Anis Bennaceur Is Teaching Sales Software to Finish the Job

The most consequential clue in Anis Bennaceur's second company was hiding in the least romantic corner of his first: the CRM. At Mixer, the professional network he co-founded for people in creative industries, every useful sales conversation produced a second job. Somebody had to remember what was said, translate it into tidy fields, write the follow-up and preserve enough context for the next person. The customer had spoken in paragraphs. The database wanted checkboxes. Between them stood a sales rep with a calendar already full.

Bennaceur understood the cost because Mixer was bootstrapped and enterprise-focused. Lean companies make waste embarrassingly visible. A missed note can become a missed deal; a slow handover can become a vanished customer. He had spent years in New York learning the unglamorous mechanics of selling software, after earlier chapters in finance and Tinder's growth effort in France. The recurring nuisance became difficult to ignore.

Then came coffee with Matthias Wickenburg. This was not an ordinary founder date. Wickenburg had run Swipecast, a competing platform for creative work. For roughly five years, the two men had occupied opposite sides of a small market. When they finally compared notes, they found the same bruise: Salesforce demanded constant care, and getting new sellers productive took too long. Competition had given them compatible evidence.

“After many back-and-forths, we decided to work together.”Anis Bennaceur, on founding Attention

The rival who understood the scar

Attention began in September 2021 with Bennaceur as CEO and Wickenburg as CTO. Their division of labor was clean. Bennaceur spoke with sales leaders and junior representatives, looking for pain rather than applause. Wickenburg built versions in parallel. Bennaceur later described hundreds of these conversations. It is hard to romanticize this method, which is one reason it is worth noticing. The early company was essentially a loop made of interviews, software and disappointment.

The first idea leaned toward real-time coaching. The language models could demonstrate the direction but were not dependable enough to carry it. The founders spent about eight months on an initial version, placed an MVP with five design partners by summer 2022, then concentrated on the simpler and sharper wedge: extract structured information from a call and put it where a sales team already worked. When improved models arrived late that year, the product's usefulness changed quickly. Attention emerged from stealth in January 2023 alongside a $3.1 million seed round.

The timing helped, but timing alone is a lazy explanation. Bennaceur had begun experimenting with GPT-3's API in 2020. He saw that a long transcript could be segmented, processed and converted into structured output before most people had a friendly chat window for the technology. The bet was not that a machine could make a summary. The bet was that messy human speech could become usable operational memory.

The removal test

Bennaceur's public advice on product-market fit is pleasingly severe. Find ten beta users. Let them use the product for two months. Then take it away. If seven are disappointed, the signal is meaningful. Compliments are inexpensive; absence has a price. Attention initially served teams with fewer than ten representatives, which kept engineering aimed at the central behavior instead of the administrative frills that larger buyers understandably request.

Only after that signal did the company lean into what Bennaceur calls growth stacking: run small experiments, retain the ones that work and automate them. Attention used its own conversation data to improve messaging. It personalized outreach from public podcast transcripts. It treated recent website activity as a reason to make an introduction warmer. The experiments accumulated rather than arriving as a single theatrical hack.

He also became his own awkwardly useful customer. After hiring two salespeople, Bennaceur returned to selling so he could experience Attention as a manager and rep. The practice clarified positioning and exposed which features mattered in the working day. His compact qualification framework asks three questions: Why do anything? Why now? Why Attention? Each removes a layer of founder vanity and puts the buyer's circumstance back in the room.

“Talk to customers, and build product - and for that, you don't need much cash to get started.”Anis Bennaceur

There is a smaller anecdote that catches his operating taste. In 2024, irritated by the process of inviting selected contacts to follow Attention's LinkedIn page, he spent about an hour building a browser extension. It saved his team roughly an hour each month. He published the tool for personal, non-commercial use. The arithmetic is modest. The reflex is the point: recurring annoyance is a request for software.

After the notetaker

By 2026, recording a call and writing a summary had become ordinary. Bennaceur's newer argument is that the value has moved one step down the chain. Software should draft the follow-up, update the system of record, identify the risky deal, recommend what happens next and execute with human approval. Attention calls this an action loop: act, measure the result, learn, then act with better context.

That distinction gives the product a different question to answer. A passive tool can report that a deal slipped. An action system can show what it attempted, whether the seller approved it and what happened afterward. In June 2026, Attention said it was running more than 20 million agent actions each month and that revenue had increased fourfold over the preceding year. The company announced a $30 million Series B led by RTP Global, following a $14 million Series A in 2024.

Bennaceur wrote that the amount raised was not the signal he cared about most. Customers who used the product also invested. It is a founder's favorite form of symmetry: the people paying the invoice volunteer to own a piece of the machine. The round is funding Proactive Insights, designed to rank a representative's next moves by revenue at stake and carry out the approved ones.

Intensity, stated plainly

Bennaceur is unusually explicit about the personal cost he associates with company-building. In a 2023 essay, he advised most people against starting a startup and described the necessary obsession in terms that leave little room for artisanal leisure. His co-founder, he wrote, was the last person he texted before sleep and the first after waking. A week without learning or a milestone counted as a failed one.

The view is austere, and it connects directly to how he talks about hiring. He favors people who learn quickly, advocates paying above market for strong talent and says the company hires slowly. He has also said choosing investors is personal, once reducing the judgment to “just vibes.” The phrase sounds casual, but his earlier writing gives it teeth. Mixer taught him to choose backers carefully, avoid false economies in hiring and resist growth before the product earns it.

There is play in the machinery, too. Bennaceur learned to code at about ten, and his instinct to make things survived the transition into management. The small LinkedIn extension sits beside essays, experiments and a podcast in which he interviews operators about sales, growth and product. These side doors into his thinking matter. They show a CEO who does not reserve curiosity for the quarterly plan. If a process is clumsy, he may build around it. If a useful operator knows something, he may put a microphone between them and ask.

That habit also explains why customer conversation became more than raw material for Attention. Bennaceur treats it as a management instrument. Calls inform product priorities, sales coaching and marketing language. The same evidence can travel through the company instead of being stranded in a transcript. His software thesis and operating thesis meet here: information becomes valuable when it returns quickly to the people who can act on it.

His path is not a tidy ascent. Attention's first conception did not hold. The underlying models were early. The useful product appeared through narrowing, waiting and repeated contact with customers. Even the co-founder relationship required revising an old judgment: a rival became the person on the other end of those first and last daily texts.

Now Bennaceur's aspiration is broader than automating CRM chores. He wants Attention to help revenue teams operate with fewer administrative gaps and more memory, first by building strength in the United States and later by expanding internationally. The ambition is large; the object of attention remains wonderfully mundane. A customer says something important. The meeting ends. Someone still has to do the work.

The wager behind Attention is that software can finally pick up that work without making the seller behave like a database clerk. Bennaceur arrived at it through a decade of enterprise friction, one former competitor and the discipline of asking users what hurts. In the noisy business of artificial intelligence, his clue was a quiet one: listen carefully, then finish the job.