The AI Offer Writer skill for Claude

The AI Offer Writer is a free skill from our own skill library. A skill is an instruction file, SKILL.md, that gives an AI assistant a fixed way of working for one task. Not software you install, not a subscription, not a connection to your bookkeeping: one text file of 2,565 words that tells Claude how a proposal should be put together, which information it has to ask for, in which order the document is built and what it may never make up itself. You download the zip at the top of this page, put it in your Claude environment, and from that moment Claude writes proposals along this logic.

You probably recognise the problem the skill solves straight away. You had a good conversation, the client wants a proposal, and then the shuffling starts. Half a day goes into a document that opens with your company history. The price sits in it bare, with no context, and after you send it there is silence. Or worse: a yes does come, but only on the cheapest option. Those are exactly the points it takes on. It builds the document from the conversation instead of from your company, it puts the price into a build up based on research into how people decide, and it closes with one next step with a date, so you know when silence stops being waiting.

We wrote it for service businesses in the broad sense: freelancers, agencies, construction and installation firms, coaches and consultants. Anybody who has to put something on paper after a conversation that needs a signature under it. It is part of the skill library we make available free from our AI and automation service, with no account and no sales email afterwards.

One thing to know in advance: the skill writes, it does not sell. The quality of the proposal stands or falls with the quality of your conversation and your notes. No structure saves a bad conversation, and the skill says so itself: without an intake, every proposal becomes an empty template.

Anyone who loses a lot of proposals usually looks for the cause in the price. Almost always it sits somewhere else. Inside sits a separate review function for existing proposals, and its checkpoints read like a list of what goes wrong in practice.

The reading order is wrong. Almost every client scans first and reads afterwards, and whoever scans jumps to the price. If the result of the work only appears after the price in the document, the client reads an amount without knowing what stands against it. Every amount then feels too high. That is why the skill always puts the result before the investment: first where the client will be, and only then what that costs.

The document is about the wrong party. Count in your last proposal how often we, us and I appear against you and your. In proposals that do not get signed, that ratio is almost always skewed. This one turns it around: the opening is entirely about the client, with at least one literal quote from the conversation, and the block about you is five lines at most and comes almost at the end.

The promises cannot be checked. We optimise your processes: nobody can verify whether that worked. Your quotes go out within a day: anybody can verify that. It rewrites every promise into the second form, and in the review function it marks every sentence that could have come from any supplier at all.

It does not say what is not included. Most of the argument about extra work arises because the proposal only describes what you do. The block with what falls outside the job is compulsory in this skill: it delivers no proposal without one.

The ending is limp. We hope this proposal meets your needs: that is not a next step, that is a sigh. The document ends with one action, one date, and what happens if the client does nothing. It rejects two next steps, because two requests at once is one too many.

In diff form, with lines from the example in the skill: what goes out and what comes back in its place.

A skill is only as good as its instructions, so we simply describe them here. The file opens with frontmatter saying when Claude should pick the skill up. Not only on the obvious questions such as write an offer, make a proposal or draw up a quote, but also on complaints: I lose half a day on every quote, I never hear back after sending it, they always jump straight to the last page, I only ever get a yes on the cheapest option. Say something like that to Claude while the skill is loaded and you get it automatically.

Then comes the core: a fixed offer logic of six steps. Acknowledgement, situation, result, approach, investment, next step. While writing, the skill carries out eight fixed actions. It pulls the client's literal words out of your notes, because those words go back into the document. It opens with acknowledgement, with no summary of your company. It describes the situation with a strict separation between what the client said and what you infer from it. It phrases the result as something the client can check.

It cuts the work into two to four phases. It builds the investment into three options, from high to low. It closes with one next step with a date. And every field that needs a number, a name or a reference you did not supply, it marks as [FILL IN].

Before anything is written, the skill runs through a required input checklist: your notes or intake report, what you deliver and what you explicitly do not, your pricing, the name of the client and the contact and whether that person decides, the lead time wanted, the evidence available, the conditions that have to be named and the tone you want. It only asks for what is missing, so you do not get a form but at most two or three targeted questions.

A striking number of rules are about language. Right at the top it forbids itself brochure language and proposal cliches: tailor made, unique approach, we take the worry off your hands, a partner rather than a supplier. It alternates short sentences with long ones, chooses a figure over an adjective and writes results in a form the client can check. And it has a separate review mode: hand it an existing proposal and you get an analysis of the reading order, a count of me against you, a mark on every cliche sentence and every promise that cannot be checked, the missing blocks, and a rewritten version of the three weakest blocks you can paste straight in.

The file closes with a source list of eight titles, from Minto to Rackham. More on those below, because that theory is the reason the structure is what it is.

Writing proposals is largely craft with no single inventor, but the choices in this skill can be traced back to research and to authors who wrote about it explicitly. The SKILL.md names its sources, and that is deliberate: it lets you read them and decide for yourself whether you agree.

The order comes from Barbara Minto. Her Pyramid Principle, developed at McKinsey and described in 1987, says: start with the conclusion, not with the run up. For a proposal that means the result first, and only then the approach and the price. A client who has to discover on page four what they get has already jumped to the price on page two. If you want to apply that way of thinking more widely than proposals, there is a separate Pyramid Principle Writer skill.

The anchor comes from Tversky and Kahneman. Their 1974 research into decision heuristics showed that the first number somebody sees colours their judgement of every number that follows. That is why the most expensive package sits at the front, and why the proposal names what the problem costs the client before the price of the solution drops.

The third option comes from Huber, Payne and Puto. Their 1982 research described asymmetric dominance: an extra option that is clearly less attractive than one of the other two shifts the choice towards that other one. That effect is used here with an explicit honesty rule attached: every option has to be an offer you can and want to deliver.

The pricing philosophy comes from Weiss and Enns. Value Based Fees and Pricing Creativity turn on the same point: the amount belongs next to the value of the problem solved, not next to your hourly rate. How you work out that value and which amount then goes in the place of [FILL IN] is the territory of the Pricing Strategy Coach skill: those two have been kept apart deliberately.

The evidence block leans on Cialdini, the groundwork on Rackham. Influence from 1984 explains why one concrete, checkable piece of evidence weighs more than five superlatives, and why the skill would rather leave the block out than fill it with air. And Neil Rackham's SPIN Selling from 1988 is about the conversation before the proposal: ask out the cost of the problem first, and only then start on the solution. That conversation has its own skill, the SPIN Selling Conversation skill, and anyone who wants to learn the psychology behind persuasion more broadly can go to the Cialdini Persuasion skill.

Even without installing the skill you can improve your own proposal with these principles: put the result before the price, separate what the client said from what you assume, and dare to write down what is not included.

The zip contains one folder with the SKILL.md in it. Installing is a matter of putting the file in the right place, and that place differs per environment. SKILL.md has been an open standard since December 2025, so the same skill also works in Codex, Cursor and Gemini CLI. You are not downloading a Claude file but a working instruction any modern AI assistant can read.

After that, using it is simple: paste your conversation notes and ask for a proposal. The rawer the notes, the better, as long as there is real information in them. Whatever is missing, it will ask for itself. Should you be new to skills and want to understand how these files work first, the knowledge base has the wider explanation of working with AI.

This one is at its strongest on programmes and consultancy work where the scope can still be steered: that is where the full structure with three options comes into its own. For a well defined delivery where the client knows exactly what they want, it falls back to one price with no options. For small amounts where a client signs without consulting anybody, a shortened one page proposal is enough: acknowledgement, result, price and next step.

And if several people sign, it writes a separate paragraph for the financial decision maker and one for the decision maker on substance. For follow up work with existing clients it skips the Why us block and starts with what has changed since the previous job.

There are also situations where you are better off leaving it alone. Legal documents are one of them: a proposal is not a contract, and the skill refuses to write text meant to serve as terms and conditions. Public tenders with formal requirements are a second: there the specification sets the structure, not sales logic. And the most important limit is the intake itself. If you have not spoken to the client yet, the skill has nothing to work with and the result becomes an empty template with nice headings. First the conversation, then the proposal. To get your conversation reports in better shape, start with the Intake Summary skill.

One more honest limit: the skill improves your proposal, not your offer. If the price really is too high for the market, or the offer does not match what the client needs, a better structure changes nothing about that. What it does do is make sure a good offer no longer falls over on a bad document.

A better proposal raises the chance of a signature, but there has to be somebody to send it to. If too few enquiries come in, the real work sits before this document: being found by the people looking for you. The free SEO scan shows in seconds where your site stands. And if you want to talk about what AI can do for your business beyond this, from separate skills to full automation, we simply do that in a conversation.