly expanded into five markets and had to force himself back to one.
But that was the company.
The company was not born from certainty. It was born from contact.
That is the alternative life.
Not Arseny as a different person. Arseny with a different rule:
Do not wait until the idea feels safe.
Find a large ugly market.
Find the painful transaction.
Ask the magic-wand question.
Sell before building.
Keep the first product brutally narrow.
Watch retention, not compliments.
Turn messy workflow into moat.
Stay close to money.
Let everything else drop.
In that life, doubt still appears every morning.
But it is no longer the CEO.