From quote generator to
deal communication platform

The product started from "AI maakt je offertes. Jij stuurt ze": AI as a faster way to produce a document. Watching real vendors work shows the document is the easy 20%.

The hard 80% sits around it:

• Scattered inbound. Interest arrives over phone, WhatsApp, e-mail and messenger.
• The intake grind. Rounds of questions, photos, sometimes a site visit, waiting on details.
• After the quote is sent. Questions, objections, change requests, silence, chasing.

That surrounding exchange decides whether the deal happens, and it is exactly the part no tool carries today.

So the reframe: stop optimizing document production, and carry the whole exchange instead. Every step of that exchange can be enhanced, or fully run, by AI agents. That is a different product with a different ceiling:

The build already leans this way without naming it: reply-threaded e-mail conversations, the Klant tab, and WhatsApp intake sharing one conversational brain. This document names the destination.

Every deal moves through the same stages, whatever the trade. The strategic claim is the bracket: lead generation before it, and invoicing, payment and job execution after it, are other tools' products. Offerte.cc owns everything from first contact to agreement, and hands execution the packet it needs (invoicing at agreement stays the parked, natural extension).

Two things in this picture matter more than the stages themselves:

• The negotiation loop. Discuss feeds back into Draft until both sides agree. That loop is where deals are won or lost today, with zero tooling.
• The agreed deal is a context asset. By the time the deal is done, the thread contains the job description, the constraints, the photos and the financials that execution needs. Nobody should retype that.

A quote prices work you already know how to do. A proposal designs the "how" first. The test: do you already know how the problem will be solved? Then it's a quote. If not, you're writing a proposal.

The sharper insight is that certainty lives per side. The vendor has painted 200 m² of living room a hundred times: for them it's a quote. The customer buys this once: they need the proposal story. What will happen, in what order, why this price, what's included and what isn't.

Product implication: the customer-facing document should be generated for the reader, not formatted from the ledger.

• One deal, two renderings. The vendor's view is line items, internals and margins; the customer's view is a story wrapped around a price.
• The first expressions exist. The brief and /klantvriendelijk (restructure the quote for the customer) already point this way.
• The endgame. Every deal automatically gets both renderings, and the further right a job sits on the spectrum, the more the customer view earns its "proposal" name.

Two axes sort every target business: job complexity (do you need photos, a site visit, real discovery, or have you done this exact job a million times?) and deal value (with volume roughly its inverse). Different quadrants need different products, and they pay differently.

Reading the map:

• Complexity decides which product the customer meets: an instant configurator, or a patient agent that collects photos over days.
• Value decides what the vendor will pay. The bottom-left runs on volume and speed; the top-right runs on win rate and deal size.
• The bottom-right corner is not a target (complex jobs that pay little), unless agents make intake nearly free.

v2, the US mapping: the model transfers unchanged, only the example dots move country.

• Bottom-left: pressure washing, gutter cleaning, window cleaning. The co-founder's proven outreach territory.
• Top-left: tiny-house builders and prefab/ADU sellers. Configurable but high-ticket, which is why a configurator closed the first paid deal.
• Top-right: remodelers and GCs, unchanged.
• The seam: the paid proof and the outreach proof live in different quadrants, and discovery (not preference) picks which one the first cohort tests.

The customer picks the channel; the product never makes them move. Whatever they use, WhatsApp, e-mail or the web, it all lands in one deal thread that remembers everything.

WhatsApp is the flagship because it is naturally asynchronous: the agent does not need to hold a live transcript. During quote creation a background agent keeps working the deal: collecting missing information, fetching external pricing, and when a new question pops up, reaching out again through the same channel. The thread is staffed even when nobody is looking at it.

Where this bites today:

• The pieces exist (web intake transcripts, WhatsApp intake sharing the same brain, e-mail replies threading into quote comments), but they live on separate surfaces.
• There is no single vendor-side inbox where "the deal with Pieter" is one continuous thread across its channels and stages.
• v2 changed when, not whether. The model holds as the destination; demand pulls the unification in (roadmap chunk 4), it is no longer the first thing we lay down.

How automated a deal can be is not a product setting, it is a property of the job: how certain is the price? When confidence is 99.9%, or the pricing model is simply fixed, the quote can be instant. When it's low, an agent needs to go gather the ground truth first.

The instant end is bigger than a speed win. A wedding photographer's customer can play with hours, travel time and the number of edited photos and watch the price update live. The vendor closes the deal at inbound speed; the customer gets to explore "what would it cost me if...". Both sides win, and being first with a number wins deals on its own.

The dial also resolves the automation-trust question cleanly: nothing is forced.

• A vendor who wants to review everything sets the gate at zero.
• A vendor with fixed packages lets instant quotes fire straight from an inbound WhatsApp message.
• Confidence is already partially measurable (per-item confidence scores, pricing-issue flags); the dial turns that signal into a feature.

The same dial, three customer experiences

The dial doesn't only decide who does the work; it decides what the customer should see. Getting to a price is a different experience per zone, and the surface should match:

• The configurator. When the price is 100% certain, put the quote builder in the customer's browser: play with the options, watch the price update live, book on the spot. Right for a wedding photographer or a fixed window round.
• The itemized quote. When the line items are the product (painting, plastering), the job gets walked through or visited, and the deliverable is the list: groups, quantities, unit prices. The customer reads it line by line, so the list has to carry the deal.
• The proposal. When the work needs a story, the deliverable is a presentation: photos, narrative, options to choose between, with the prices living inside the story instead of stacked at the end of a table.

This is the product implication of the quote-proposal spectrum in section 3: the customer view is not one template that stretches. The vendor picks the surface per service (or the confidence signal suggests one), and the same deal thread powers all three.

Is this an administration tool that makes quote paperwork easier, or a sales tool where quote generation is the core of how a business wins work? The two identities have different features, different buyers, and a 10x price gap.

An admin tool prices against Word and Excel, which cost nothing: the ceiling is €20 to €30 per month. A sales tool prices against revenue won. The proof point exists: a freelancer sells custom-built versions of exactly this workflow at €300 to €400 per month, to businesses that happily pay it because it wins them deals.

Mapped against the lifecycle. The honest picture: drafting is genuinely strong, the channels exist individually, and everything that makes this a platform rather than a set of features is still open.

Not a sprint plan. v1 ordered these by architecture: build the platform premise (the unified thread) first. v2 orders them by demand: money has changed hands for the configurator + follow-up pair, and nobody has paid for the inbox. The thing the market bought goes first; the foundation gets pulled in when the deals demand it.