The 30-second version:

On June 24, 2026, a four-year-old New York scale-up closed $30 million to sell an idea that is simple to state and hard to ship: sales software should not stop at telling you what happened on a call, it should do the next thing. Attention drafts and sends the follow-up, updates the CRM, launches the next sales step. Behind the round sits a deeper shift in sales tech: value is leaving the layer that observes for the layer that acts.

The fact: $30 million for AI that acts instead of observes

Attention announced a $30 million Series B on June 24, 2026, led by RTP Global, with returning investors Aglaé Ventures, Eniac and Alven, new investor Linea Ventures, and a group of angels drawn from its own customer base, including Kirill Bigai (CEO of Preply) and Sam Jacobs (CEO of Pavilion) (GlobeNewswire release). Founded in late 2021 in New York by Anis Bennaceur (CEO) and Matthias Wickenburg (CTO), the company will use the funds to expand its agentic offering and move upmarket into large revenue organizations.

The numbers it puts forward show the traction. Attention runs more than 20 million agent actions per month since launching the capability, its annual recurring revenue has grown 4x year over year, and average contract value is up 10x over two years (citybiz). The platform claims more than 500 customers, among them Abridge, Scale, Lovable, Preply and BambooHR.

The product promise fits in one line: where most tools listen to the call and write up the summary, Attention takes the next action. The software drafts and sends the follow-up email, updates the record in the CRM, triggers the next commercial move, then ties the outcome back to its own work (TechEdge AI). Anis Bennaceur, co-founder and CEO, sums up the thesis:

"Most software watches the call and writes up what happened. We take the next best action, and because we take it, we can see what actually worked, and we get smarter every time we do."

Why Attention targets the shift from system of record to system of action

Attention is attacking the line between the system of record and the system of action. For twenty years, sales software has been a system of record: the CRM logged the state of the pipeline, conversation-intelligence tools like Gong and Clari transcribed and summarized calls. All of them described reality; none of them changed it. Execution stayed manual: the rep wrote the follow-up, updated the record, and picked the next move.

Attention's bet is that this execution layer becomes the real product. Twenty million agent actions a month are not generated summaries; they are follow-ups sent, CRM fields filled, sales steps triggered with no human in the loop. The difference is economic as much as technical: a tool that summarizes is priced like an assistant, a tool that executes is priced like a team member. Average contract value up 10x in two years tells exactly that story, the move from a productivity gadget to revenue infrastructure.

This is not an isolated move. Salesforce pushes Agentforce as a layer of autonomous agents able to chain several steps of a sales workflow. HubSpot announced in May 2026 that it would open its CRM to external AI agents, with a commitment to API parity and a Model Context Protocol (MCP) server so every feature can be driven programmatically (SAHM Capital). The incumbents are turning their system of record into an execution platform; Attention was born on the other side of that line.

What the closed execution loop reveals

The most interesting part of Attention's thesis is not the action, it is the loop. "Because we take it, we can see what actually worked": a tool that only observes never knows whether its recommendation was good, because a human then decides whether to follow it. A tool that executes closes the circuit. It proposes an action, performs it, measures the result on the pipeline, and adjusts the next proposal. Attribution stops being an after-the-fact estimate; it is wired to the act itself.

That closed loop is what separates a copilot from an agent. A copilot suggests and waits; its quality depends on the discipline of whoever uses it. An agent acts and learns from the consequence; its quality improves with volume. At 20 million actions a month, Attention accumulates a learning signal that a plain transcription tool will never see, because that tool stops at the exact moment the decision gets made. It is the same logic that made recommendation engines powerful: whoever executes the choice owns the data on what converts.

One big caveat remains. An execution layer wired into the CRM is exactly the ground Salesforce and HubSpot want to hold with their own agents. Attention claims a usage lead, but its long-term defense will depend on executing better than the native agent shipped for free inside the CRM the customer already pays for. The question for the next twelve months is not traction, already proven by the 4x ARR, but depth of integration: how far Attention can act before it becomes just a feature of the platform it runs on.

Where this meets my day-to-day

This line between recording and executing is one I run into on revenue-ops work. On the bilingual Condition Report for Horus, built on Pipedrive, the value of a CRM is not the beauty of the pipeline but what fires on its own: a stage cleared that creates the follow-up, a field filled that notifies the right person. A pipeline that only records is a spreadsheet in disguise; what changes a team's life is the automation that acts on that state.

On the automation side, Attention's logic matches what I build with n8n: don't stop at capturing a signal, go all the way to the act that follows from it. On the IA Brew newsletter, the workflow doesn't just aggregate sources, it produces and ships the content end to end. An agent that drafts the follow-up and actually sends it, rather than suggesting it in a corner of the screen, is the same principle at the scale of a sales team: the value isn't in the suggestion, it is in the action carried through to the end.

The takeaway

Attention just raised $30 million on a precise line: sales software that acts is worth more than software that only observes. For a small business, the right question is no longer "which tool summarizes my calls best" but "what, in my sales stack, actually executes the next step for me".

Frequently asked questions

What does the Attention platform do?

Attention is an AI platform for revenue teams that does more than record calls: it drafts and sends follow-ups, updates the CRM and triggers the next sales play. Founded in late 2021 in New York, it runs more than 20 million agent actions per month for over 500 customers.

How much did Attention raise and who led the round?

Attention announced a $30 million Series B on June 24, 2026, led by RTP Global, with returning investors Aglaé Ventures, Eniac and Alven, new investor Linea Ventures, and angels drawn from its own customer base. The funds will expand its agentic offering and push upmarket into enterprise revenue teams.

How is Attention different from a tool like Gong or Clari?

Gong and Clari listen to, transcribe and summarize sales calls; they observe. Attention acts: it sends the follow-up, updates the CRM and then ties the sales outcome back to its own action. That is the shift from a system of record, which logs, to a system of action, which executes and measures what worked.