Agent for generating simple quotations: catalogue and PDF template
Taking the prices from the catalogue, working out the quantities, laying out a clean quotation on the company template: a simple task, repeated several times a week. Your agent puts the quotation together from your catalogue and your PDF template, ready to read over before it goes out. Hosted in France — on local inference or an isolated resource — your prices and your client data stay with you. The salesperson approves before anything is sent.
Updated on
Layout applied on your PDF template, legal wording and validity period included.
Ready to read over — nothing has gone to the client.
✎ Action · quotation ready in draft — you approve before it is sent
New version ready to read over before it goes out.
✎ Action · quotation recalculated — you approve before it is sent
A Blue Lemon Agent quotation generation agent puts a quotation together from the client's request, your price catalogue and your PDF template — part numbers, quantities, any discounts and the legal wording. The document is laid out to your brand guidelines, ready to read over before it goes out. It runs on local inference or is hosted in France: your prices and your client data are never exposed to a foreign service, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity. The salesperson moves from formatting to reading over. Live in two to three weeks,.
These figures describe our offer, not results measured at a client: how large the gain is on your volume of quotations is confirmed by a pilot.
What does an AI agent bring to preparing your quotations?
Putting together a correct quotation means taking the right prices, working out the quantities and respecting the company template. But entrusting your catalogue and your prices to a consumer tool means losing control of them.
! The issue
Every quotation means taking the catalogue in force, checking the quantities and respecting the expected layout — a simple task, but one that piles up as client requests come in. Yet most consumer AI tools amount to entrusting prices, discount structures and your clients' contact details to a third party, often hosted outside Europe and subject to the Cloud Act.
✓ Our answer
AI is only of interest for generating quotations if it is sovereign and faithful to your catalogue. Local inference or an isolated resource hosted in France, systematic human oversight, nothing sent automatically: time gained on formatting is never paid for in a pricing error or a leak of your commercial data. The aim is not to replace the salesperson, but to hand them the quotation already put together.
Your catalogue and your client data: sovereignty & compliance
A quotation contains your prices, your discounts and your client's contact details. Here is how the architecture of our agents protects them, quotation by quotation.
Local inference
The agent can run on a machine belonging to the company: no price leaves the network, nothing passes through a public cloud.
Hosting in France
Otherwise, a dedicated and isolated resource, hosted in France under French law — your catalogue: processing and access within the European Union targeted by the architecture.
Reduced extraterritorial exposure
For your prices and your client data, 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 immunityeven when hosted in Europe by an American provider.
Isolated resource
No pooling: an environment strictly dedicated to your company and your catalogue.
Template & legal wording respected
Your PDF template and your legal wording are applied as they stand; encryption, role-based access and logging.
AI Act: governed deployment
The agent is strictly in support; no quotation is sent 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.
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.
· 41 requests matched no line in your price list. I priced none of them — I handed them back, and I will say why.
· 18 quotes issued carry no validity period. The oldest is eleven months old.
· Three prices on your public website differ from your internal price list. Your customers see the website.
· 23% of quotes never received a reply, and none was chased. morning-watch_4-flags.pdf41 off-list requests · 18 quotes with no validity
⛓ Source · 612 quotes, price list, published prices
What I record: 41 requests concern work that does not exist as such in your price list — a variant, a volume off the grid, a combination of two lines.
What I could have done: take the nearest line, apply a rule of three, or add two lines together. All three give a number that looks right.
Why I do not: a quote is an offer. The figure I write, your customer can accept — and at that moment it is owed. An extrapolated price is not an estimate: it is a commitment nobody decided to make.
What I do instead: I prepare all the rest of the quote — identity, scope described, terms, required particulars — and I leave the price line empty, with what is missing to fill it. The quote goes to a salesperson, not to the customer.
What it cost in time: nothing. The 41 came back priced in under two hours on average, because only one figure remained to write. 41-requests_empty-price-line.pdfAn extrapolated price is a commitment nobody decided
⛓ Source · 41 off-grid requests, 3 possible extrapolations set aside
Routing follows who can act: an off-grid request goes to the salesperson concerned, with the quote ready except the price; a quote with no validity period to whoever issued it; a gap between website and price list to whoever owns pricing, not to the salesperson; an unanswered quote to whoever issued it, once.
With a monthly chase. Then a monthly summary: off-grid requests by nature, quotes missing particulars, price gaps, quote reply rate.
Three guarantees hold this up, and the first is the most important in this trade. Every line of a quote comes from your price list — prices, standing discounts and charges, applied line by line, with no rekeying and no drift. An off-grid request goes to the salesperson with the quote already built, everything except the price: a new discount carries your signature, never mine. Solvency is checked at your end — a perfectly filled quote may go to somebody who will not pay, which is why nothing goes out without the salesperson's yes. And on particulars I hold to your template, the one you validated and dated: I flag what appears there and is missing from the quote, rather than a rule recited from memory.
✎ Framework · no price set, no discount, no law recited
What I am asked, in practice: "apply the usual discount", "they are a good customer", "the salesperson agreed". Across 612 quotes, 47 times.
Why I hand back today: those three phrasings are not rules. "The usual discount" is 5 % on one service and 12 % on another; "good customer" is written nowhere. I do not improvise a commercial gesture — I apply the one you wrote.
What it takes to have me handle it: a ceiling rate, the conditions that trigger it — volume, account age, term commitment, payment on delivery — and who may go beyond. Half a page.
What it gives you back: of the 47 requests, 39 fall into the three commonest cases. Once the grid is written, they go out within the minute instead of waiting for approval. The other 8 escalate, as today.
What I add along the way: every discount granted carries the condition that triggered it. After a quarter, you know what your discount policy actually costs — and today, nobody does. discount-grid_39-of-47.pdfHalf a page to write · 39 requests handled within the minute
⛓ Source · 47 requests across 612 quotes, 39 covered by three conditions
What I record: three lines of your internal catalogue show a different price from the one published on your site. The gap runs from 4 % to 19 %.
What I do today: I flag it and hand back, because nothing tells me which of the two governs.
What only needs settling once: designate the source of truth — internal catalogue or published price — and I apply it without coming back to you. Either way I keep flagging the gap to whoever maintains the other, so it gets corrected rather than endured.
Why it matters beyond three lines: a customer who saw a price on your site and receives a higher quote opens a discussion you lose in advance, whatever the right price is. On the three lines concerned, the published price was the lower one in two cases.
What I do meanwhile: I quote at the published price and flag it on the quote, in one line. You lose no deal, and the gap stays visible. 3-gaps_source-of-truth-to-designate.pdfOne decision to take once · the gap stays flagged
⛓ Source · 3 divergent lines, gaps of 4 % to 19 %, published price lower in 2 cases
What I record: of 612 quotes, 18 carry no validity period. Your quote template provides for one — these 18 were produced outside the template, in a spreadsheet or a word processor.
What that leaves open: a price proposed eleven months ago, in a context that has changed. Your customer can come back with it, and the discussion will happen with no written basis.
What I stand on, and it is firmer than a recited rule: your validated, dated template. A point of law recited from memory carries no weight in a commercial discussion; your template is enforceable internally. The applicable text is asked of your counsel, in one question — and the day their answer enters the template, it applies at once to every quote that follows.
What I do: I compare each quote with your template, the one you validated, and I flag what appears there and is missing from the quote — validity period, scope, payment terms, identity particulars.
What that gave: the 18 flagged to whoever issued them. 14 were reissued from the template, 4 concerned closed deals. And two salespeople said they did not know a template existed. 18-quotes-off-template_14-reissued.pdfCompared with the company template, not with the law
⛓ Source · 612 quotes, company-validated template, 18 gaps
What I record: 141 quotes out of 612 never received a reply, neither acceptance nor refusal. None was chased.
What that produces in your figures: they count as lost. Your conversion rate is calculated on quotes nothing is known about — which is not the same as a refusal, and does not warrant the same conclusion.
What I do: one chase, only one, as the validity period nears its end: a reminder of the quote, its date, its expiry, and an open question. No new proposal, no discount, no second chase.
Why only one: because the second no longer addresses somebody who might not have seen it. It addresses somebody who did not reply, and insisting is not an argument.
What that gave: of 141 chased, 38 replies — 17 acceptances, 21 refusals. The 21 refusals are the most useful part: they come out of the fog and sometimes say why. 141-unanswered_17-accepted-21-refused.pdfInsisting is not an argument
⛓ Source · 141 unanswered quotes, 38 replies after one chase
What the layout assembles, in the template's order: your letterhead and logo, the company's full identification, the quote number from the current series, the issue date, the customer details taken from your records, the catalogue lines with reference, quantity and unit price, the net totals, VAT by rate and the total payable.
Three statements are carried by the template and no longer depend on anyone: the legal notices in your footer, the validity period — 30 days, the value you entered in the template —, and the acceptance wording with the signature block. That is exactly what the 18 quotes with no validity period were missing: they had been produced outside the template, in a spreadsheet.
What the layout does not do on its own: it moves no block, changes no typeface, adds no line the template does not provide for. A quote that does not look like the other 611 gets noticed, and noticed the wrong way.
The PDF comes out in 4 seconds, named by your rule — number, customer, date — and filed in the deal folder. Of 612 quotes, 612 now carry the validity period and the legal notices, against 594 before.
One layout that failed, and I am publishing it: 7 quotes came out with the line table running onto a second page, that page carrying no header and no number — all quotes over 24 lines, which the template had never met. The template now repeats the header and number on every page, and I flagged the change for approval before applying it: it touches your document, and it was not mine to decide alone.
✎ Framework · laid out on your PDF template, with no invented variant
What I send alone, as soon as you open it: quotes made entirely of lines from your catalogue, under the ceiling you set, to an already registered customer, with your template's validity period. Those four bounds cover 531 of last year's 612 quotes.
What escalates: the 41 off-catalogue requests, discounts not covered by a written grid, prospects never invoiced, and any quote above the ceiling. 81 quotes out of 612, about one per working day.
What it gives you back: issuing time falls from three days to twenty minutes on the 531. Of 141 quotes left unanswered last year, some of it was the sending delay — a customer kept waiting three days has time to ask somebody else.
What the quote carries so it is defensible: the catalogue version applied, the validity period, and the name of the person it goes out under. You remain the issuer; I am the tool that composes and sends.
What I keep doing even on direct send: following up at the validity deadline. None of the 141 unanswered quotes had ever been followed up. direct-send_4-bounds_531-quotes.pdfWhat goes alone · what escalates · what the quote carries
⛓ Source · 531 of 612 quotes within the four bounds, 3 days down to 20 minutes
What is kept: the quotes issued and their lines, the catalogue version applied — the one from that day, not today's —, the validity period, the requests left unpriced and why, the discounts granted with the condition that triggered them, and the follow-ups sent.
Why the catalogue version is the key item: a quote contested six months later is defended with the rate in force the day it went out. Without that version, the discussion happens on today's rate — and you are wrong by construction.
What it gives you after a quarter: the real cost of your discount policy, the acceptance rate by service family, and the average time between sending and reply. Three figures nobody at your end can produce today.
What I do not keep, for want of use: nothing about people. Not who asked for a discount, nor who approved what in how long — except the name of the quote's issuer, which is a signature and appears on the document itself. what-you-keep_quotes.pdf6 items kept · a quote reproduced identically two years later
⛓ Source · catalogue version timestamped, 3 new indicators per quarter
Your case is not here? That is exactly what a 15-minute conversation is for. Book the free audit →
What does the agent actually do?
One task, carried out from your catalogue and your template. All the uses work in support, subject to your approval.
Reading the client's request
Identifies the part numbers, quantities and options asked for, from the client's message.
Calculation from the catalogue
Applies the prices, discounts and charges in force in your price catalogue, with no re-keying.
Layout on the PDF template
Puts the final document together on your template, legal wording and validity period included.
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.
Qualifying sales enquiries
To sort incoming emails before quotation, a dedicated agent qualifies the sales enquiries received.
Sales email qualification agent from 504 € excl. VAT / month Sales email qualification →CRM follow-up
To keep the client record up to date after the quotation goes out, a dedicated agent feeds your CRM.
Sales agent (meeting notes, follow-ups, CRM) from 592 € excl. VAT / month CRM & follow-up →In 15 minutes we identify the most relevant agent — without oversizing the project.
How much time can a sales team recover?
By automating the taking of the catalogue and the layout, the time spent on simple quotations comes down, reinvested in the conversation with the client. How large the gain is depends on your volume and remains to be confirmed by a pilot.
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.
One package, one agent
A quotation generation agent (catalogue, PDF template), installed and operated for you.
Setup + controlled subscription
- Installation, configuration and training for your teams
- Operation, human oversight, updates and support
- Sovereign hosting in France, a dedicated and isolated resource
All inclusive, no setup fee
- Setup included (installation, configuration, training)
- Operation, human oversight, updates and support
- Sovereign 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)
Four guarantees that matter to your sales activity
Related resources
Your questions, our answers
Does the agent send the quotation to the client by itself?
Can the agent get the prices wrong?
Are our prices and client data protected?
Does it connect to our catalogue and our CRM?
How long does it take to deploy this agent?
Does the agent handle complex quotations with configuration?
Other agents for sales and client relations
Let's size up the potential in your quotations
15 minutes to identify your priority catalogue — hosted in France, supervised, with no commitment.