● B2B offer — Catalogue & e-commerce

Product page agent: your specifications, your template

A growing catalogue creates a writing backlog: items online with no description, or described in haste. Your agent puts the page together from the product's real specifications, in your template and with the basics of search optimisation — without adding a selling point you have not supplied. Hosted in France — local inference or an isolated resource — your forthcoming launches stay confidential. The catalogue manager approves before publication.

Free audit (15 min) → See the prices

Hosted in France Catalogue never exposed GDPR & AI Act: governed deployment Human oversight

Updated on 07/09/2026

Deployed in a few weeks

Product pages · hosted in France

Write the page for the new item from the specification table.

Page put together from the specifications in the table alone: description, key points, technical specifications.
Template applied, title and meta description filled in according to your search optimisation rules.
Three fields are empty in the source: I flag them rather than filling them in.
✎ Action · page to read over, nothing is published

Can you fill in the missing fields?

These two fields are flagged rather than filled: they concern composition and warranty, where an inaccurate statement would engage your liability as the seller.
The rest of the page is complete and ready to publish; all that is left is for you to fill in those two points.
✎ Action · the agent protects your product claims

Local inference · no data outside the EU

Catalogue 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 catalogue manager decidesThe agent prepares, never rules

✦ In brief

A Blue Lemon Agent product page generation agent writes your descriptions from the real specifications of each item, in your template, with the basics of search optimisation. It adds no specification absent from the source and flags the missing fields instead of filling them in. It runs on local inference or is hosted in France: your catalogue and your forthcoming launches stay with you, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity. Live within a few weeks,.

100%

hosted in France in the target architecture

transfer outside the EU in the target architecture

catalogue uses ready to deploy

decision taken without human approval

These figures describe our offer, not results measured at a client: how large the gain is on the size of your catalogue is confirmed by a pilot.

The context

What does an AI agent bring to enriching your catalogue?

An item with no description sells badly and ranks badly. But a page that invents a specification engages the seller's liability.

! The issue

Every new item needs a description, key points and search tags — repetitive work that piles up and delays going live. Yet a general-purpose automatic writer invents plausible specifications, which engages your liability as the seller, and along the way absorbs your catalogue and your unlaunched items into a third party subject to the Cloud Act.

✓ Our answer

Automated page writing is only of interest if it is sovereign and strictly bounded by the source. Local inference or an isolated resource hosted in France, writing from the supplied specifications alone, empty fields flagged: you catch up the backlog without risking an inaccurate claim. A specification that is absent stays absent.

The decisive point

Your catalogue and your unlaunched products: sovereignty & compliance

Your catalogue and your forthcoming launches reveal your product strategy. Here is how the architecture of our agents protects them.

Local inference

The agent can run on a machine belonging to your organisation: no item leaves the network, no forthcoming launch passes through a public cloud.

Hosting in France

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

Reduced extraterritorial exposure

For your catalogue and your unlaunched products, 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 whatsoever: an environment strictly dedicated to your company and its catalogue.

Template and search rules respected

Your page template and your tagging rules are applied as they stand; encryption, role-based access and logging of every page produced.

AI Act: governed deployment

The agent is strictly in support; no specification is invented and no page is published 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 catalogues containing regulated products or launches under embargo, 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.

Product pages · hosted in France

Four flags this morning on your catalogue.
· One hundred and eighty-seven pages show a characteristic that appears in no supplier data. It comes from a text, not a source.
· Forty-one pages carry a numerical claim with no origin — "up to 30%", "twice as much".
· Three hundred and forty pages share the same paragraph. It describes no product in particular.
· Twenty-nine references changed at the supplier and their page did not move. morning-watch_4-flags.pdf187 unsourced characteristics · 41 figures with no origin
⛓ Source · 2,100 pages, supplier data, reference history

One hundred and eighty-seven invented characteristics?

Not invented — copied from an older text nobody knows the origin of.
What I record: those 187 pages show a value — a strength, a compatibility, a capacity, a duration — that appears in no supplier datasheet, no catalogue, no document you opened to me.
Where they come from: an earlier draft. Tracing back, 143 are found word for word in an older page in the same catalogue, itself unsourced.
Why it is not a detail: a characteristic shown on a page is what a customer decides on. A false compatibility produces a return, a false strength produces a complaint, and in both cases it is written in black and white on your site.
The rule I apply: no characteristic that does not come from a dated supplier source. Where the data is missing, the field stays empty — it is neither filled in nor guessed from a neighbouring product.
What that gives: of 2,100 pages rewritten, 412 fields stayed empty. Each carries a note of what is missing and who to ask. 187-characteristics_412-empty-fields.pdfA characteristic shown is what a customer decides on
⛓ Source · 187 unsourced characteristics, 143 copied from an earlier page

How do you work, and who do you tell?

I write from supplier data, and I report everything I could not write.
What each page produced carries: for every characteristic, the source and its date, the fields left empty with what is missing, and the version of the supplier reference used.
Routing follows who can act: missing data goes to whoever manages supplier relations, with the list of references concerned; a numerical claim with no origin to whoever wrote it, never removed silently; a reference modified at the supplier to the catalogue owner; a paragraph shared by hundreds of pages is flagged as text that describes nothing.
With a monthly chase. Then a monthly summary: empty fields by supplier, pages with at least one unsourced characteristic, and the average age of the data used.

Three guarantees. Every claim on a page is tied to a dated source — that is what makes the page defensible before a customer as before an inspection, and what lets you find in one second where a figure written two years ago came from. Price and availability come from your systems: they change faster than pages, and a page freezing them would be lying by the next day. And the lawfulness of a claim is validated at your end, on a file that is ready: I hand over the list of claims with their source or its absence, sorted by risk — the person responsible rules on forty lines, not on a whole catalogue.
✎ Framework · no unsourced claim · no legal validation

"Up to 30% saving" — what is wrong with that?

The form, not the figure: "up to" is true whatever the actual result.
What I record: 41 pages carry a numerical claim whose origin I find nowhere — no test, no supplier data, no documented comparison. The commonest wordings: "up to X%", "twice as much", "best in its class".
What happens to them, and it is not a silent removal: each goes back to its author, by name. A claim silently withdrawn suggests it was false, when it may be accurate and merely undocumented — and the only way to know is to ask whoever wrote it.
What I do: I list them, with the page, the exact wording, and the question: what does this figure rest on?
What came out of it: of the 41, 12 rested on a real test nobody had tied to the page — citing the source was enough. 9 came from a supplier document once found. 20 had no identifiable origin and were rewritten by their author.
What I write in their place: the measured value, the measurement condition, and the source. Less spectacular, and checkable by the customer. 41-claims_12-9-20.pdf"Up to" is true whatever the actual result
⛓ Source · 41 numerical claims, 12 tests found, 20 rewritten

Do those wordings also sit in the tags?

Yes, and that is where they show most: the tag is what a search engine displays before anyone opens the page.
What I record: of the 41 claims, 26 also sit in a title or a meta description, copied word for word. And 806 meta descriptions are empty: the engine then makes one up itself, out of whatever paragraph comes first — often the one 340 pages share.
Why keyword stuffing is not merely useless: it takes the room that belongs to the reference, the dimension and the compatibility — the three things a buyer actually types. 212 of your titles repeat the brand three times or more, and not one of those 212 carries the capacity.
What I write instead: a title built on the reference, the distinguishing characteristic and the capacity, in the order and the length you set; a meta description that restates what the page actually says, with no promise absent from it.
The figure that does not flatter me: across 2,100 pages, I could only write the meta description of 1,688. The remaining 412 are exactly those with a field left empty: summarising an incomplete page would amount to filling that hole in the tag, which the rule forbids everywhere else. What becomes of them: they go out with the data requests to the supplier, and the tag is written the day the field is filled — with nobody having to come back over it. tags_1688-metas-written.pdf806 empty · 26 claims copied across · 412 waiting on a field
⛓ Source · 41 claims of which 26 in a tag, 806 empty metas, 1,688 written

The same paragraph on 340 pages — is that bad for search?

Yes, and it is the lesser of the two problems.
What I record: an identical introductory paragraph, word for word, on 340 pages from different families. It speaks of quality, standards and satisfaction.
The search problem: real — a search engine seeing 340 pages carry the same content keeps only one, and you do not choose which.
The real problem: that paragraph fits 340 different products because it says nothing about any of them. It is a simple test, and I apply it everywhere: if a sentence stays true when you change the product, it does not belong on a product page.
What I do: I replace that block with what actually distinguishes the product — dimensions, material, compatibilities, box contents — taken from supplier data. Of the 340, I could write 288 distinct pages.
The other 52: the available data does not distinguish them. I flag that rather than manufacture an editorial difference — two genuinely identical products are not told apart by style. 340-pages_288-distinct.pdfA sentence still true with another product does not belong on the page
⛓ Source · 340 pages, 288 rewritten distinct, 52 flagged

And the 29 references that changed?

The supplier modified the reference, the page stayed as it was — in one case for three years.
What I record: 29 references whose supplier datasheet was updated while your page did not move. In 11 cases, a characteristic shown on your site no longer matches.
What that produces: a perfectly written, coherent page that describes a product no longer sold in that form. Nothing signals it: it is neither broken, nor empty, nor suspect.
What I do: I compare the version of the supplier data used at writing with the current version, across the whole catalogue. That is what recording the version on each page makes possible — without it, this comparison cannot be made.
What I have prepared for the 11 that no longer match: the rewritten page is already there, beside the current one, with the changed characteristic highlighted and the supplier version that carries it. Publishing stays with the catalogue owner: a supplier change can be a typo fix or a product change, and the two look alike in a file — never in a live catalogue. Eleven side-by-side comparisons, and the catalogue is current by the afternoon.
What I report: the reference, what changed, how old your version is, and whether the modified characteristic is shown on your site. Of the 29, 11 need rewriting, 18 do not. 29-references_11-to-rewrite.pdfAn outdated page is neither broken, empty nor suspect
⛓ Source · 29 modified references, 11 shown characteristics outdated

Can you publish the product pages straight online?

Yes, and that is where you recover the most time. The publishing condition is written, not left to my judgement.
What I publish alone: a page where every characteristic carries a dated supplier source, with no unsourced numeric claim, no paragraph shared with another page, and whose supplier reference has not changed since the last update. Those four conditions are mechanically checkable — I judge nothing, I observe.
What it covers on your catalogue: once the 187 unsourced characteristics are handled and the 340 paragraphs rewritten, most of the catalogue publishes with no intervention. Today every page waits for a read-through that takes four days on average.
What still waits for you: the 41 numeric claims whose origin I cannot find, prices — which are not a characteristic but a commercial commitment — and the 29 references whose supplier sheet has changed: there, somebody must decide whether the product being sold is still the same one.
What I do as soon as a source moves, even after publishing: I unpublish the characteristic concerned and tell you. A live page asserting something out of date is an assertion made to the public — an empty field beats a false one. direct-publishing_4-conditions.pdfWhat publishes alone · what waits · what gets unpublished
⛓ Source · 4 checkable conditions, 4-day read-through removed on most of the catalogue

What do you keep?

The full path from every assertion back to its source — and that is what protects you the day it is contested.
What is kept: for each characteristic, its source and that source's date, the supplier reference version at the time of publishing, the text published as it went out, the characteristics I could not write and why, and the unpublishings with their reason.
What it is for, concretely: a complaint about a capacity, a compatibility or a rating is settled by opening the supplier's technical sheet as at the publication date. Without that trace, the discussion is about today's sheet — which is not the one the customer read.
What it has already shown: the 187 unsourced characteristics were not invented. They were copied from an older text nobody could trace — and that mechanism repeats at every catalogue overhaul until the source is attached to the value.
What I do not keep: nothing about people. Not who read through, nor how many pages a day — except the name of whoever approved a numeric claim, because a claim binds the company and somebody must answer for it. what-you-keep_product-pages.pdf5 items kept · a complaint settled on the page as it stood
⛓ Source · source and date per characteristic, supplier version timestamped

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 life of the catalogue. 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 489 € excl. VAT / month

Writing the page

Puts together the description and key points from the supplied specifications alone.

Search tags

Fills in the title and meta description according to your rules, with no keyword stuffing.

Flagging what is missing

States which fields are empty in the source instead of filling them in by inference.

Controls and safeguards These 3 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 Human validation before publication, with versions retained

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.

Link content to performance, costs and campaign objectives

Other needs our agents cover Each card says where the matching agent stands: available, on quote, or still being architected.

Multilingual translation

Rendering approved product pages in another language is not included here: it is the service of a dedicated agent, which sets out the languages, the formats and the human validation.

On quote 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.

Marketing content

For editorial content around the product, a dedicated writing agent takes over.

Writing & marketing content assistant from 444 € excl. VAT / month Copywriting →

Does your need fall outside this?

In 15 minutes we identify the most relevant agent — without oversizing the project.

Book the free audit → Build your agent

The gain

How much backlog can a catalogue catch up?

By automating the first draft and the tagging, items go live faster, with reading over becoming the bulk of the work. How large the gain is depends on your volume and remains to be confirmed by a pilot.

Writing a page

Today · written from scratch

Review and approval

Search tags

Today · keyed in by hand

Prepared by the agent, to approve

Spotting missing information

Today · found after publication

Prepared by the agent, to approve

Qualitative, non-contractual comparison: the proportions shown illustrate the shift of the work towards review, 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

Audit & scoping

15 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.

Pricing

One package, one agent

A product page generation agent (templates, basic search optimisation), installed and operated for you.

Agility

Setup + controlled subscription

4,630 € ^{excl. VAT} setup

then 489 € ^{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

749 € ^{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

7,965 € ^{excl. VAT} setup

then 689 € ^{excl. VAT}/month · + hardware from 1,058 € (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.

See the full price list →

Estimate my gains in 1 minute →

Our commitment

Four guarantees that matter to your catalogue

Your catalogue does not leave the companyLocal inference or an isolated resource hosted in France; no item and no forthcoming launch entrusted to a third party, no data used to train a model.

Data in France, under French lawYour catalogue and your unlaunched products: minimisation and location in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.

The catalogue manager keeps the decisionThe agent produces written product pages, which can be checked and altered; no approval is automated.

Human oversight & traceabilityOn the size of your catalogue: systematic logging and tracking, in line with the AI Act.

Going further

Related resources

AI agents for business →Our full offer Sovereign AI hosted in France →Our confidentiality commitment

Frequently asked questions

Your questions, our answers

Can the agent invent a product specification?

It is designed not to: it writes from the supplied specifications alone and flags the empty fields. An inaccurate claim engages your liability as the seller, so reading over remains essential.

Does the agent publish the pages itself?

No. It produces a page ready to read over; publication remains a human act, in your catalogue management tool.

Is search optimisation taken into account?

Yes, on the basics: title, meta description and structure, according to your rules. This is not a search optimisation service, but clean tagging from the moment the page is created.

Is our catalogue 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 items are not used to train a third-party model.

Does it connect to our catalogue management tool?

Yes, to the common product management tools and e-commerce platforms, with no migration. The integration is defined at the design stage.

How long does it take to deploy this agent?

A few weeks as a rule, depending on the structure of your catalogue and the richness of your template, after a free audit then a phase of design, integration and testing.

Related use cases

Other agents for the catalogue and online sales

E-commerce →Product advice and support Translation →Rendering the catalogue Marketing copywriting →Content around the product

Let's talk

Let's size up the potential in your catalogue

15 minutes to assess your volume of items — hosted in France, supervised, with no commitment.

Free audit (15 min) → Write to us