Case study: Formula. prospects with its own AI agent

Formula. uses its own agent to find its clients.

The result: a 40% reply rate on first messages, 52 ongoing conversations and 11 booked calls. Data pulled on July 23, 2026, from the product dashboard.

Here is the context, the setup step by step, and every number.

Why does Formula. prospect with its own agent?

Laura Terriou, founder of Formula., is in the middle of a launch. Product, content, clients: zero time left for manual outreach.

That is exactly the problem our clients live with. According to Salesforce's State of Sales research, salespeople spend less than 30% of their time actually selling. A founder mid-launch gets even less.

The target: French-speaking consultants and B2B SMBs, with an offer already validated.

Instead of telling you what the agent can do, we let it prospect for us. And we publish the numbers, dated.

How the agent was set up, step by step

1. Agent calibration. It ingests the website, the LinkedIn profile, past conversations, posts and competitors. Duration: a few dozen seconds.
   Result: an agent that writes like Formula., without a single line drafted by hand.
2. 8 campaigns and their personas. One persona per segment: independent consultants, SMB executives, consulting firms.
   Result: 3,734 prospects sourced, all aligned with the target.
3. The agent runs. Up to 28 invitations per day, sent between 8am and 6pm, with alternating human-like behavior.
   Result: 587 invitations sent, 179 accepted, a 30% acceptance rate.
4. Swipe validation. The agent writes every message and drafts every reply. Laura validates with a swipe, 3 seconds, from her phone.
   Result: 283 first messages, 114 replies, 52 ongoing conversations, 11 booked calls.

What do the numbers look like as of July 23, 2026?

On the Formula. account, 283 first messages generated 114 replies, a 40% reply rate, with 52 ongoing conversations and 11 booked calls. Data pulled on July 23, 2026, from the product dashboard. Classic manual outreach typically sits at 7 to 11%.

• 8 active campaigns
• 3,734 prospects sourced
• 587 invitations sent, 179 accepted: a 30% acceptance rate
• 283 first messages sent, 114 replies: a 40% reply rate
• 52 ongoing conversations
• 11 booked calls

Nothing extrapolated. It is the dashboard, as is, on the date shown.

Best campaign: French-speaking SMB executives

On this campaign, 59 invitations were accepted and 54 prospects replied. That is a 92% reply rate.

Why that score? A very precise persona, an offer that speaks directly to their problem, and messages calibrated on real conversations.

That is the lesson of this case: targeting precision beats volume. Formula. clients observe 30 to 60% reply rates depending on sector and offer, and this campaign shows the top of the range.

What timeline should you expect for your own results?

1. Week 1: calibration and first sends. The agent learns your offer, the first invitations go out the same day.
2. Weeks 2 to 4: volume ramps up. The agent climbs toward its maximum pace within LinkedIn limits: 28 invitations per day, 840 per month. The first serious conversations start.
3. Month 2: cruising speed. The funnel stays full. Clients observe 5 to 10 meetings per week depending on sector and offer, and some go past 10.

The full mechanics are on the Method page.

Which KPIs should you track every week?

• Acceptance rate on invitations: the health of your targeting. Formula. reference: 30%.
• Reply rate on first messages: the quality of the calibration. Formula. reference: 40%. What counts as a good rate in 2026.
• Active conversations: your real pipeline, not a file of cold contacts.
• Meetings per week: the only number that pays. How many to expect per month.

"Every morning I open the app and see my own agent's numbers on my phone. 114 replies out of 283 messages, so I know exactly what is working and why. My only gesture is a swipe between two meetings."

Laura Terriou, founder of Formula.

The misconception to correct: "an AI agent spams"

No. The Formula. agent sends 28 invitations per day maximum, between 8am and 6pm, with alternating human-like behavior: message, invitation, profile visit.

That stays below LinkedIn's official invitation limits. No mass blasting, no bursts.

Spam is 200 identical messages in one hour. This is 283 first messages over several weeks, each one calibrated on the prospect.

The honest alternatives: Waalaxy, Lemlist, agencies

Waalaxy, Lemlist: solid sequence tools. You write the messages, you schedule the sends, and you take over as soon as a prospect replies.

Lead gen agencies: they do the work, but at a higher cost, often several thousand dollars per month.

An agent runs the whole journey: it finds, contacts, qualifies and drafts every reply until the meeting. The full 2026 comparison is here.

Key takeaways

• Formula. prospects with its own agent: full transparency, numbers dated July 23, 2026.
• Real funnel: 283 first messages, 114 replies (40%), 52 conversations, 11 booked calls.
• Best campaign: French-speaking SMB executives, 54 replies out of 59 accepted invitations, 92%.
• Setup: calibration in a few dozen seconds, 8 campaigns, 28 invitations per day max.
• Your only gesture: validate with a swipe. The agent prospects, you collect the meetings in your calendar.
• Prerequisite: an offer already validated through manual sales, from about 5k euros of monthly revenue.

Frequently asked questions about this case study

Are these numbers verifiable?

Yes. Every number was pulled on July 23, 2026, from the Formula. product dashboard, the same screen every client sees.

Nothing is extrapolated or rounded up.

How long until I see comparable results?

Week 1: calibration and first sends. Weeks 2 to 4: volume ramps up within LinkedIn limits. Month 2: cruising speed.

Clients observe 5 to 10 meetings per week depending on sector and offer, and some go past 10.

How much time per day does the agent require?

A few minutes. The agent finds the prospects, writes the messages and drafts every reply.

Your only gesture: validate with a swipe, 3 seconds per message, from your phone.

What kind of business is Formula. built for?

B2B founders, consultants and SMBs with an offer already validated through manual sales, from about 5k euros of monthly revenue.

The agent amplifies an offer that already sells, it does not invent demand. Check that your offer is ready before you start.