+33 (0)1 87 66 00 65 · Monday to Friday, 9am–6pm Free audit (15 min)
● B2B offer — Personal executive assistant

The AI executive assistant: your digital chief of staff

A managing director opens the inbox before everyone else and closes it after everyone else. In between, what matters is buried under what does not. Your agent reads overnight, sorts, matches deadlines to their clauses and hands you, at a fixed time, a briefing that fits on one screen: what needs a decision today, what can wait, what is already written. It reaches you where you already are — Teams, WhatsApp or your inbox. Hosted in France, on local inference or an isolated resource. You approve every message sent.

Hosted in France Inbox and calendar protected GDPR & AI Act: governed deployment You approve every message sent

Updated on

Live in two to four weeks
Executive assistant · hosted in France
7:15 — your briefing. 128 messages since 6 p.m. yesterday. 96 filed with no action needed (copies, suppliers, newsletters), 32 concern you.
4 decisions are waiting today, one of them before noon: the Nancy site needs a delivery date your ERP already provides.
3 deadlines fall this week, each with the contract clause behind it.
Five replies are written and waiting for your word.
⛓ Sourced · executive inbox, calendar, ERP, live contracts
The decision before noon — what exactly is it?
Confirm or move the 14 October installation on the Nancy joinery lot. The site manager asked you yesterday at 5:42 p.m.; your ERP shows the workshop free from the 9th, and the contract sets a penalty beyond the 20th.
The reply is written, four lines, with the date and the lot reference. It goes as soon as you say so — or right now, if you give me the mandate to confirm the lead times your ERP already covers.
✎ Framework · nothing goes without your word, or a written mandate you withdraw with a word
Local inference · no data outside the EU
Data hosted in France
Sovereign by designLocal inference or hosting in France
GDPR & AI Act: governed deploymentTraceability & human supervision
TurnkeyDesigned, installed and run for you
You approve every message sentThe agent writes, you sign
✦ In brief

A Blue Lemon Agent executive assistant reads your inbox and calendar before you do, sorts what calls for a decision, matches every deadline to the clause behind it and hands you a daily briefing at a fixed time on Teams, WhatsApp or your inbox. It writes the routine replies, prepares the files for the day's meetings and keeps the list of what is pending. Hosted in France, on local inference or an isolated resource, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity. You approve every message sent — and the day you want it to send one precise category on its own, that is written into a named, capped, dated mandate you withdraw with a word. Live in two to four weeks.

100 %
hosted in France in the target architecture
0
transfer outside the EU in the target architecture
3
briefing channels to choose from: Teams, WhatsApp, email
0
message sent without your word or your mandate

Benchmarks describing our offer, not results measured at a client. How much you gain on your own day is confirmed by a pilot on your own inbox.

The context

Why the director's inbox is the bottleneck of the company

A pending decision from the director usually blocks several people downstream. Yet those decisions arrive mixed with hundreds of messages that call for none — and nothing in an inbox tells them apart.

! The stake

The head of an SME is the crossing point of a flow that no tool ranks: client messages, internal chasers, contractual deadlines, meeting invitations, copies for information. They read early, reread late, and the single most expensive item — the one that called for a dated decision — is the easiest to lose. Handing that inbox to a consumer tool would expose executive correspondence, commercial terms and live negotiations to a third party, often hosted outside Europe and subject to the Cloud Act.

Our answer

An executive assistant is only worth having if it is sovereign and confidential by design. Local inference or an isolated resource hosted in France, an overnight reading of inbox and calendar, deadlines matched to contracts and to the ERP, a briefing at a fixed time on the channel you choose, routine replies already written, files for the day's meetings prepared. You approve every message sent: the agent prepares and quantifies, you decide. And when you want it to send one precise category on its own, that is not guessed — it is written into a named, capped, dated, logged mandate you withdraw with a word.

The decisive point

An executive inbox is the most sensitive in the company: sovereignty & compliance

Commercial terms, live negotiations, employment matters, correspondence with counsel: the director's inbox concentrates the most confidential material the company holds. Here is how our architecture protects it.

Local inference

The agent can run on a machine inside the company: no message, no attachment, no calendar entry leaves the network.

Hosting in France

Otherwise, a dedicated, isolated resource hosted in France under French law — your executive correspondence: processing and access within the European Union targeted by the architecture.

Reduced extraterritorial exposure

Architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity: the analysis of your inbox depends on a subcontracting chain and remote access documented for the configuration chosen.

One inbox, yours

The agent serves the person who employs it, on their own inbox and their own calendar. It produces no indicator about colleagues from what it reads there.

Short briefing outside EU messengers

On WhatsApp the message travels through an operator outside the European Union: the briefing goes there in short form, with no client name, no amount and no file reference. The full version stays in Teams or in your inbox.

AI Act: governed deployment

Strictly a support agent; no message sent without your word or a written mandate; a full decision log and human supervision 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 events logged, their content, their retention period and who may access them are defined for the deployment chosen.
For the most sensitive correspondence — employment matters, capital operations, exchanges with counsel — SecNumCloud and strictly local inference are available 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.

The company in this demonstration

Fictional company

Sorel Menuiseries — maker of bespoke aluminium windows and doors, private and public contracts

Sector
Manufacture and installation of aluminium joinery, as finishing-trade lots
Headcount
62 staff, including 1 managing director and 1 part-time executive assistant shared with accounts
Market served
Builders, architects and public buyers in the north-east of France
Order of magnitude
128 messages a day in the director's inbox, 11 meetings a week, 34 live sites
Tools in place
Microsoft 365 mail and calendar, Teams, a production ERP, a CRM, site contracts in the document store
Who decides
The managing director approves every message sent; the executive assistant keeps the shared calendar
Room for improvement
1 h 50 a day spent in the inbox, and 3 contractual deadlines missed over the last half-year

Sorel Menuiseries has doubled in size in six years without changing how it decides: everything comes up to the managing director, and the inbox has become the company's compulsory crossing point. The executive assistant keeps the calendar but has no access to site files, and nobody rereads the contracts before the deadline. The agent is connected to the mailbox, the calendar, the production ERP and the contract store; it reads overnight and delivers a briefing in the morning. Nothing goes without the director's word, except what a written mandate entrusts to it.

This company, its figures and the exchanges that follow were invented for the demonstration. They illustrate a common situation; they describe no real client.

Executive assistant · hosted in France
I reread your last 90 days of inbox and calendar. Your three missed deadlines this half-year were not an oversight: all three come from the same place in your day.
What I counted: 11,520 messages received — 128 a day — of which 806 called for a reply from you, that is 7 %. Your read timestamps give 1 h 50 a day spent in the inbox.
And here is the gap that explains everything. Your 806 replies fall into two very unequal populations:
· 141 went out in under 2 hours. Every one is about a live site, and every one comes from a correspondent you know.
· 665 waited more than 24 hours, of which 203 more than four days.
All three of your missed deadlines are in the second group. Each was announced in a message that stayed 4, 6 and 11 days unopened — and in all three cases the subject line carried no word of urgency. It was not your attention that failed, it was the sorting: nothing in an inbox tells a routine chaser from a contractual formal notice.
My proposal: a briefing at 7:15, fourteen lines, separating what calls for a dated decision from what calls for none — and matching every announced date to the contract clause behind it. Approve it for a week, we measure it, and you change it as much as you like. diagnostic_90-days-of-inbox.pdf141 replies within 2 h against 665 beyond 24 h
⛓ Sourced · executive inbox (90 days), calendar, read timestamps, contracts in the document store
How do you decide that one message calls for a decision and another does not? That is where it plays out.
On three questions asked of every message, and never on the tone of the message — the tone is precisely what cost you those three deadlines.
The three questions:
· Is someone waiting on you to move forward? I see it from an explicit request and from the absence of a third-party reply in the thread.
· Does a date commit the company? I then match the message to the site contract in the document store and display the article. A date without its clause cannot be ranked.
· Does the information already exist elsewhere? If the ERP gives the workshop availability being asked for, the reply is written before you open the message.
Here is this morning's briefing, exactly as it went out: 128 messages since 6 p.m. yesterday, 96 filed with no action needed, 32 concern you, 4 decisions today including one before noon, 3 deadlines this week, 5 replies written.
And the figure I owe you: replayed over the 90 days, my sorting would have been wrong 41 times — 29 messages filed as « no action » that you had in fact handled, 12 raised as decisions that were not. That is 0.36 % of messages, and I would rather tell you before you find it. All 29 misses come from new correspondents: I propose raising every first message from an unknown correspondent for a month, then measuring again. briefing_of-the-day_07-15.pdf14 lines, 4 decisions, 3 deadlines
⛓ Sourced · sorting rules replayed over 90 days, site contracts, production ERP
Why 7:15? And I want it on WhatsApp, that is what I open first.
7:15 because it is your hour, measured — and on WhatsApp yes, with one nuance I would rather set out now than at the first incident.
The hour: over 90 days, your first action of the day is at 7:22 in median, and it happens on your phone 6 days out of 7. A briefing at 7:15 arrives just before, not during.
You asked me last week to move it to 6:30: I advise against it, and here is the arithmetic. 4 of the 14 lines in yesterday's briefing came from messages that arrived between 6:30 and 7:10 — two site managers write before leaving for the site. At 6:30 you would have read a briefing already a quarter wrong.
The channel, and what travels on it: all three work, but they are not equivalent. On WhatsApp the message travels through an operator located outside the European Union. On that channel I therefore propose the short briefing: « 4 decisions, 1 before noon, 3 deadlines, 5 replies ready »no client name, no amount, no site reference. One word from you and the full version opens in Teams or in your inbox, where it has never left your perimeter.
What you gain concretely: you see in three seconds whether to open it, and the information that would be worth something to a competitor does not travel on a consumer network. Convenience and confidentiality are not traded off here, they add up. briefing_three-channels-compared.pdfFull in Teams, short on WhatsApp
⛓ Sourced · first-consultation timestamps (90 days), message arrival times, data egress policy
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 the executive assistant actually does

Each use corresponds to a capability we configure with you. All of them work in support: you approve every message sent.

Included in your agent The 7 capabilities essential to this promise are included, at no extra cost.
From 652 € excl. VAT / month

Daily briefing at a fixed time

One screen, not a list: what needs a decision today, what can wait, what is already written.

On Teams, WhatsApp or in your inbox

The briefing arrives where you already are in the morning, in full or short form depending on the channel.

Drafted replies, ranked

Routine replies are written, sourced on your own documents and ranked by what they unblock.

The day's meetings prepared

One file per meeting: points left open, up-to-date figures, what has changed since the last session.

Deadlines matched to their clause

Every date carries its contract, its article and its consequence — a deadline without its clause cannot be ranked.

Named, withdrawable sending mandates

The day you want it to send one precise category on its own, the scope, the cap and the duration are written — and withdrawn with a word.

The company's memory

Instantly retrieve a contract, a term or the history of a file quoted in the briefing.

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 Sources, access rights and handling of questions with no answer Data protection, rights and human oversight
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.
Integration with existing tools without double entry
The gain

How much of a director's day?

By taking on the reading, the sorting, the matching of deadlines and the writing of routine replies, the agent shifts the work towards the decision. Actual gains depend on your message volume and the number of your meetings.

Reading and sorting the executive inbox
Today · done by hand
Briefing read
Writing routine replies
Today · done by hand
Review and approval
Preparing the day's meetings
Today · often skipped
File ready at the door
Tracking contractual deadlines
Today · from memory
Every date with its clause
A qualitative, non-contractual comparison: the proportions shown illustrate how the work shifts towards the decision, they represent no measurement. Every output of the agent is reviewed and approved by a competent person.
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

An executive assistant, installed and run for you

An agent connected to your mailbox and calendar, with the briefing channel of your choice, tuned to your files and your deadlines. Prices excluding VAT — annual subscription, the time it takes for the gains to settle.

Agility

Setup + controlled subscription

8,005 € excl. VAT setup
then 652 € 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
Order →
The simplest Serenity

All inclusive, no setup fee

1,097 € 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
Order →
100% Sovereign

On site, you own it

11,745 € excl. VAT setup
then 856 € 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)
Order →
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 for an executive inbox

Your correspondence never leavesLocal inference or an isolated resource hosted in France; no message or attachment entrusted to a foreign third party, no content used to train a model.
Data in France, under French lawInbox, calendar and attachments: minimisation and location in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
You approve every message sentThe agent writes, quantifies and prepares; no message goes without your word, or without a written mandate, capped, dated and withdrawn with a word.
Human supervision & traceabilityEvery briefing, every draft and every message sent under mandate is logged, as the AI Act requires.
Frequently asked questions

Your questions, our answers

How does it differ from the email assistant?
The email assistant handles a shared service address — contact@, sales@, support@ — and produces drafts for the team that runs it. The executive assistant serves one person, on their own inbox and their own calendar: it does not merely sort, it matches messages to your contracts, your ERP and your meetings to produce a decision briefing. The two complement each other.
Which channel does the briefing arrive on?
The one you already use in the morning: Teams, WhatsApp, or simply your inbox. One important point: on WhatsApp the message travels through an operator located outside the European Union. The briefing therefore goes there in short form — number of decisions, urgent items, cut-off times — with no client name, no amount and no file reference. The full version stays in Teams or in your inbox, where it does not leave your perimeter. Telegram is unavailable for this processing — qualification under way: the Telegram API terms prohibit using, accessing or aggregating data obtained from the platform for the development, enhancement or deployment of artificial intelligence.
Does the agent send messages in my name?
You approve every message sent. That is the default: the agent writes, you review, you send. This is not a technical limit but a question of mandate: sending in your name commits you. The day you want it to send one precise category on its own — acknowledgements, forwarding an already approved document, confirming a free slot — that is written down: named scope, daily cap, duration, a log you can read, and withdrawal with a word. The approval stays yours; you give it once in writing instead of giving it again message by message.
Does the agent measure my colleagues?
No, and that is not what it looks at. It serves one person on their own inbox: what it reads there serves to prepare your decisions, not to describe those who write to you. No individual indicator — response time per person, message volume per department — is produced from your mailbox. Should your company one day wish to measure an activity, that is a separate project, with the prior information of the people concerned and the consultation of the works council required by French employment law.
What if the agent gets a priority wrong?
You correct it in the briefing, and the correction becomes a dated rule you can read back and undo. Every month the agent publishes the share of its proposals you rewrote or downgraded: that is the indicator that says whether it is improving, and it is measured on your corrections, not on its output.
Is our 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, in compliance with the GDPR. Your correspondence is not used to train a third-party model.
How long does deployment take?
Two to four weeks depending on the connections (mailbox, calendar, briefing channel, ERP or CRM if you want deadlines matched to your files), after a free audit that scopes the perimeter, then a design, integration and testing phase before go-live.
Let's talk

Let us size the potential on your day

15 minutes to scope your message volume, your deadlines and your briefing channel — hosted in France, supervised, no commitment.