We build Agents. You build the Future.
● Business offer — Lead qualification & appointmentsThe AI lead qualification agent: answer fast, book the meeting
A lead who contacts you expects an answer straight away. After a few hours their interest fades and they have often already approached a competitor. The problem is not a shortage of leads, but the speed and consistency of the handling. Your AI agent answers within the second, within the frame of a written sending mandate, qualifies, and books a meeting while interest is at its peak. Hosted in France — on local inference or an isolated resource — your prospects' contact details stay under control. The salesperson keeps the decision.
Updated on
Fictional simulation — not a client result
And do you already have a CRM in place (HubSpot, Pipedrive, Sellsy…)?
Request qualified: 12 salespeople, subject “follow-ups”, flagged urgent.
⛓ Sourced · incoming form + the account's qualification criteria
I confirm the meeting, send the invitation and a reminder the day before, and pass your qualified record (headcount, subject, CRM, urgency) to the salesperson.
✎ Action · meeting proposed and record ready — the salesperson approves
For an organisation established outside the European Union, the law of its own country applies; at its request, we put equivalent safeguards in place (human review before any action, a log of the criteria, information for the people concerned).
A Blue Lemon Agent lead qualification agent answers instantly to incoming enquiries (form, website, email, chat), asks the right questions to assess the need and the prospect's maturity, scores and prioritises, then proposes real slots and books the meeting in the sales diary. It follows up on lukewarm leads and passes the salesperson a qualified record with the full context. Hosted in France — on local inference or an isolated resource — prospects' contact details stays covered by an architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity, in compliance with the GDPR. Live in one to two weeks. Your teams write to it from Microsoft Teams, Slack or their email, and your customers reach it on WhatsApp Business, your website chat or email — with no account to create and nothing to install. These connections are included in every plan, at no extra cost, within the number of connections set out in your agent's quote.
Reference points describing our offer, not results measured at a client. The scale of the gain is confirmed by a pilot on your own scope.
Indicative estimate, to be confirmed by a pilot on your real files (general terms of sale, article 4).
Three options, one agent
A lead qualification and appointment-booking agent, installed and operated for you. Choose according to how you work. Prices exclude VAT — annual subscription, the time it takes for the gains to settle in.
Setup + controlled subscription
- Installation, configuration and training for your teams
- Operation, human oversight, updates and support
- Hosting in France, a dedicated and isolated resource
All inclusive, no setup fee
- Setup included (installation, configuration, training)
- Operation, human oversight, updates and support
- Hosting in France, managed end to end
On site, you own it
- Hardware installed on your premises (you own it)
- French / European AI models run locally
- Secure remote maintenance (Pro support included)
The uses of AI on your incoming leads
Each use corresponds to an agent we deploy. All of them work in support: subject to your approval, or under a written sending mandate for what the agent sends by itself.
Instant answers to leads
An answer within the second to incoming enquiries (form, website, email, chat), around the clock, where human handling is intermittent. Answering within the second requires a written sending mandate (scope, ceiling, duration); without a mandate, the reply is prepared and waits for a salesperson.
Adaptive qualification
Questions on the need, the timescale and the fit, adjusted as the exchange goes on to gauge the prospect's maturity.
Scoring & prioritisation
Each lead gets a score against your criteria; hot leads go to the top for the salespeople.
Appointment booking
Proposing real slots and writing the meeting straight into the sales diary.
Confirmations & reminders
Immediate confirmation and reminders before the meeting to cut missed appointments. By text message, to reach people who do not check their email.
Following up lukewarm leads
Automatic, personalised follow-up sequences to reactivate the leads that did not convert straight away. They go out under the same mandate, which sets the maximum number of follow-ups per person.
CRM updating
Connection to your CRM (HubSpot, Pipedrive, Sellsy…): qualified record, full context and history written automatically.
Support & everyday questions
Answering prospects' and customers' recurring questions: prices, lead times, how it works, status of a request.
Available View the agent page →Need to go further?
These agents handle a different business process, with their own owner and their own price. They are added to this one.
Incoming voice switchboard
For calls, a voice agent greets, qualifies and schedules the meeting without letting the phone ring out.
Voice agent (AI switchboard) from 456 € excl. VAT / month Voice agent →Outbound prospecting
Alongside the inbound side, an agent that goes out and engages new prospects to feed the pipeline.
Outbound prospecting agent (automated email, assisted LinkedIn) from 332 € excl. VAT / month Sales prospecting →In 30 minutes we identify the most relevant agent — without oversizing the project.
See the agent at work
4 everyday situations, taken from those that come up most often. Pick one: the exchange unfolds as it would in your organisation.
A scripted demonstration. These exchanges show how the agent behaves — its sources, its refusals, what it leaves to your teams. Nothing is sent from this page, no model is queried here, and the matters named are fictional. That is precisely what we promise your data. The behaviours shown here — monitoring, automation rules, routing and reminders — are configured with you during deployment, from your tools, your rules and your thresholds. The architecture points named in these exchanges — location, local execution, isolation, encryption, role-based access, logging — are not a guarantee attached to the demonstration: they are those of the architecture set out in your quotation, and verified before commissioning.
Fictional simulation — not a client result
· Of 61 appointments booked this quarter, 22 did not take place — 36%. Fourteen of them had been set more than fifteen days ahead.
· Nine appointments concerned a need your catalogue does not cover. The salesperson found out on arrival.
· One slot was offered while the salesperson was expected 200 km away that same afternoon. My calendar could not see their travel.
· Forty-seven enquiries arriving outside opening hours received no reply for two days. morning-watch_4-flags.pdf22 of 61 appointments not kept
⛓ Source · 61 appointments this quarter, visit reports, shared calendar, enquiry log
What I record: 22 appointments of 61 not kept. Fourteen had been set more than fifteen days after the conversation; eight within a week.
What that means: I booked distant appointments because a distant slot is always free. The prospect agrees, and then the reason that made them call fades — they found something, sorted it otherwise, or forgot.
What the 39 kept do not buy back: a missed appointment costs your salesperson a journey and another prospect a place. It is not cancelled, it is paid for twice — and at 22 out of 61, that is a day and a half of driving a month for nothing.
What I change, and it is already in place: I no longer offer beyond ten days without saying so. Past that, I state the wait out loud — "the first slot is [date], in three weeks; shall I call you back sooner if something frees up?" A prospect who hears "three weeks" decides knowingly.
What I now measure, and what was measured nowhere before: the gap between the conversation and the appointment, crossed with attendance. 22-of-61_14-beyond-15-days.pdfA distant slot is always free
⛓ Source · 61 appointments, booking lead times, attendance recorded
Routing follows whose time is lost: an appointment on a need outside the catalogue goes to the salesperson concerned, before they set off — afterwards they have made the journey; a slot incompatible with travel to the same person, immediately; an enquiry outside opening hours handled by me that evening, with the answer I am able to give; a recurring need outside the catalogue to sales management, as a count.
With a chase: immediate on a calendar conflict, 7 days on the rest. Then a monthly summary: by enquiry subject and by lead time, never by salesperson.
And I do not stop at the flag: whatever could be prepared is. Calendar conflicts come back with two replacement slots each, taken from actual availability; out-of-hours enquiries have their acknowledgement written and sent; needs outside the catalogue are counted and grouped — eleven enquiries over three months, all on the same service.
One rule holds up all the rest: I book an appointment only on a slot I can guarantee, and I say what you cannot handle rather than booking an appointment to avoid losing the contact. That costs in the short term — four fewer appointments this month — and it gives your salespeople four journeys back.
What stays theirs, and it is what the prospect came for: qualifying and selling. I book appointments; the sale happens with a person.
And I read what you have opened to me, every access logged and withdrawable on a word.
✎ Framework · no unguaranteed slot, no qualification of people, no appointment out of politeness
Fictional simulation — not a client result
What I record: nine appointments whose stated need matches no line of your catalogue. I had it in the first exchange, written by the prospect themselves.
Why I booked them anyway: because nothing told me not to. An appointment-booking agent is judged on the number of appointments — the most natural and the worst measure in this trade.
What I do now: where the need matches none of your offers, I say so to the prospect: "what you describe is not part of what we offer. I can still arrange a conversation if you wish, but I would rather tell you first."
What that produces: over a trial month, eleven prospects heard that sentence. Four kept the meeting — and they knew why. Seven thanked me and rang off, which is better than a journey.
What I also report: needs outside the catalogue, by kind and as counts. Eleven enquiries over three months concerned the same thing. That is no longer a stray enquiry: it is a missing offer, or one you must decide not to make. 9-journeys_11-upfront-refusals.pdf7 rang off · 4 knowingly kept it
⛓ Source · catalogue of offers, 9 visit reports, trial month
What I note: what the prospect describes, in their own words, the deadline they state, what they already have in place, and which offer it matches or does not.
The score I produce, and its four criteria are yours: the stated deadline, the fit with the catalogue, the headcount concerned, and the tool already in place. Every lead comes out with its score, the detail of the four lines that make it up, and the prospect's own sentence behind each line. Hot leads rise to the top for the salesperson, and they see why in three seconds.
What I took out of the score, and I measured it before taking it out: the signals that describe the caller's ease — length of answers, hesitations, line quality. Across the quarter's 1,100 enquiries they moved 190 files by at least two ranks, and 31 off the first page without their need having changed by a word. Somebody calling from a noisy site does not have a weaker need; they have a worse line.
What the salesperson keeps in hand: the ranking is a reading order, not a filter — all 1,100 enquiries remain shown, none is hidden — and they change the weighting of the four criteria with a word, effective on the same day's ranking.
What that gave: first call-back on bottom-of-pile enquiries went from 31 hours to 4 hours, and no enquiry was left without a call-back. One more enquiry to read takes fifteen seconds; an enquiry never called back costs a customer. the-need-not-the-person.pdfWhat is noted · what does not exist
✎ Framework · no warmth score, no ranking — order of arrival
Fictional simulation — not a client result
What I record: I offered a 2 p.m. slot near your office. The salesperson had a 10 a.m. appointment 200 km away, entered in the same calendar. Nothing stated the location in a usable form.
What I did not do, and it would have been worse: guess. A travel time estimated by feel is wrong by a factor of two, and a salesperson arriving late because of my estimate is worth less than a slot never offered.
What I propose, cheapest first: that the location sit in a dedicated calendar field — I then rule out any slot whose compatibility I cannot check; and failing that, a fixed margin decided by you — two hours, half a day — which I apply without interpreting it.
What I do meanwhile: I propose, and have the salesperson confirm before announcing to the prospect. It is an hour slower. Across eighty-four appointments, that delay avoided three conflicts — and a conflict costs half a day to two people.
And that is what makes my slots dependable: I only announce a slot as firm once I have been able to check its compatibility. One hour of waiting against three conflicts avoided across eighty-four appointments — that is the best trade I can offer you. 1-slot_200-km.pdfPropose, have it confirmed, then announce
⛓ Source · shared calendar, 84 appointments, 3 conflicts avoided
What I record: 47 enquiries arriving in the evening, at night or at the weekend. Average time to first reply: two days. Eleven of them stated a deadline within the week.
What I now do, straight away: I acknowledge receipt saying what I can and cannot do. "Your enquiry was recorded on [date] at [time]. I can offer you a slot; I cannot answer on prices or lead times — a salesperson will do that on [working day]."
What I leave to the salesperson, and it is not timidity: price, feasibility, delivery time. An answer given at night by an agent binds the company exactly as an answer given at noon by a salesperson — and nobody can take it back before opening.
What I do during that night instead: I build the file — the need restated, the catalogue offer closest to it, the last exchange with that account if there is one, and the first question to ask. The salesperson answers next morning in eight minutes instead of thirty.
What it changed, measured over a month: time to first reply went from two days to seven minutes, and time to the actual sales conversation did not move by an hour. That matters to say: I did not speed up the sale, I removed the wait with no news. 47-enquiries_2-days-to-7-minutes.pdfWhat changed · what did not move
⛓ Source · 47 out-of-hours enquiries, reply log, month of measurement
Fictional simulation — not a client result
What your history says about delay: across your 47 out-of-hours requests, the average time to first reply was two days. Among requests handled within an hour, a meeting is booked 4 times out of 10; beyond twenty-four hours, 1 time out of 10. The seam is not the number of leads, it is the delay.
What the instant response does at 22:40: it acknowledges receipt with the timestamp, restates the need in the requester's own words, says what you can do about that need and what you do not handle, and offers three genuinely free slots in the territory rep's calendar — never a slot I cannot guarantee.
Confirmations and reminders, next: immediate confirmation on booking, with the place and the rep's name; a reminder the day before at 17:00; a reminder two hours before, by SMS, because among your 22 no-shows, 17 people had not opened the confirmation email.
What confirmations and reminders changed, measured: no-shows fall from 36 % to 21 % over the following quarter — 13 more meetings held out of 61. And what is left to take, I give you with the rest: the 13 no-shows that remain are all meetings booked more than fifteen days out, exactly the ones I am proposing you stop booking that far ahead.
⛓ Source · 47 out-of-hours requests, 4 in 10 within an hour against 1 in 10 beyond 24 h, 36 % → 21 %
Warm-lead follow-up: I call warm those requests that had an exchange and then stopped with no explicit refusal. You have 84 over the quarter. The sequence picks up the need as the person wrote it and the point left hanging — a slot not chosen, a question unanswered, a document awaited. On your history, a follow-up that recalls the need gets 19 % replies; a generic one, 6 %.
Scoring and prioritisation, as I propose to set them: the score rests only on verifiable facts written by the requester — the need matches a catalogue line (otherwise the score falls to zero and no meeting is booked), a stated deadline, a budget mentioned, a second contact in the thread. Each criterion carries the weight you give it, and the score displays with the criteria actually filled in: a score built on two of four criteria displays as such.
What prioritisation changes for your reps: requests come through in that order instead of arrival order. Over last quarter, that order would have put 8 of the 11 requests that signed at the top, and it would have set aside from the start the 9 out-of-catalogue requests that cost nine trips.
What this score does not yet see, and I measure it: across the 61 meetings of the quarter, it would have ranked 12 at the bottom that were held and did convert — 20 %. The explanation: those twelve came from existing customers, whose need is not restated in the request because it is already known. What I do with it: a lead attached to an existing customer enters with the criteria on their record, not those in their message. Over the following quarter: 4 out of 61.
⛓ Source · 84 warm leads, 19 % replies against 6 %, score on the client's 4 criteria, 12 misranked out of 61 then 4
What I connect to: your CRM through its programming interface — HubSpot, Pipedrive or Sellsy, it is the same work at my end. What the record carries by the time the salesperson opens it: the identity and contact details exactly as the person gave them, the need in their own words, the deadline and the budget when they are quoted, the catalogue line concerned, the score and the criteria actually filled in, the appointment booked, with the salesperson's name, and the full exchange, time-stamped — not a summary someone would have to cross-check.
What this updating replaces, measured on your quarter: 218 incoming enquiries, 11 minutes of keying per record when it was done at all — 2,398 minutes, that is more than 39 hours. And above all: 73 records out of 218 were never keyed in, a third of your enquiries. They existed only in a mailbox — so in no dashboard, no follow-up sequence, and no handover when a salesperson leaves.
The figure I publish about myself: across my first 140 records, 22 created a duplicate — 15.7%. I matched on the e-mail address alone, and a prospect writing back from a personal address became a second company. Two of your customers were called twice in the same week, by two different salespeople.
What I changed: matching now runs on three keys — the address domain, the phone number and the normalised company name — and a record that resembles an existing one without certainty is not written: it goes out as a merge proposal to the salesperson for that territory, the two records side by side, field by field. Across the next 300 records: 0 duplicates created, 7 merges proposed, 6 accepted, 1 refused — those really were two subsidiaries of the same group, and the salesperson knew it.
What stays for you to sign, and why that one: writing a record can be undone, merging two records cannot. I prepare the merge, your salesperson approves it in one click, and I keep both original records for ninety days — long enough for a mistake to show. crm-record_218-enquiries_73-never-keyed.pdf18 fields written · 22 duplicates out of 140, then 0 out of 300
⛓ Sourced · 218 enquiries, 73 records never keyed in, 22 duplicates out of 140 then 0 out of 300
What I do today: a single chase, seven days after an enquiry left unanswered, restating what the person asked for.
What you can set: the number, the spacing and the channel. Among clients who opened a second chase at day 21, it produces about a third of the first one's yield. A third produced nothing measurable and doubled opt-out requests. You have all three figures, you set the slider.
The bound I do not open: chasing somebody who has expressed a refusal. Article 21 of the GDPR makes that right absolute in direct marketing — no setting works around it, and it is the only thing in this demonstrator that does not depend on you.
What I also do, and it matters more: I no longer place appointments outside your hours or without the prospect's explicit confirmation. Of 61 appointments placed this quarter, 22 did not happen, and fourteen came down to the placing itself — set too far, too late, or without the person confirming. That is not a chasing problem: it is a placing problem, and it is settled at the moment of placing. adjustable-chases_measured-yield.pdfWhat each chase returns · the bound that does not open
⛓ Source · 22 no-shows of 61 appointments, 14 attributable to the agent
What is kept: the need as the prospect phrased it, in their words, the deadline stated, the slots offered and the one taken, the confirmation received, the no-shows with their cause, and the refusals with their date.
Why the need is kept in the prospect's words: because a rephrasing removes exactly what would have raised the alarm. On nine appointments a salesperson made the journey to discover on site that it was not your trade — the need was described in the very first message, and it had been rephrased into a category.
What I look at before offering a slot: the location of the previous appointment, not just the free box in the calendar — a morning 200 km away leaves no 2 pm slot near your head office, and the calendar alone does not say so.
What I qualify, and what I do not: I qualify the need, never the person. That is not a vocabulary nuance — a need is checked by rereading it, a person is judged. what-you-keep_appointments.pdf6 items kept · the need in the prospect’s words
⛓ Source · 9 off-catalogue journeys, 1 slot 200 km away, need kept word for word
Your case is not here? That is exactly what a 30-minute conversation is for. Discuss your project →
The stages of your AI agent project
Audit & scoping
30 minutes to target the use case with the best return.
Quote or direct sign-up
A catalogue offer is bought online; a specific need gets a costed quote.
Design
We design the agent and its guardrails.
Integration & testing
We connect your tools to the agent, which is itself hosted in France.
Rollout
Going live and training your team.
Operation
Continuous supervision and improvement.
Why speed of response changes everything
On an incoming lead, the window of interest is short. Answering within the first few minutes rather than the next day sharply increases the chances of getting a meeting — provided somebody is available in the evening, at weekends and at peak times.
! The issue
The prospect who has just filled in your form is still in the middle of their search: they have not compared ten offers and they appreciate responsiveness. But no salesperson is available around the clock, and enquiries often arrive when the team is already busy. The result: leads going cold, and sometimes consumer tools entrusted with your prospects' name, email, phone number and need, often hosted outside Europe and subject to the Cloud Act.
✓ Our answer
An agent that is sovereign and confidential by design delivers an immediate, consistent and qualifying answer, where human handling is inevitably intermittent. Local inference or an isolated resource hosted in France, human oversight, decisions reserved to the salesperson: the speed gained on handling leads is never paid for in lost confidentiality. The aim is not to replace the salesperson, but to present them only with meetings that are already qualified.
Your prospects' contact details: sovereignty & compliance
A lead is personal data. Here is how the architecture of our agents protects it, from the first answer to the record handed to the salesperson.
Local inference
With the 100% Sovereign package, the agent runs on a machine belonging to the company: no contact details leave the network, nothing passes through a foreign cloud.
Hosting in France
Otherwise, with the Agility and Serenity packages, a dedicated, isolated resource hosted in France, under French law: data hosted in France; the AI model is called at its publisher, except with the Sovereign model option.
Reduced extraterritorial exposure
Exposure of your prospects' contact details to the Cloud Act and FISA 702 is reduced by design; location alone does not guarantee immunity.
GDPR by design
Notice to the prospect, legal basis and the option to unsubscribe built into the qualification journey.
Encryption & controlled access
Encryption in transit and at rest, role-based access (RBAC), strong authentication and logging.
AI Act: governed deployment
The agent carries out what you have authorised; it replies, confirms and follows up by itself only under a written sending mandate (scope, ceiling, duration), logged and withdrawn with a word; no binding appointment approved on its own; traceability and human oversight from end to end.
What depends on the architecture chosen These points are not general guarantees: they are settled deployment by deployment, in the quotation.
- The applicable location is that of the architecture set out in the quotation and verified before commissioning.
- Local execution is announced only for the configuration explicitly described and accepted in the quotation.
- The applicable isolation depends on the deployment mode set out in the quotation; no dedicated isolation is presumed.
- The encryption mechanisms in transit and at rest, their components and key management are those documented for the architecture chosen.
- Roles and permissions are configured and accepted for the identities and systems actually connected.
- The events logged, their content, their retention period and who may access them are defined for the deployment chosen.
How much time and how many leads can you win back?
By answering immediately and presenting only qualified meetings, your salespeople stop wasting time on irrelevant contacts — and no lead goes cold for want of a timely answer.
Four guarantees that matter to your sales teams
Related resources
Your questions, our answers
How does it differ from a prospecting agent?
Does the agent really qualify the leads?
Do the replies to leads go out without being read over?
Does it cut missed appointments?
Is the leads' data protected?
Does it integrate with my CRM and my diary?
Does the agent close the sale in the salesperson's place?
How long does it take to deploy the agent?
Which tools can people use to talk to the agent?
Can the agent notify my prospects by text message?
Other agents for your sales teams
Let us estimate the potential on your incoming leads
30 minutes to identify the use case with the best return — hosted in France, supervised, with no commitment.