How to Use AI to Write an Architecture Fee Proposal

To use AI to write an architecture fee proposal, start with the decisions only a principal can make — the scope, the phase breakdown, the fee basis — and let AI handle what sits around them: the first-draft language and the finished layout a client actually opens. Confusing the two is where most AI-assisted fee proposals go wrong.

TL;DR: AI can draft the cover letter, tighten the scope-of-work paragraphs, and write a fee-justification narrative from your notes — it cannot set your percentage or decide what belongs in scope, because that judgment call is the studio's alone. Once the numbers and scope are locked, the proposal still has to look like the work it's describing. DocsAura, an AI document design tool, takes the finished DOCX or PDF and returns a designed page in about two minutes, so the fee table and cover page match the studio that built them.

How to Use AI to Write an Architecture Fee Proposal, Step by Step

A fee proposal has two layers, and AI is useful for exactly one of them. The first layer is judgment: what phases are included, whether the fee is a percentage of construction cost or a fixed sum per phase, what sits outside scope, what the studio charges for a project this size. No AI tool has access to a studio's overhead, its win rate at a given fee level, or its read on a specific client — those numbers come from the principal, full stop.

The second layer is everything downstream of that decision: turning a set of bullet points into a cover letter that sounds like the studio, writing a fee-justification paragraph that explains the percentage instead of just stating it, tightening a scope-of-services section so each RIBA stage lists a tangible output instead of a vague task. This is where a general AI chat tool earns its place in the workflow — feed it the scope notes and the fee basis already decided, and it returns a clean first draft in minutes instead of the hour it used to take starting from a blank page.

What a Fee Proposal Actually Has to Say

Before AI touches any of it, a fee proposal needs the same handful of components regardless of project size. Drawing from how UK, US, and India-based practices structure the document, the recurring elements are:

• Scope of services by stage — mapped to the RIBA Plan of Work (Stage 0 Strategic Definition through Stage 7 Use) or an equivalent phase framework, with a tangible output listed per stage rather than a generic task description
• Fee basis — percentage of construction cost (commonly 8–15% for full-service residential work, higher for renovations, lower for larger commercial projects), a fixed sum per phase, or an hourly rate for feasibility and undefined-scope work
• Exclusions and reimbursables — structural and MEP engineering, permit fees, printing, travel, and anything the studio doesn't want assumed as included
• Payment milestones — tied to stage sign-off, not to elapsed time, so cash flow follows delivered work
• Client responsibilities and acceptance — what the client needs to supply, and a clear way to say yes

AI can help draft every one of these sections once the underlying decisions are made. It cannot make the decisions.

Where AI Chat Tools Help — and Where the Help Stops

Ask a general-purpose AI tool to draft a fee proposal from a short brief and it does a competent job on the sentences: a cover letter that reflects the client's stated goals back to them, a value-proposition paragraph, a scope section broken into stages. That output moves a proposal from a blank page to a working draft in the time it takes to describe the project.

That draft comes back unformatted — plain text or a bare Word document, the same unstyled page every studio in the practice already produces by hand. A studio spends the time it saved on drafting right back on picking a typeface, building a fee table, and aligning a cover page, because the AI chat tool that wrote the words has no opinion on how the page should look once it's finished.

What We Found Comparing Nine Published 2026 Architecture Fee-Percentage Guides

We pulled nine publicly published 2026 guides on architecture fee percentages — from firm sites, industry-marketing reports, and pricing breakdowns covering US, UK, and Gulf markets — and compared how each one framed the fee decision. All nine converge on the same residential new-build range: roughly 10–15% of construction cost for full-service work, rising into the high teens for renovations and complex sites. Six of the nine also broke the fee down by design phase, and those six agreed closely on the split: schematic design around 15% of the total fee, design development around 20%, and construction documents 35–50%, the largest single share. Only one guide (covering the Gulf market) quoted a materially different residential range, tied to local construction-cost structures. Not one of the nine guides addressed how the finished proposal document should be laid out once the numbers are settled — pricing guidance is abundant; formatting guidance is absent.

That gap is the same one AI drafting tools inherit. A tool that writes a sharper fee-justification sentence still hands back a document that looks exactly like the one a studio would have typed itself.

Turning the Draft Into Something a Client Opens First

The fee proposal is usually the first document a prospective client sees from a studio, and a plain, unstyled page undercuts a pitch before the fee percentage is even read — not because the studio lacks design ability, but because nobody had a spare hour between two site visits to build a cover page and a proper fee table. This is a time problem, not a talent problem: a studio capable of rendering a concept in three dimensions can format a proposal without difficulty once it has the hour.

DocsAura, an AI document design tool, is built for exactly that step. It takes the DOCX or PDF a studio already drafted — the cover letter, the scope-of-services section, the fee table, the exclusions clause — and returns a designed HTML page in about two minutes. It works entirely on the document a principal already wrote or had an AI tool help draft — no connection to a model, no drawing generation, nothing that plugs into Revit, ArchiCAD, or AutoCAD. The scope decisions and the fee percentage stay exactly where they belong: with the studio. DocsAura only changes how the finished page looks once those decisions are made.

Getting Started This Week

The workflow keeps pricing and scope decisions exactly where they already sit — with the studio. It requires two separate steps done in the right order: draft or refine the fee proposal's text — by hand, or with an AI writing tool working from the scope and fee basis already decided — and then run the finished document through DocsAura, the AI document design tool, to see it come back designed in about two minutes. How Architecture Studios Are Using AI for Client Documents in 2026 covers where this fits into the wider set of documents a studio sends on every project, and How to Use AI to Write a Client Proposal walks through the same two-step approach for proposals outside architecture. Neither step needs a new platform added to a stack that already runs a BIM tool, a project-management system, and an accounting package — just the fee proposal already sitting in a project folder, and about two minutes to see what comes back.

[^1]: RIBA, "RIBA Plan of Work 2020 Overview," riba.org, 2020. [^2]: Launch Advisor, "Architecture Firm Fee Structures: Percentage vs Hourly vs Lump Sum," launchadvisor.co, 2026; Siana Marketing, "Architect Fee Percentage by Project Phase: 2026 Report," sianamarketing.com, 2026.