● B2B offer — CRM & business development
CRM: complete records, leads prioritised on your criteria
A lead handled quickly and well starts with a complete record: sector, size, contact, history of exchanges. Your agent completes these elements from your own sources, applies the priority criteria you have set and presents the leads in that order, each one with what justifies its position. Hosted in France: your prospect data stays with you. Sales management adjusts the criteria and assigns the leads.
Hosted in France
Prospect data protected
GDPR & AI Act: governed deployment
Human oversight
Updated on
Deployed in a few weeks
CRM enrichment · hosted in France
Complete and prioritise the leads received this week.
Records completed from your internal sources: sector, size, contact identified, history of exchanges attached.
Your priority criteria are applied: target sector, company size, channel of entry, how long the enquiry has been open.
Every lead shows what explains its position.
🔗 Sourced · the CRM and internal sources
Your priority criteria are applied: target sector, company size, channel of entry, how long the enquiry has been open.
Every lead shows what explains its position.
🔗 Sourced · the CRM and internal sources
One lead is in the wrong place, to my mind.
The detail of the calculation is shown: criterion by criterion, what was taken into account and where the information came from.
Adjusting a criterion changes the order for every lead that follows — that is your setting, not mine.
✎ Support · criteria adjustable by you
Adjusting a criterion changes the order for every lead that follows — that is your setting, not mine.
✎ Support · criteria adjustable by you
Local inference · no data outside the EU
CRM data hosted in France
Sovereign by designLocal inference or hosting in France
GDPR & AI Act: governed deploymentTraceability & human oversight
TurnkeyDesigned, installed and operated for you
The sales management decidesThe agent prepares, never rules
✦ In brief
A Blue Lemon Agent CRM agent completes your records from your internal sources and prioritises your leads according to the criteria you have set, showing the detail of the calculation for each one. The criteria remain adjustable by sales management. It runs on local inference or is hosted in France: your prospect data stays with you, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
100%
hosted in France in the target architecture
0
transfer outside the EU in the target architecture
6
sales uses ready to deploy
0
decision taken without human approval
These figures describe our offer, not results measured at a client. How large the gain is on your volume of leads and the richness of your CRM is confirmed by a pilot.
The context
What does an AI agent bring to your CRM?
A salesperson moves fastest when the record is complete and the order of play is clear. Both of those can be prepared.
! The issue
A lead handled well means a complete record and a readable order of play. Completing records is steady work, and prioritising means applying the same criteria to everyone. The agent does both: it enriches from your sources and applies your criteria, showing for each lead what explains its position.
✓ Our answer
Sales management keeps control of what matters: the priority criteria, their weighting and the assignment of leads. The detail of the calculation is visible for every record, which makes the ranking open to challenge and to adjustment. Local inference or an isolated resource hosted in France: your prospect data and the make-up of your portfolio are entrusted to no third party.
The decisive point
Your prospect data and your portfolio: sovereignty & compliance
Your prospect portfolio is one of your most valuable assets, and your records contain personal data. Here is how they are protected.
Local inference
The agent can run on a machine belonging to your organisation: no prospect data and no CRM record leaves the network.
Hosting in France
Otherwise, a dedicated and isolated resource hosted in France, under French law — your CRM records and your incoming leads: processing and access within the European Union targeted by the architecture.
Reduced extraterritorial exposure
For your prospect data and your portfolio, the architecture aims to reduce exposure to the Cloud Act and FISA 702; being located in France or in the European Union does not, on its own, guarantee immunity.
Isolated resource
No pooling: an environment strictly dedicated to your company and its commercial criteria.
The detail of the calculation always visible
Every priority shows the criteria taken into account and the source of each piece of information; encryption, role-based access and logging of the enrichments performed.
AI Act: governed deployment
The agent is strictly in support; no lead is set aside and none is assigned to a salesperson automatically; 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.
- 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.
For strategic accounts and deals under negotiation, SecNumCloud and reinforced hosting are options depending on your requirements. A single architecture is designed to answer both the GDPR and extraterritorial exposure. Designed for deployment in line with the GDPR and the AI Act, after the processing, roles and context-specific risks have been assessed.
Demonstration
See the agent at work
4 real 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.
CRM enrichment · hosted in France
Three flags this morning across your CRM's 1,200 records.
· 61 duplicates: the same company, two records, two salespeople calling it.
· 210 records carry a headcount over three years old, 34 of them on companies that have changed bracket.
· 190 fields have no recorded source. I do not know where they came from — and nobody does. morning-watch_3-flags.pdf61 duplicates · 190 fields with no source
⛓ Source · 1,200 records, public company register, enrichment log
· 61 duplicates: the same company, two records, two salespeople calling it.
· 210 records carry a headcount over three years old, 34 of them on companies that have changed bracket.
· 190 fields have no recorded source. I do not know where they came from — and nobody does. morning-watch_3-flags.pdf61 duplicates · 190 fields with no source
⛓ Source · 1,200 records, public company register, enrichment log
"No source": what does that mean for a CRM field?
That it can neither be checked, nor corrected, nor defended.
What I record: 190 fields — headcounts, revenues, directors' names — state neither their origin nor their date. They were keyed in by hand, imported from a file, or taken from a provider whose contract has ended.
Why it is a concrete problem: a salesperson calling with a wrong headcount loses credibility in the first minute. And a person asking where information about them came from is entitled to know — that is an obligation, not a courtesy.
What I do: every field I write carries its source and its date. Public register, the company's website, the prospect's own statement, or "keyed in by a salesperson on …". A field with no source is not written.
What I do with the 190: I mark them as unsourced, visible as such on the record. I do not erase them — some are right, and erasing accurate information because it is badly traced is a pure loss. Nor do I use them to compute a score. 190-fields-with-no-source.pdfA field with no source can be neither checked, corrected nor defended
⛓ Source · 190 fields with no origin, marked and excluded from scoring
What I record: 190 fields — headcounts, revenues, directors' names — state neither their origin nor their date. They were keyed in by hand, imported from a file, or taken from a provider whose contract has ended.
Why it is a concrete problem: a salesperson calling with a wrong headcount loses credibility in the first minute. And a person asking where information about them came from is entitled to know — that is an obligation, not a courtesy.
What I do: every field I write carries its source and its date. Public register, the company's website, the prospect's own statement, or "keyed in by a salesperson on …". A field with no source is not written.
What I do with the 190: I mark them as unsourced, visible as such on the record. I do not erase them — some are right, and erasing accurate information because it is badly traced is a pure loss. Nor do I use them to compute a score. 190-fields-with-no-source.pdfA field with no source can be neither checked, corrected nor defended
⛓ Source · 190 fields with no origin, marked and excluded from scoring
What exactly is the agent for?
It keeps records current and flags situations. It ranks nobody and decides no priority.
What it does: groups duplicates, updates what is public and dated, flags stale fields, and attaches to each record the verifiable signals — a site opening, an announced hiring drive, a published tender.
Routing follows who can decide: a duplicate goes to both salespeople concerned, together, because merging two records is a commercial decision; a stale field to whoever owns the account; a strong signal to whoever covers the territory; an unusual score to nobody — it sits on the record, with what produced it.
With a monthly summary: fields corrected and their source, duplicates handled, conversion rate of the leads I scored badly, and fields still unsourced.
Three settings that belong to you. The score is about a company situation, and the record always says which fact produced it — you can add the contact's role if you judge it carries weight. You decide the collection scope: mine stops at the professional context, and I will tell you why widening it costs more than it earns. And I can order a salesperson's day if you authorise it — Nathalie Roussel switched it on for her team, with the list built every morning at 7.
✎ Framework · 3 openable settings · prioritisation live on one team
What it does: groups duplicates, updates what is public and dated, flags stale fields, and attaches to each record the verifiable signals — a site opening, an announced hiring drive, a published tender.
Routing follows who can decide: a duplicate goes to both salespeople concerned, together, because merging two records is a commercial decision; a stale field to whoever owns the account; a strong signal to whoever covers the territory; an unusual score to nobody — it sits on the record, with what produced it.
With a monthly summary: fields corrected and their source, duplicates handled, conversion rate of the leads I scored badly, and fields still unsourced.
Three settings that belong to you. The score is about a company situation, and the record always says which fact produced it — you can add the contact's role if you judge it carries weight. You decide the collection scope: mine stops at the professional context, and I will tell you why widening it costs more than it earns. And I can order a salesperson's day if you authorise it — Nathalie Roussel switched it on for her team, with the list built every morning at 7.
✎ Framework · 3 openable settings · prioritisation live on one team
Is your score often wrong?
Enough that I publish the gap rather than let it close on its own.
The flaw of any learned score: it learns from your won deals. It therefore scores high whatever resembles your current customers — and low anything that resembles them little, including a segment you have never approached.
What makes the flaw invisible: a low-scored lead is not called back. Not being called back, it does not convert. So the score is right, and it will be right indefinitely. That is a prophecy, not a prediction.
What I do to break it: 5 % of low-scored leads are drawn at random and worked like the others, without the salesperson knowing they come from that draw.
What the control group gave: 9 % conversion on low-scored leads called anyway, against 14 % on high-scored leads. The gap is real — and it is four times smaller than the score suggested. 22 deals were signed on leads I had scored badly. control-group-5-percent_9-against-14.pdfA prophecy, not a prediction
⛓ Source · 5 % control group, 9 % against 14 % conversion, 22 deals signed
The flaw of any learned score: it learns from your won deals. It therefore scores high whatever resembles your current customers — and low anything that resembles them little, including a segment you have never approached.
What makes the flaw invisible: a low-scored lead is not called back. Not being called back, it does not convert. So the score is right, and it will be right indefinitely. That is a prophecy, not a prediction.
What I do to break it: 5 % of low-scored leads are drawn at random and worked like the others, without the salesperson knowing they come from that draw.
What the control group gave: 9 % conversion on low-scored leads called anyway, against 14 % on high-scored leads. The gap is real — and it is four times smaller than the score suggested. 22 deals were signed on leads I had scored badly. control-group-5-percent_9-against-14.pdfA prophecy, not a prediction
⛓ Source · 5 % control group, 9 % against 14 % conversion, 22 deals signed
Why publish a figure that works against you?
Because it is the only one that shows what the rest is worth.
What the figure says: my score sorts usefully — 14 % against 9 % is not nothing when organising a week. What it does not say: that a low-scored lead is worthless. One lead in eleven signs anyway.
What that changes in use: the score orders a list, it does not remove lines from one. A low-scored lead stays on the list, at the bottom, with the reason for its score. A lead removed from the list never comes back.
What the record always shows: the two or three facts that produced the score, each with its source and date. "Headcount up from 12 to 40 in eighteen months, public register, March 2026." A salesperson can therefore disagree with the score, and their disagreement is information: I count the times they were right.
What I do measure, and it is the more useful of the two for you: how right the score is, not how obedient the salespeople are. A score you are obliged to follow stops being verifiable — nobody ever disproves it again, and you lose the only way of knowing what it is worth. Your salespeople keep control of their day, and you keep a score that corrects itself. what-produces-a-score.pdfA score you are obliged to follow stops being verifiable
⛓ Source · score explained fact by fact, disagreements counted
What the figure says: my score sorts usefully — 14 % against 9 % is not nothing when organising a week. What it does not say: that a low-scored lead is worthless. One lead in eleven signs anyway.
What that changes in use: the score orders a list, it does not remove lines from one. A low-scored lead stays on the list, at the bottom, with the reason for its score. A lead removed from the list never comes back.
What the record always shows: the two or three facts that produced the score, each with its source and date. "Headcount up from 12 to 40 in eighteen months, public register, March 2026." A salesperson can therefore disagree with the score, and their disagreement is information: I count the times they were right.
What I do measure, and it is the more useful of the two for you: how right the score is, not how obedient the salespeople are. A score you are obliged to follow stops being verifiable — nobody ever disproves it again, and you lose the only way of knowing what it is worth. Your salespeople keep control of their day, and you keep a score that corrects itself. what-produces-a-score.pdfA score you are obliged to follow stops being verifiable
⛓ Source · score explained fact by fact, disagreements counted
And the order my salespeople see leads in — where does that come from?
Prioritisation follows your criteria, written by your sales director on 14/03, and the arithmetic is shown on every record.
Your six criteria and their weights, as you set them: target sector 25, headcount between 50 and 500 20, competitor equipment identified 20, inbound rather than outbound 15, area covered by a salesperson 10, age of last contact 10.
In practice, on one record: "71 out of 100 — sector 25 (activity code, public register, read 02/08); headcount 20 (142 staff, same source); competitor equipment 0 (no mention); inbound 15 (web form of 28/07); area 10; recency 1 (last contact 11 months ago)". The salesperson reads the rank and the reason for the rank on the same line.
The criteria are adjustable, and you see the effect before you adjust them. Say you want "competitor equipment identified" to go from 20 to 30: I run the rule back over your last twelve months before you approve it. It would have lifted 84 records into the top quarter, including 7 of the 22 deals you won — and it would have pushed 3 others down, also won. You decide on those figures, not on a hunch; an approved criterion applies to every lead from the next record on.
One adjustment turned against me, and I am publishing it: the weight on recency was 25 in June. It lifted dormant records with no buying signal at all: 96 leads prioritised, 4 meetings. Moving it from 25 to 10 gave 41 leads prioritised for 9 meetings over the following period. What prioritisation still does not do: it orders company situations, never people — and none of your six criteria bears on an individual.
✎ Framework · prioritisation by your criteria — the arithmetic is shown lead by lead
Your six criteria and their weights, as you set them: target sector 25, headcount between 50 and 500 20, competitor equipment identified 20, inbound rather than outbound 15, area covered by a salesperson 10, age of last contact 10.
In practice, on one record: "71 out of 100 — sector 25 (activity code, public register, read 02/08); headcount 20 (142 staff, same source); competitor equipment 0 (no mention); inbound 15 (web form of 28/07); area 10; recency 1 (last contact 11 months ago)". The salesperson reads the rank and the reason for the rank on the same line.
The criteria are adjustable, and you see the effect before you adjust them. Say you want "competitor equipment identified" to go from 20 to 30: I run the rule back over your last twelve months before you approve it. It would have lifted 84 records into the top quarter, including 7 of the 22 deals you won — and it would have pushed 3 others down, also won. You decide on those figures, not on a hunch; an approved criterion applies to every lead from the next record on.
One adjustment turned against me, and I am publishing it: the weight on recency was 25 in June. It lifted dormant records with no buying signal at all: 96 leads prioritised, 4 meetings. Moving it from 25 to 10 gave 41 leads prioritised for 9 meetings over the following period. What prioritisation still does not do: it orders company situations, never people — and none of your six criteria bears on an individual.
✎ Framework · prioritisation by your criteria — the arithmetic is shown lead by lead
What do you enrich from?
Public company sources, the prospect's own website, and what they told you themselves.
What I use: the public company register (legal form, declared headcount, directors, sites), the company's website, published tenders, and what the prospect entered in your forms. Every field carries its source and date.
What I do when an address is missing, and I have already done it on your base: I go and find it where the company published it itself — legal notices, contact page, tender response form, register filing. Across the 190 fields with no source, I re-sourced 147 work addresses published by the company itself, each with its web address and the date it was read; the remaining 43 go to the switchboard, with the role targeted in the subject line — 31 useful replies in six weeks. A published address beats a guessed one: it can be quoted the day the recipient asks where you got it.
What stays off the path, and it is not caution: deducing an address from a pattern (first.last@…) — that is not data, it is a hypothesis sent to somebody; scraping a personal profile on a social network; recording what a person publishes privately; buying a list whose origin is undocumented — article 14 of the GDPR requires the source to be stated by the first message at the latest, and a list with no origin makes that sentence impossible to write.
The limit, in one sentence: I enrich a company, not a person. A person appears on a record because they hold a role and have spoken to you, not because I reconstructed their history.
What that costs you: your records are less filled in than a data vendor promises. What is on them is verifiable, dated, and you can answer the question "where did you get this?". what-is-collected_what-is-not.pdfA guessed address is a hypothesis sent to somebody
✎ Framework · no deduced address · no personal profile · no undocumented origin
What I use: the public company register (legal form, declared headcount, directors, sites), the company's website, published tenders, and what the prospect entered in your forms. Every field carries its source and date.
What I do when an address is missing, and I have already done it on your base: I go and find it where the company published it itself — legal notices, contact page, tender response form, register filing. Across the 190 fields with no source, I re-sourced 147 work addresses published by the company itself, each with its web address and the date it was read; the remaining 43 go to the switchboard, with the role targeted in the subject line — 31 useful replies in six weeks. A published address beats a guessed one: it can be quoted the day the recipient asks where you got it.
What stays off the path, and it is not caution: deducing an address from a pattern (first.last@…) — that is not data, it is a hypothesis sent to somebody; scraping a personal profile on a social network; recording what a person publishes privately; buying a list whose origin is undocumented — article 14 of the GDPR requires the source to be stated by the first message at the latest, and a list with no origin makes that sentence impossible to write.
The limit, in one sentence: I enrich a company, not a person. A person appears on a record because they hold a role and have spoken to you, not because I reconstructed their history.
What that costs you: your records are less filled in than a data vendor promises. What is on them is verifiable, dated, and you can answer the question "where did you get this?". what-is-collected_what-is-not.pdfA guessed address is a hypothesis sent to somebody
✎ Framework · no deduced address · no personal profile · no undocumented origin
And if a prospect asks what you hold on them?
The record prints as a readable statement, field by field, with each source.
What the statement contains: each field, its value, its source and the date it was obtained, the record's origin (form filled in, trade show, inbound call, list), the exchanges recorded, and the score with the facts that produced it.
Why the score is in it: because it is data about them, just as their headcount is. A score you would not show is a score they could not contest.
What I do with an objection: the record leaves commercial work and I keep strictly what is needed not to recreate it next month — that is, enough to know they must not be approached again, and nothing more.
What I have already prepared when an objection arrives: the effect is immediate — the record leaves commercial work within the second, before any review, because waiting for a decision would already be approaching them. And the file goes with it: date and channel of the request, origin of the record, the 3 approaches received over twelve months, and the exact sentence the objection rests on. It is notified to whoever owns the account and to whoever answers for data at your end, with its date. What is left to that person is the final word — well founded or not — and it is given in two minutes on a complete file, not by reconstructing a history.
What that implies at your end: a person these requests reach. Without them, the statement I produce is sent by nobody. statement-of-a-prospect-record.pdfA score you would not show is a score they could not contest
✎ Framework · full field-by-field statement · objection notified, never handled silently
What the statement contains: each field, its value, its source and the date it was obtained, the record's origin (form filled in, trade show, inbound call, list), the exchanges recorded, and the score with the facts that produced it.
Why the score is in it: because it is data about them, just as their headcount is. A score you would not show is a score they could not contest.
What I do with an objection: the record leaves commercial work and I keep strictly what is needed not to recreate it next month — that is, enough to know they must not be approached again, and nothing more.
What I have already prepared when an objection arrives: the effect is immediate — the record leaves commercial work within the second, before any review, because waiting for a decision would already be approaching them. And the file goes with it: date and channel of the request, origin of the record, the 3 approaches received over twelve months, and the exact sentence the objection rests on. It is notified to whoever owns the account and to whoever answers for data at your end, with its date. What is left to that person is the final word — well founded or not — and it is given in two minutes on a complete file, not by reconstructing a history.
What that implies at your end: a person these requests reach. Without them, the statement I produce is sent by nobody. statement-of-a-prospect-record.pdfA score you would not show is a score they could not contest
✎ Framework · full field-by-field statement · objection notified, never handled silently
We want conversion rates per salesperson. Can that be done?
Yes, it is lawful and I set it up with the file that makes it safe. First, here is what I have observed.
What I supply to install it: the prior information notice to the people concerned (art. L1222-4 of the Labour Code), the works council consultation file and a proportionate scope. The French DPA restates that an employer has the power to frame and monitor staff activity — it is the conditions that are regulated, and they are prepared.
What I have observed elsewhere: an individual conversion rate makes low-odds quotes disappear. The rate rises, the number of deals falls, and the dashboard shows an improvement. The number of deals signed does not produce that effect — same steering, without the distortion.
So what I propose: both tables, with the second brought forward. You decide.
One point to know if you open the first: my 5 % control group — the badly scored leads worked anyway — does not survive an individual measure on conversion. It brought in 22 deals this half-year. On deals signed, it holds. opening-individual-indicators.pdfWhat opens · works council file ready · the indicator that does not distort
⛓ Source · French DPA, art. L1222-4 · works council file supplied · control group worth 22 deals
What I supply to install it: the prior information notice to the people concerned (art. L1222-4 of the Labour Code), the works council consultation file and a proportionate scope. The French DPA restates that an employer has the power to frame and monitor staff activity — it is the conditions that are regulated, and they are prepared.
What I have observed elsewhere: an individual conversion rate makes low-odds quotes disappear. The rate rises, the number of deals falls, and the dashboard shows an improvement. The number of deals signed does not produce that effect — same steering, without the distortion.
So what I propose: both tables, with the second brought forward. You decide.
One point to know if you open the first: my 5 % control group — the badly scored leads worked anyway — does not survive an individual measure on conversion. It brought in 22 deals this half-year. On deals signed, it holds. opening-individual-indicators.pdfWhat opens · works council file ready · the indicator that does not distort
⛓ Source · French DPA, art. L1222-4 · works council file supplied · control group worth 22 deals
And after six months, what have we gained?
A database you can defend, and twenty-two deals you would not have had.
What was corrected: 61 duplicates resolved — same company, two records, two salespeople calling it —, 210 headcounts brought up to date, 34 of them out of bracket, and 190 fields with no origin identified and marked.
What the control group returned: 5 % of low-scored leads are drawn at random and worked like the others. 9 % conversion against 14 % on high-scored leads — the gap is real, and four times smaller than the score suggested. 22 deals signed on leads I had scored badly.
What that guarantees you: the score orders a list, it never removes lines from one — and a lead removed from a list never comes back. You keep the whole of your addressable market, simply sorted by likelihood.
What you have on screen for each record: every field, its value, its source and its date. A prospect asking where information about them came from gets the answer in seconds — and a salesperson calling with a correct headcount does not lose credibility in the first minute. what-you-get-back_crm.pdf6 items to hand · addressable market intact
⛓ Source · 61 duplicates, 210 headcounts, control group 9 % vs 14 %, 22 deals
What was corrected: 61 duplicates resolved — same company, two records, two salespeople calling it —, 210 headcounts brought up to date, 34 of them out of bracket, and 190 fields with no origin identified and marked.
What the control group returned: 5 % of low-scored leads are drawn at random and worked like the others. 9 % conversion against 14 % on high-scored leads — the gap is real, and four times smaller than the score suggested. 22 deals signed on leads I had scored badly.
What that guarantees you: the score orders a list, it never removes lines from one — and a lead removed from a list never comes back. You keep the whole of your addressable market, simply sorted by likelihood.
What you have on screen for each record: every field, its value, its source and its date. A prospect asking where information about them came from gets the answer in seconds — and a salesperson calling with a correct headcount does not lose credibility in the first minute. what-you-get-back_crm.pdf6 items to hand · addressable market intact
⛓ Source · 61 duplicates, 210 headcounts, control group 9 % vs 14 %, 22 deals
Local inference · no data outside the EU
Your case is not here? That is exactly what a 15-minute conversation is for. Book the free audit →
Use cases
What does the agent actually do?
One agent, several moments in the sales cycle. All these uses work in support, subject to your approval.
Included in your agent
The 3 capabilities essential to this promise are included, at no extra cost.
From 542 € excl. VAT / month
Enriching the records
Completes sector, size, contact and history from your internal sources.
Prioritising on your criteria
Applies the criteria you have set, with the detail of the calculation.
Adjustable criteria
Adjusting a criterion updates the order for every lead that follows.
Controls and safeguards
These 4 controls are built into the agent: they frame what it does, whatever plan you pick. They are not chosen and are not added to your order.
Human validation, exceptions and escalation
Status, safe closure and audit trail
Prioritise opportunities on explicit criteria
Personalise without breaching consent or prospecting rules
What the agent must be connected to
This connection is required for the agent to work. It concerns your information system and is scoped during the audit.
Prepare the next action and write back to the CRM without loss
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.
Sales & CRM
For meeting notes and follow-ups, a dedicated sales agent takes them on.
Sales agent (meeting notes, follow-ups, CRM) from 592 € excl. VAT / month Sales & CRM →Qualifying sales emails
To sort incoming enquiries, a dedicated qualification agent takes over.
Sales email qualification agent from 504 € excl. VAT / month Email qualification →Sales proposals
Once the lead is qualified, a dedicated proposals agent prepares the offer.
Sales proposal agent from 489 € excl. VAT / month Sales proposals →
Does your need fall outside this?
Book the free audit →
Build your agent
In 15 minutes we identify the most relevant agent — without oversizing the project.
The gain
How many leads can a team handle in a useful order?
By taking on the enrichment and the ordering, the effort shifts towards contact and live qualification. How large the gain is depends on your volume and remains to be confirmed by a pilot.
Completing the CRM records
Putting the leads in order
Finding the context of a lead
Indicative figures, not contractual, to be confirmed by a pilot on your volume of leads and the richness of your CRM. The priority criteria and the assignment of leads belong to sales management: the agent applies and explains, it does not decide.
How it works
The stages of your AI agent project
1
Audit & scoping
15 minutes to target the use case with the best return.
2
Quote or direct sign-up
A catalogue offer is bought online; a specific need gets a costed quote.
3
Design
We design the agent and its guardrails.
4
Integration & testing
We connect your tools to the agent, which is itself hosted in France.
5
Rollout
Going live and training your team.
6
Operation
Continuous supervision and improvement.
Pricing
One package, one agent
A CRM agent (enrichment, prioritisation, traceability), installed and operated for you.
Agility
Setup + controlled subscription
5,545 € excl. VAT setup
then 542 € excl. VAT/month — you invest at installation and pay a reduced subscription. Ideal for keeping the cost under control over time.
- Installation, configuration and training for your teams
- Operation, human oversight, updates and support
- Sovereign hosting in France, a dedicated and isolated resource
The simplest
Serenity
All inclusive, no setup fee
847 € excl. VAT /month
all inclusive, immediate start. No upfront investment: a single subscription. Ideal for starting quickly and simply.
- Setup included (installation, configuration, training)
- Operation, human oversight, updates and support
- Sovereign hosting in France, managed end to end
100% Sovereign
On site, you own it
8,935 € excl. VAT setup
then 746 € excl. VAT/month · + hardware from 2,491 € (one-off purchase, in addition) — a sovereign computer installed on your premises, maintained remotely. Models run locally, your data returned at the end of the contract. 36-month commitment.
- Hardware installed on your premises (you own it)
- French / European AI models run locally
- Secure remote maintenance (Pro support included)
Not included in the packages: AI consumption (model tokens), re-invoiced at real cost with no margin, and tracked in real time in your client area. Maintenance and supervision subscription for an initial term of 12 months for the Agility package, 24 months for the Serenity package and 36 months for the 100% Sovereign package, renewable; support levels (SLA 72 h / 24 h / 4 h) optional. Bespoke development, additional integrations or exceptional volumes are quoted separately. Support Monday to Friday, 9am to 6pm. Prices exclude VAT.
AI model:
none of the AI models offered currently carries a fixed surcharge. When the selected model carries a cost, that cost is shown when you choose it, before you order, and re-invoiced at the cost incurred, with no mark-up; usage is billed at the publisher's price. Publishers' prices are published in US dollars: the amount re-invoiced is the amount in euros actually borne by Blue Lemon Agent on the publisher's invoice, at that invoice's exchange rate, with no commission or mark-up.
Included components and additional components Components included in the base offer: the Blue Lemon Agent software foundation, the AI models listed in the order journey, the standard channels (Microsoft Teams, Slack, WhatsApp Business, email, website chat, calendars, Microsoft 365 / Google Workspace, file storage, market VoIP telephony, professional social-media pages and accounts, Google Business Profile), hosting in France for the package chosen, backups, supervision, updates and support. If adapting the AI agent to your constraints, your needs or your requests requires other paid components — a third-party publisher's software licence, paid API access to one of your applications, hosting of health data, for which French law requires an HDS-certified host (art. L. 1111-8 of the French Public Health Code), SecNumCloud-qualified hosting, a speech synthesis service, particular hardware —, they are offered to you as an option or on quotation and re-invoiced at the cost incurred; nothing is committed without your written agreement. Where the artificial intelligence model you choose entails an additional cost, that cost is shown to you before you order and re-invoiced to you at the cost incurred, with no margin.
What to expect
Go-live
2 to 3 weeks
Agent designed, channels connected, team trained.
Steady state
4 to 7 weeks
After a few weeks of real use, once the agent's behaviour matches what you expect.
Indicative estimate, adjusted to the options you keep. It is not a delivery commitment.
Our commitment
Four guarantees that matter to your portfolio
Your prospect data stays with youLocal inference or an isolated resource hosted in France; no record and no prospect data entrusted to a third party, no data used to train a model.
Data in France, under French lawYour prospect data and your portfolio: minimisation and location in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
The sales management keeps the decisionThe agent produces completed records and prioritised leads, which can be checked and altered; no approval is automated.
Human oversight & traceabilityOn your volume of leads and the richness of your CRM: systematic logging and tracking, in line with the AI Act.
Going further
Related resources
Frequently asked questions
Your questions, our answers
What is the prioritisation based on?
On the criteria you set: target sector, company size, channel of entry, how long the enquiry has been open. Every lead shows the detail of the calculation, criterion by criterion.
Can the criteria be adjusted?
Yes, at any time. An adjustment updates the order for every lead that follows, which lets you tune the prioritisation to your commercial reality.
Where does the enrichment information come from?
From your internal sources: the CRM, the history of exchanges, data already in your possession. The source of each piece of information added is indicated.
Does the agent assign leads to salespeople?
No. It prepares and puts them in order; assignment rests with sales management, which knows each salesperson's workload and affinities.
Is our prospect data protected?
Yes. The agent is hosted in France, on local inference or an isolated resource, with the deployment objective of processing and access operated within the European Union and an architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity. Your portfolio is not used to train a third-party model.
How long does it take to deploy this agent?
A few weeks as a rule, depending on the volume of leads and the integration with your CRM, after a free audit then a phase of design, integration and testing.
Related use cases
Other agents for sales
Let's talk
Let's size up the potential in your CRM
15 minutes to frame your criteria and your CRM — hosted in France, supervised, with no commitment.