.svg)
Synopsis
An engagement with a B2B organisation of 80 employees, focused on producing sales proposals for large-account tenders.
For a similar project, discover our custom AI agent development offer.
The engagement combined a phase modelling the pre-sales know-how with the deployment of an AI agent dedicated to production.
Over six months, the production cycle was cut from several days to under an hour, and the sales leadership was able to reinvest the time saved in higher-value phases!
The scope of the engagement
The organisation produces an average of 120 sales proposals a year, mainly in response to large-account tenders. Each proposal took up an account director for two to three full days.
When the sales leadership approached us, the reason for the engagement initially seemed technological: the aim was to introduce an AI agent into the pre-sales chain to shorten lead times.
The audit phase revealed a different reading of the issue.
Diagnosis
Observing how the account director worked revealed a pattern common in organisations that have deployed generative AI at individual level, without modelling how it is used.
Producing a proposal involved several chat interfaces at once, each used for one fragment of the deliverable. Prompts were rebuilt for every bid. Reusable material - past proposals, client presentations, price grids - was retrieved by hand, with no document governance or knowledge capture.
More fundamentally, the organisation's sales know-how - its pricing logic, its scope structure, its strategic signature - was not formalised anywhere. It lived in the heads of the account directors, and was replayed identically, by hand, for every bid.
This diagnosis changed the scope of the engagement. The issue was not introducing a tool. It was first formalising an undocumented intellectual asset, a prerequisite for deploying an AI agent in a meaningful way.
Approach
The engagement took place in two stages.
Phase 1 - Modelling
Over six weeks, we rebuilt the pre-sales production chain with the sales leadership, formalising three levels of objects: the pricing logic and its variables; the standard structure of a proposal by type of bid; and the organisation's strategic and stylistic signature.
This phase did not involve any technology.
It involved senior staff time, and an interviewing and mapping methodology.
In our view, it is the core of the value of an engagement of this kind.
Phase 2 - Industrialisation
Once this model was in place, deploying the AI agent became an extension of the in-house process rather than the introduction of a third-party tool.
The agent takes the transcript of a discovery meeting as input and, within minutes, returns a proposal in the organisation's format, in line with its pricing standards and narrative structure.
The human phase shifts: the account director now spends around thirty minutes reviewing and making strategic adjustments.
The judgement calls that give a large-account proposal its value - nuance, differentiated recommendations, strategic stance - remain entirely in their hands!
Lab engagement
Assess the maturity of your pre-sales chain We offer the sales leadership of large groups a 45-minute scoping interview, led by a Lab partner, to establish an initial diagnosis of how mature the modelling of their pre-sales processes is.
Request a scoping interviewMetrics
|
Production cycle for a proposal from 2.5 days on average to 45 minutes |
×96 |
|
Annual volume of proposals handled with the same sales headcount |
+40% |
|
Tender response rate 48-hour response window met |
×1.8 |
|
Senior time redeployed to strategic value discovery, recommendation, negotiation |
~2,500 h/year |
Beyond the quantitative metrics, the main contribution of the engagement is qualitative.
The sales leadership regained its ability to make choices on its bids: selecting the highest-value tenders, deepening the discovery phases, reworking strategic recommendations ahead of closing.
All activities that the pressure of the production schedule had until then pushed to the margins.
Reading the engagement
Three lessons deserve to be put into perspective, beyond this particular case.
First, the bottleneck in a pre-sales chain is rarely technological. It lies in the lack of formalised sales know-how - which makes every deliverable handcrafted, and therefore slow and structurally unscalable.
Second, generic AI tools, deployed at individual level, can make this underlying problem worse. They create one-off productivity gains that hide the lack of systemic modelling, without capitalising on the organisation's intangible assets.
Finally, the value of a custom AI agent is measured less by speed of execution than by how senior time is requalified. The point is not that the proposal is produced faster. The point is that the hours saved are reinvested in the judgement calls that make the difference in a large-account buying decision.
It is on this condition that an AI agent deployment stops being a tooling topic and becomes a lever for transforming the sales function.
AI Sisters Lab
Start thinking about your pre-sales chain The Lab supports the executive and sales leadership of large groups on modelling and AI agent deployment engagements with high quality requirements. Engagements are custom-built, run over several months, and overseen by a Lab partner. Initial conversation with Chloé Pariente, CEO of AI Sisters Lab.
Book a callFurther reading
AI Sisters Lab engagements More Lab case studies AI consulting for large groups
AI Sisters Lab engagementsMore Lab case studiesAlso read: Molotov.TV's journey to AI with the AI Sisters method and OpenAI's AgentKit: features and limitations
Where does your company stand with AI?
Take the free AI assessment in 5 minutes and get your priority actions.
Take the assessment
.avif)
