We build Agents. You build the Future.
● Business offer — Small & very small businessesAI agent for small businesses: level 1 customer support, without a technical team
This page sells one thing, and one only: the level 1 AI customer support agent for a small business, at its own price — answering frequent questions, order and case tracking, resolving simple requests, escalation to your team, three channels in one service language, ticket or CRM updating. It is presented here for a small business that receives more requests than its few office staff can absorb. The review that picks and scopes the first use case is offered separately, on quotation; the other needs of a small business — phone, appointments, prospecting, accounting, tenders — each belong to another agent, listed below with its own page and its own price.
Updated on
Fictional simulation — not a client result
⛓ Source · your standard replies, your terms and your order tracking
Every exchange is logged, with its source and, where relevant, the reason for the handover.
✎ Action · request log kept — your team decides the sensitive cases
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 small business can deploy a Blue Lemon Agent customer support agent without a technical team: it answers frequent questions from your documents, gives the state of an order or a case, settles simple level 1 requests and hands over to your team for everything else, with the full history. Data hosted in France; the AI model is called at its publisher, except with the Sovereign model option. With that option, customer files, contracts and employee data are never exposed to a foreign service, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity. Your customers reach it on WhatsApp Business, your website chat or email, in the service language you set, with no account to create and nothing to install; it updates your ticket or your CRM. This scope is that of level 1 customer support, with nothing added and nothing removed; the price is this page's own, calculated on its own cost line.
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).
One price, this page's own
The scope of this page is that of level 1 customer support; its price is its own, calculated on its cost line: a set-up fee, then a monthly subscription depending on the plan, prices excl. VAT. The review and scoping of the first use case are quoted separately, once per client. The other agents listed on this page each carry their price on their own page.
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)
Level 1 customer support, for a small business
Six capabilities included, those of level 1 customer support, then a scoping option and seven neighbouring agents that can be added to it. The agent carries out by itself what you authorise it to do; out-of-scope cases are passed to your team, never decided by the agent.
Answering your customers' frequent questions
Opening hours, lead times, terms, warranty, documents to provide: the agent answers 24/7 from your standard replies and your documents, citing its source.
Order, quote & case tracking
Looks up the state of an order, a quote or a case in your management software and gives an exact, dated answer, in real time.
Level 1 resolution of everyday requests
Settles simple, documented requests: duplicate invoice, address change before dispatch, access reset, everyday steps.
Escalation to your team, with the history
Passes anything outside the frame — a complaint, a goodwill gesture, a quote request — to the right person on your team, with the full history of the exchange.
Multichannel, one service language
Works on WhatsApp Business, your website chat and email, in the service language you define; a further language is an option at checkout, billed per language opened.
Ticket & CRM updating
Records every request and its answer in your tracking tool or your CRM, and gives you a monthly summary: volumes, reasons, resolution rate, handovers.
Before starting: scoping the first use case
This service does not change the agent: it picks and bounds its first scope on your premises. It is offered separately, on quotation, once per client.
The list of requests that come back every day, the scoping note — where the data lives, which tools are connected, what the agent will not do — and the step-by-step deployment plan, submitted to the owner. Offered separately, once, per client.
On quotation — once, per clientThe other needs of a small business
Phone, appointments, prospecting, sales follow-up, accounting, tenders, document base: each belongs to a distinct agent, with its own page and its own price. They are ordered separately.
Switchboard
Picking up, qualifying and routing calls, including outside opening hours — so nothing is missed again.
Voice agent (AI switchboard) from 456 € excl. VAT / month Voice switchboard →Appointment booking
Qualifying incoming requests and automatically fitting appointments into the diary.
Lead qualification / appointment booking agent from 312 € excl. VAT / month Qualification & appointments →Outbound prospecting
Feeding the sales pipeline: identifying targets, making contact and booking meetings.
Outbound prospecting agent (automated email, assisted LinkedIn) from 332 € excl. VAT / month Prospecting →Sales follow-up & CRM
Keeping the CRM up to date, following up on opportunities and preparing follow-ups at the right moment.
Sales agent (meeting notes, follow-ups, CRM) from 307 € excl. VAT / month Sales & CRM →Administration & accounting
Automating the entry and checking of documents, alongside your chartered accountant.
Accounting agent (summaries, anomalies) from 406 € excl. VAT / month Accounting agent →Tender responses
Preparing and pre-drafting the responses from your references and your templates, for approval.
Tender response agent (drafting) from 354 € excl. VAT / month Tenders →AI knowledge base agent
The information exists in your company — but it is scattered across procedures, contracts, an intranet and the memory of a few people.
Document agent (FAQ, knowledge base) from 353 € excl. VAT / month Discover the agent →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.
The company in this demonstration
Fictional companyMaisonneuve Travaux — a second-fix building company in the Loiret, fictional company, twelve employees
- Sector
- Second-fix building work, all trades — renovation sites for local councils, social landlords and private customers
- Headcount
- 12 employees, 4 of them in the office — the owner, the executive assistant, a field sales rep and a site manager — and 8 on site
- Public served
- Councils and landlords following their sites, private customers waiting for a quote, a visit date or an invoice
- Order of magnitude
- 412 customer requests in three months, by email, WhatsApp and the website chat — 137 a month, 9 minutes each; 6 requests out of 10 concern the state of a site, a quote or an invoice
- Tools in place
- The shared contact@ mailbox, WhatsApp Business, the website chat, the shared “Sales” folder holding the terms and conditions, the lead-time grid and the standard replies, the management software and its order tracking, the CRM
- Who decides
- The owner rules on complaints and goodwill gestures; the executive assistant approves the standard replies and picks up handed-over requests; the field sales rep picks up quote requests; the site manager answers on visit dates
- Points for improvement
- 84 requests a month out of 137 can be settled from what is already written; 12 diverging answers in three months, given to the same customer by two people; 9 quote requests left more than two weeks without an answer in the shared mailbox; a dozen hours a month to give back to the office
Maisonneuve Travaux keeps twelve people busy on second-fix building sites. Four people in the office receive everything that comes in: questions from private customers, chasers from landlords, quote requests, invoices to resend. The same question — “where is my site up to?” — is asked ten times a week, to three different people, on three channels, and each answers as best they can. The level 1 customer support agent takes on the mass of these requests from what the company has already written; whatever commits the company goes up to the office, with the history.
This company, its figures and the exchanges that follow were invented for the demonstration. They illustrate a common situation; they describe no real client.
Fictional simulation — not a client result
· One question has recurred 23 times since Monday — “what date is the team coming to my place?” — and no standard reply covers it: everyone answers from memory.
· 9 quote requests have waited more than two weeks in the shared mailbox, 2 of them for six weeks.
· Customer B has written three times about the same invoice, by email then by WhatsApp, and received two answers that do not give the same date.
· The lead-time grid in the “Sales” folder dates from January: it announces a five-day quote lead time, your answers this week say ten. morning-watch_4-flags.pdf4 flags · 23 requests grouped
⛓ Source · shared mailbox, WhatsApp Business, website chat, “Sales” folder, order tracking
What I looked for: an answer in your standard replies, your terms and conditions and your order tracking. The visit date is nowhere in them: it is in the site manager's head and in his diary.
What I could have done: answer “within a fortnight”, which is plausible. A customer who plans their leave around a plausible and wrong date comes back angrier than one who was told we were checking.
What the 23 received: “I am passing your question to the site manager; you will have a date within 24 h”, and the question reached him grouped, as a single list of 23 sites, not as 23 messages.
What I propose: that the planned visit date be entered in the order tracking at each scheduling. The day it is there, I give it myself, and the 23 become zero.
⛓ Source · 23 requests since 21/09, standard replies, order tracking
What I did: each of the 9 is escalated to the field sales rep with the original request, the date received, the channel and what the customer has written since. A dated holding reply went to the customer, without promising a time frame nobody has set.
What I prepare for the sales rep: the file of each request — customer, site, what is asked for, documents received, lead time announced by the grid — so that he quotes from a complete file, in minutes. A quote commits the company: its signature stays with the sales rep.
With a chase: 24 h on a quote request without an answer, 4 h on a customer request waiting 48 h. Then a monthly summary in your CRM: by reason and by channel, never by person.
✎ Proposal · chase thresholds to be set — the owner decides
Fictional simulation — not a client result
Over the past month: 137 requests. I handled 84 end to end, each with the document cited — terms and conditions, lead-time grid, standard reply — and its date. 53 went to the office.
What I answer: opening hours, announced lead times, payment terms, the ten-year building guarantee and what it covers, the documents to provide for a landlord's file. I give the document, and the customer can ask for it.
What I hand over, and why the 53 split into three: 21 because the answer exists nowhere; 19 because the request commits the company — a quote, a discount, a complaint; 13 because the customer is unhappy, and unhappiness is handled by a person.
The third category deserves saying: I detect irritation from the wording, and I hand over before answering, not after the fact. handover_file-passed-on.pdf1 request passed on · 11 days' gap
⛓ Source · 137 requests this month, documents cited
The wording used: “Your request for a goodwill gesture is being passed to management with the history of your site; you will have an answer from them.”
Why I do not answer myself, even to refuse: granting a discount commits the company, and so does refusing one — a refusal written by an agent becomes a document the customer keeps. The owner decides, before a third party, under his signature.
What I pass on, and it is measurable: the request as it was worded, the actual site dates taken from the order tracking, the lead time announced in the quote, and the gap between the two — 11 working days. The owner does not start from nothing.
Across the 19 requests handed over this month because they commit the company: the office took on average 6 minutes to handle them, against 14 on those arriving without that file.
⛓ Source · 19 requests handed over, measured handling times
Fictional simulation — not a client result
What I read: order no. 2026-0418, stage “second fix — painting”, updated yesterday at 5:40 pm by the site manager; next stage “handover”, planned for 09/10.
What the landlord receives: those three facts, dated, and a note that the handover date is the one in the schedule and will be confirmed by the site manager.
If the tracking has not moved for 48 hours, I say so and give the time of the last reading, rather than repeating a stale state.
For the month: 51 requests on the state of a site, a quote or an invoice, 47 handled end to end; the remaining 4 passed to the site manager because the tracking contradicted what the customer had in front of them.
⛓ Source · order tracking no. 2026-0418, schedule, 51 status requests this month
This month, 38 requests of that kind, 34 resolved: a duplicate invoice sent from the management software, a ten-year guarantee certificate attached, a billing address corrected before issue, a handover appointment confirmed on the slot already scheduled.
Where level 1 stops, and the line is written: as soon as a request touches a refund, a discount, a deadline to renegotiate, a complaint or a quote, it goes to the office with the full history. Level 1 resolution is judged on what it passes on as much as on what it resolves.
The figure I publish about myself: of the 34 resolutions, 2 produced a second request from the same customer within seven days. Both concerned the guarantee certificate: the template in the “Sales” folder dated from last year. I flagged it to the executive assistant, and meanwhile I answer with the up-to-date template she pointed me to.
⛓ Source · 38 level 1 requests, 34 resolved, 2 repeats
Fictional simulation — not a client result
What multichannel means concretely: email, WhatsApp Business and the website chat share the same standard replies, the same order tracking and the same handover. What changes between channels is the form, not the substance: on WhatsApp the answer runs to a few lines; by email it carries the case reference and the document cited.
Why one base and not three: customer B wrote three times about the same invoice, on two channels, and received two different dates from two people. With one base, the third request arrives with the first two in view.
The service language, and I would rather be plain: it is single, and you set it — French here. The landlord writes in English: I answer in French, saying so, and I hand over if they cannot read it. I open another language the day you order it, and it is billed per language opened. What I guarantee: no answer in a language nobody proofreads on your side. monthly-summary_137-requests.pdf137 requests · 5 indicators
⛓ Source · 3 channels on one base, 12 diverging answers in three months
Ticket and CRM updating: each request creates or completes a ticket — the channel, the question, the answer and the document cited, or the handover and its reason, then what the office answered. The ticket closes when the answer has gone, never before, and never without an answer — and every handover is logged.
What I give you back every month: volumes by channel, the most frequent reasons, questions with no written answer, the level 1 resolution rate and the handover reasons. Five indicators about your requests, none about people.
What is not kept: no customer profile built from their requests, no statistics per office member, and no analysis of a customer's tone beyond the current request. I detect irritation in order to hand over immediately — and that detection does not outlive the conversation.
✎ Framework · retention periods to be set by the company
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 AI matters to small businesses — and why they hesitate
A small business has few hands for many repetitive tasks that come round every day — and everyday customer requests are the first of them. AI is an opportunity to win that time back, provided you keep control of the sensitive cases and of your data.
! The issue
The small business is caught between a load of repetitive tasks piling up on the same shoulders and the impossibility of hiring to absorb it all. Yet most consumer AI solutions amount to entrusting customer files, contracts and employee data to a third party, often hosted outside Europe and subject to the Cloud Act.
✓ Our answer
AI is only of interest to a small business if it is simple to start and hosted in France. You begin with the most frequent case — everyday customer requests —, measure, then extend with other agents. Local inference or an isolated resource hosted in France, systematic human oversight, decisions reserved to the owner: the time saved is never paid for in lost confidentiality. The aim is not to replace the team, but to give it back time.
A small business has to protect its data too
It is sometimes thought that sovereignty only concerns large groups. That is wrong: a small business handles customer files, contracts, employee data. Here is how the architecture of our agents protects them.
Local inference
With the 100% Sovereign package, the agent runs on a machine belonging to the company: no file leaves the network, nothing passes through a 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 customer files to the Cloud Act and FISA 702 is reduced by design; location alone does not guarantee immunity.
One isolated resource per company
No pooling of customer files: an environment strictly dedicated to your business.
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; no sensitive action approved 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.
- 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 can a small business win back?
Everyday customer requests are the most repetitive task of a small office. The agent takes on the mass of them; your team keeps the cases that call for a decision. The scale of the gain depends on your volume and is confirmed by a pilot.
Four guarantees that matter to a small business
Your questions, our answers
Does a small business really need an AI customer support agent?
Do you need an IT person to run an AI agent?
How much does an AI customer support agent cost for a small business?
What happens when the agent does not know the answer?
Is my customer data protected?
Does the agent state that it is an artificial intelligence?
How long does it take to deploy a first agent?
Which tools can people use to talk to the agent?
Does the agent also take calls?
48 AI agents Blue Lemon Agent deploys for this scope
Do you work in a specific field?
Let us estimate the volume of requests in your business
30 minutes to measure what the customer support agent would take on — hosted in France, supervised, with no commitment.