How to prompt AI for proposal writing (with real examples)

Last updated: May 19, 2026

AI is supposed to make writing easier and more efficient. But many AEC proposal writers encounter the same block: You spend 30 minutes painstakingly inputting your RFP requirements, details about past projects, and the style and tone you’re going for, and it still comes out generic.

While AI tools like ChatGPT or Claude can create polished writing, they often fail at conveying the details that make or break the success of your proposal. Your firm’s style gets lost, crucial details get overlooked or tangled into hallucinations, and the end product reads like something that any firm could’ve written. 

As a result, many AEC marketers end up spending more time tweaking the output than they would’ve spent writing the proposal from scratch—leading to more work, not less. 

While this type of experience may tempt you to write off AI tools as a whole, don’t be so quick—AI is still a tremendously useful tool. The real secret to getting the most out of it is prompting.

This guide breaks down how to prompt AI for proposal writing: What to include, how to structure your prompts, real examples for every major proposal section, and the mistakes that quietly tank your outputs.

Key takeaways

• Prompt quality determines output quality. A vague prompt produces generic results, while a structured prompt with role, context, task, format, and tone produces something you can actually use.
• AI is most useful when applied at every stage of the proposal workflow, from RFP analysis and outline creation to drafting, personalization, and compliance checks.
• The most common prompting mistakes are easy to fix. Look out for missing client context, ignoring tone, and asking for too much at once. 
• Agentic AI is shifting the proposal writer’s role from doer to manager. Strong prompting habits are what prepare you for that transition.

Table of contents

Why prompting matters in proposal writing

The structure of a high-performance AI prompt

• Role
• Context
• Task
• Format
• Tone
• Full AEC proposal prompt intro: Example

How proposal teams should use AI across the workflow

• RFP analysis
• Outline creation
• Drafting sections
• Personalization
• Editing and compliance

AI prompt examples for each proposal section

Bad vs. good AI prompts for proposals (with examples)

AI prompting best practices for proposal writers

Common AI prompting mistakes to avoid

Looking toward the future: How agentic AI can help proposal writing

How OpenAsset Shred improves the AI prompting process

Final takeaways: AI prompts in proposal writing

Why prompting matters in proposal writing

A vague prompt is like going to a deli and ordering “a sandwich”— if you refuse to specify what kind of bread, toppings, or protein you want, the deli will make something up. 

When you write “draft a proposal cover letter,” the model has no idea who your client is, what they actually care about, what your firm has done before, or what differentiates you. It has to guess, and for AI, “guessing” means creating plausible-sounding sentences that say very little. 

Inputting relevant information helps, but only if it’s done properly: submitting a mountain of examples, assets, and won’t help if your AI tool doesn’t understand the context that connects them all. 

Research from MIT has found that AI productivity gains are highly dependent on how well users interact with the tool — not just whether they use it. The gap between a mediocre prompt and a strong one can mean the difference between an hour of editing and a section you can actually use.

For proposal teams, the stakes are quite high. Deadlines are tight: According to our State of Proposals In AEC Marketing, 64% of proposal teams report they regularly submit the day they’re due, and only 25% of teams meet 100% of their submission targets. Clients read dozens of submissions, and competition is fierce — A proposal that reads like it was written for anyone will lose to one that reads like it was written for them.

This means it’s crucial to produce proposals with speed, quality, and accuracy. AI can solve all those problems, but only if it’s prompted correctly. AI prompting for proposal writing is a skill just as important as writing or project management, and the sooner you start treating it that way, the better your outputs get.

The structure of a high-performance AI prompt

Strong prompts share a consistent structure of 5 key building blocks. It’s important to have all of them, or else the output suffers. 

1. Role

Tell the AI who it is. This is the most underused element in most prompts. For example, try saying:

“You are a senior proposal writer at a civil engineering firm with 20 years of experience in water infrastructure projects.”

Role-setting shapes tone, vocabulary, and perspective. An AI acting as a proposal writer thinks differently from one acting as a general assistant.

2. Context

Give the AI the situation. Who is the client? What’s the project? What are the evaluation criteria? What’s your firm’s relevant experience? What’s the deadline?

The more specific the context, the less the AI has to guess. For example, try: 

“The client is a municipal water authority issuing an RFP for a new wastewater treatment facility. Their stated priorities are regulatory compliance, long-term operational cost, and community impact. Our firm completed a similar project in [City] in 2022.”

3. Task

Say exactly what you want done — no more, and no less. Don’t leave room for interpretation, or try to combine a prompt meant for action with a prompt meant for brainstorming. If you want to use an AI for brainstorming or as a sounding board for possible directions to go in, do it in a separate prompting session. 

Avoid: “Write about our approach.” 

Use: “Write a 200-word project approach section that leads with regulatory compliance, references our [City] project as a precedent, and closes with a statement about our quality control process.”

4. Format

Tell the AI how to structure the output. Do you want bullet points or prose? How many words or paragraphs? Does it need headers? Should it mirror the RFP’s section structure? Should it be in HTML, text within the chatlog, or formatted into a Word document? 

Without format instructions, you’ll often get a wall of paragraphs when you need a bulleted list, or vice versa.

5. Tone

Proposal tone varies by audience and project type. A healthcare infrastructure RFP reads differently from a transportation solicitation. Similarly, an architecture firm may have a more stylized voice than a practical construction firm. Specify.

For example: 

“Write in a confident but accessible tone. Avoid jargon. The client is a public agency, not a technical audience.”

Full AEC proposal prompt intro: Example

Pulling it altogether, here’s what the basics of your prompt intro should look like. 

“You are a proposal writer at a mid-sized engineering firm. The client is a regional transit authority evaluating firms for a bus rapid transit corridor project. Their top priorities are schedule certainty, community engagement, and local subcontractor participation. Write a 250-word project approach section. Lead with schedule certainty, briefly reference our BRT experience in [City], and close with our local hiring commitment. Use clear prose, no jargon, third-person voice.”

Keep in mind that you will also need to include RFP requirements, resumes, and past projects you’d like to include. This can be done by manually uploading them to your AI tool, or by using a specialized AI tool for proposal writing that connects to your DAM or asset repository, like OpenAsset Shred. 

The second method results in significant time savings and smoother prompting, as it doesn’t require you to “train” the AI every time you need to tackle a proposal.

How proposal teams should use AI across the workflow

Most AEC proposal writers think of AI as a drafting tool. But the truth is, AI has a place at every stage of the proposal process. Here’s how to put it to work.

RFP analysis

Before you write a single word, use AI to shred the RFP. Prompt it to extract all explicit requirements, identify evaluation criteria, flag conflicts or ambiguities, and surface key dates and formatting rules.

“Review this RFP and list all mandatory requirements, weighted evaluation criteria, submission format rules, and any conflicting instructions across sections.”

This turns hours of manual compliance review into minutes. And it’s one of the highest-leverage uses of AI in the entire proposal workflow.

Be cautious, though: Free AI tools typically use user inputs to train their models, so using a free tool to shred an RFP can leak sensitive information. Additionally, because RFPs are so long and dense, some AI tools may not be able to handle the load of information. Opt for enterprise-level plans or specialized tools built for understanding RFPs. 

Outline creation

Once you (and your AI) understand the RFP, use AI to build your response outline — one that maps directly to the client’s evaluation criteria. For example:

“Based on these RFP evaluation criteria, create a proposal outline with section headers, a one-sentence description of each section’s goal, and a note on which criteria each section addresses.”

For help mapping an RFP’s requirements to your proposal outline, check out this article about crafting an RFP response. 

Drafting sections

This is where most people start, and where prompting quality matters most.

Don’t overwhelm the AI by trying to draft your entire proposal at once — this leads to low-quality responses, inadequate word count, fluff, and mixing up what’s required for each section.

Instead, tackle your proposal one section at a time, and use the five-part structure we covered earlier for every section. Feed the AI your relevant project data, your past narratives, and your differentiators.

Personalization

AI can help you tailor language to a specific client much faster than manually tweaking words line-by-line. Prompt it to rewrite a generic section using the client’s language, priorities, and terminology from the RFP.

For example, you can prompt: 

“Rewrite this project approach section using language and priorities drawn directly from the RFP. The client emphasized ‘resilience’ and ‘long-term stewardship’ — incorporate those terms where appropriate.”

Editing and compliance

After drafting, use AI to tighten prose, cut word count, or adjust reading level for a non-technical reviewer. Since this is a lighter request than full drafting, most AI tools can usually handle editing your full draft all at once — but you can still feed it by section to be safe.

You can also use AI tools to run a compliance pass. AI can cross-check your response against the RFP’s requirements and flag gaps. For example, prompt:

“Review this proposal draft against the attached RFP checklist. Identify any required items that are missing or addressed insufficiently.”

However, like RFP analysis, this is another use case that requires caution. Most AI tools can’t be 100% trusted when it comes to sensitive, meticulous requirements. Always double-check, and again, consider using tools built with AEC in mind. 

OpenAsset Shred, for instance, not only pulls requirements from your RFP to draft proposal outlines and sections, but cites everything it generates with the specific part of your RFP that inspired the output. This makes it easy to compare RFP requirements with your AI output. 

AI prompt examples for each proposal section

To make AI prompting easier for you as a proposal writer, here are a few example prompts you can use for common sections within a proposal. Feel free to copy and paste them into your AI tool and edit them to fit your needs. 

Remember that you still must add necessary supplemental information and assets. 

Executive summary 

“You are a proposal writer at a structural engineering firm. The client is a state DOT evaluating firms for a highway bridge rehabilitation program. Their top priorities are safety record, schedule performance, and local workforce participation. Write a 200-word executive summary that opens with a clear statement of our qualifications, references two relevant past projects (include placeholders), and closes with a direct commitment to the client’s goals. Confident tone, no jargon.”

Cover letter

“Write a one-page cover letter from our firm to [Client Name] for their [Project Name] RFP. Open with a sentence that shows we’ve read and understood the project’s unique challenges. Briefly state why our firm is the right choice — focus on [two differentiators]. Close with a direct, confident next step. Tone: professional but human. Avoid ‘we are pleased to submit’ or similar boilerplate.”

Approach/methodology

“Write a 300-word project approach section for a [project type] project. The client wants to see our process for [specific challenge from RFP]. Structure it in three phases: [Phase 1], [Phase 2], [Phase 3]. For each phase, include what we do, why it matters to the client, and one concrete deliverable. Active voice, clear language, no padding.”

Relevant projects/case studies

“Write a 150-word project narrative for [Project Name]. Include: project scope, the specific challenge we solved, our approach, and a measurable outcome. Tie the narrative back to [Client’s stated priority from RFP]. Use past tense, active voice.”

Team bios

“Write a 100-word bio for [Name], [Title]. Highlight their experience with [relevant project type], their role on this project, and one specific achievement that’s relevant to [client’s priority]. Avoid generic phrases like ‘seasoned professional’ or ‘extensive experience.’ Be specific.”

Compliance checks

“Review this proposal section against the following RFP requirements [paste requirements]. List: (1) requirements fully addressed, (2) requirements partially addressed with a note on what’s missing, (3) requirements not addressed. Return as a simple table.”

Bad vs. good AI prompts for proposals (with examples)

The difference between a weak and strong prompt is almost always specificity. Here are some examples of weak vs. strong prompts. 

Executive summary

Weak: “Here’s our RFP and our firm information. Write an executive summary for our proposal.”

Strong: “Here’s our RFP and firm information, as well as 5 past projects we’ve done. Write a 175-word executive summary for our proposal to [Client]. They care most about [priority 1] and [priority 2]. Reference our experience with [past project]. Open with a clear value statement, not a firm introduction. End with a sentence that connects our qualifications directly to their stated goals.”

Why it works: The strong version gives the AI a word count, a client focus, a structural direction, and a specific outcome to hit. 

The weak version produces a firm overview that could apply to any RFP, and overwhelms the AI with an overload of supplemental information, but no clear direction for how much of that information is important or relevant. This will cause the AI to pull in unnecessary data. For example, it might reference a past project to build a road when your proposal is for a boathouse.

Project approach

Weak: “Here’s the RFP, and our firm information. Describe our approach to this project.”

Strong: “Here’s an RFP for [project], our firm information, and 5 past projects. We are an architecture firm focused on [company mission] that is suited for this project because of [reason 1] [reason 2] [reason 3]. Describe our approach to [specific project type], structured around the three challenges the client raised in Section 3 of the RFP: [challenge 1], [challenge 2], [challenge 3]. For each challenge, explain our method and one measurable outcome. 250 words max. No jargon — this section will be reviewed by non-technical evaluators.”

Why it works: The AI now knows the structure, the word limit, the audience, and exactly which RFP section to address. The output targets the evaluator’s actual concerns instead of describing a generic process.

Team bios

Weak: “Here’s our project manager’s resume. Write a bio for her.”

Strong: “Write a 120-word bio for [Name], who will serve as Project Manager. Emphasize that they have 15 years of experience in [project type], managed [notable project], and are a licensed PE in [state]. The client is a public transit agency — highlight experience with public-sector clients and community engagement. Avoid phrases like ‘dedicated professional’ or ‘passionate about results.'”

Why it works: Specificity — name, title, experience, relevant achievements, audience context, and what to avoid — gives the AI everything it needs to produce something useful on the first pass. 

AI prompting best practices for proposal writers

Here are a few habits that consistently improve output quality when using AI for proposal writing. 

• Be specific and structured. Every prompt should include role, context, task, format, and tone. Think of each element as a constraint that narrows the AI’s options toward something useful.
• Include audience and project context. Who is reading this? What do they care about? What language did they use in the RFP? The more the AI knows about the evaluator, the more it can tailor the output toward them.
• Break work into sections. Don’t prompt for an entire proposal at once. Draft section by section. Each section gets its own context, task, and format instructions. This keeps outputs focused and manageable.
• Refine outputs in passes. Your first draft from AI is a starting point. Use a follow-up prompt to tighten it: “Shorten this by 50 words without losing the key points” or “Rewrite the first paragraph to lead with [specific point].”
• Validate everything. AI can misremember project details, misread requirements, or generate plausible-but-wrong information. Every output gets a human review. Always double-check facts, compliance, and that the narrative matches your intent.
• Keep a living prompt template library. Every time a prompt produces a strong output, save it with notes on what worked. Over time, you build a bank of tested prompts your whole team can use — and proposal quality gets more consistent across the board. 

For more on building an efficient review process, see this AI proposal writing guide.

Common AI prompting mistakes to avoid

Here are a few common AI prompting mistakes proposal writers in the architecture, construction, and engineering industry make. Avoid these mistakes, and your AI tool will become much more productive. 

• Asking AI to write the whole proposal at once. This produces something that feels coherent but lacks specificity in every section. Always break your prompting apart into sections. 
• Missing key context. If the AI doesn’t know who the client is, what they prioritized, and what your firm has done before, it invents all three. Again, this leads to something that sounds okay on first glance, but when you look into the details, the proposal crumbles. 
• Ignoring tone. Public sector proposals read differently from private developer RFPs. Healthcare clients have different sensibilities than transportation agencies. Specify tone every time.
• Over-relying on AI without review. AI outputs fail in several predictable ways: they inflate qualifications, smooth over gaps in context and knowledge, and occasionally fabricate project details. Every draft needs a human pass before it goes anywhere near a client.
• Using the same prompt template for every section. The compliance section and the cover letter serve completely different purposes. Build prompt templates for each type of section and keep them in a shared library your team can use and refine.

Looking toward the future: How agentic AI can help proposal writing

One exciting technological development that can help AI prompting for proposal writers is agentic AI. 

Agentic systems don’t just respond to a single prompt, but can plan, execute, and self-correct across an entire workflow with minimal hand-holding.

As Christopher Penn, Chief Data Scientist at TrustInsights.AI, puts it: 

“Imagine in the proposal workflow process, you hand an RFP and all of your necessary background data to an agent and say the goal is to craft a winning RFP response based on this input RFP. You would give it boilerplate and examples of past RFPs, winning RFP responses, especially winners that were submitted by a competing firm, buyer profiles, or even actual human profiles. And with the appropriate structure, you’d launch it, and come back in a few hours when it was done. Because you planned appropriately, the system behaved autonomously and continued to test itself and improve itself until it generated RFP responses that met the objective quality metrics you specified.”

For proposal writers, this means a shift in roles. Christopher Penn explains: 

Your role as the doer is being replaced by your role as the manager. The hottest skill in 2026 in AI is effective product and project management skills. If you’re great at writing plans, you’ll be great at today’s agentic AI systems.”

This technology is still in development and isn’t quite ready for full autonomous deployment in the proposal writing environment. However, many purpose-built AI systems are incorporating agentic AI into their software to automate smaller pieces of the proposal writing workflow. 

Strong prompting habits are the foundation of transitioning smoothly to these future tools. If you’re already good at structuring a pursuit, defining win themes, building a compliance matrix, and conveying that to an AI tool, you’re already thinking like an AI manager.

How OpenAsset Shred improves the AI prompting process

The biggest limitation of using a general AI tool for proposal writing is context. You have to manually feed it everything: your past projects, your team bios, your differentiators, and the RFP requirements. 

And since most general AI tools don’t store your data or learn from it, you have to do this every single time you create a new session. You can’t rely on the AI to use past sessions to understand contextual information like your firm’s tone or scope of work. Most AI tools can’t learn that a particular project or resume comes up in most proposals and suggest it for you. You have to drip-feed it context and start from scratch every time. 

Doing this over and over again for every AI session can quickly enter the territory of being so time-consuming that you might as well have just skipped the AI part and written it yourself.

OpenAsset Shred changes that.

Shred is built specifically for AEC proposal teams. Instead of starting from scratch each time, it connects directly to your firm’s proposal library, project records, and people profiles via OpenAsset DAM — so the context is already there. You’re not hunting through shared drives for the last time you answered a question about your safety record. Using context-driven agentic AI, Shred surfaces relevant past projects, boilerplate text, and other assets for you so you don’t have to spend as much time brainstorming what elements to include.

That changes how prompting works in practice. Instead of spending time building context into every prompt, you spend time on the strategic work, like nailing win themes, differentiation, and your overall narrative. 

Shred also automates the compliance work — like extracting RFP requirements verbatim, flagging conflicts, and surfacing key dates — so your prompts can focus on drafting and personalization rather than hunting for what you might have missed.

By making your firm’s institutional knowledge searchable and connected to the drafting process, stressed, deadline-driven proposal teams can reduce time spent searching for existing content — one of the most tedious, time-consuming parts of the proposal process. 

Final takeaways: AI prompts in proposal writing

AI can’t write proposals. Proposal writers who know how to use AI write better proposals, faster.

The gap between a generic output and a genuinely useful one is almost always the prompt. If you can master prompting AI with role, context, task, format, and tone, you’ll spend far less time editing and far more time on the strategy that actually wins work. 

Building templates, splitting your request into sections, refining and verifying outputs, and using purpose-built AI tools for the architecture, engineering, and construction fields is also important. 

Prompting is a skill like any other. It gets better with practice, with feedback, and with the right tools behind it.

Ready to see what AI-assisted proposal writing looks like when it’s purpose-built for AEC? See how OpenAsset Shred works — and how it’s helping proposal teams spend less time searching and more time winning.