PRIME
The Construction Prompting Framework
Five elements that turn a vague AI request into a structured brief, producing output you would actually send.
Prime your prompt before you prompt.
AI output quality is proportional to the quality of your brief.
Most people who tried AI and gave up were writing one-line prompts. The problem was never the tool. It was the briefing. PRIME fixes that, with no technical knowledge required.
Five elements. One prompt.
Each element adds a layer of specificity. You do not need all five every time. When you need something polished, all five is how you get there.
Without a persona, AI defaults to a generic assistant voice. With one, it writes from a specific professional frame — with the vocabulary, assumptions, and priorities of that role. The same request produces very different output depending on who is doing the writing.
Try: Contracts Administrator, Project Controls Analyst, Construction Finance Manager, Owner's Representative. Each shifts the lens of the entire output.
The task itself. One deliverable, clearly named. If you need two things, write two prompts. Compound requests produce compound mediocrity. AI splits attention across tasks and does none of them as well as it would with a focused brief.
Other construction deliverables this works for: budget variance explanation, draw request cover letter, schedule delay memo, subcontractor response, executive summary.
This is the scope of work for the response. The more specific your instructions, the less AI has to guess, and guessing is where it goes wrong. Think of it as the instructions you would give a competent colleague before asking them to write something on your behalf.
Structure, order, and level of detail all go here. The more specific, the less AI has to fill in.
Metrics do not describe structure. They describe who needs to read the output and what they need to walk away knowing. This is the element most people skip, and the one that separates competent output from output that is ready to send.
Write from the reader's perspective, not the writer's task list. This reframes the entire output.
Fences out the patterns AI defaults to when left unconstrained. Most useful after reviewing a first output and noticing what needs removing. Exclusions let you close the gap between what AI produced and what you actually needed, without rewriting the whole prompt.
Even one or two well-chosen exclusions can dramatically sharpen an otherwise decent output.
What a complete PRIME prompt looks like
All five elements, assembled into one complete prompt. This produces a budget variance explanation a CFO would send to an owner.
As a project controls analyst at a commercial general contractor,
write me a budget variance explanation.
Include:
a) A one-line summary of the total variance amount and whether the project is over or under budget
b) The top two causes of the variance, with the specific line item and dollar impact for each
c) The recovery plan or mitigation steps being taken, with owners and timelines
The owner should be able to read this in under two minutes and understand exactly where the budget stands without requesting a follow-up call. The project executive should be able to walk into an owner meeting with this document and need nothing else.
Do not use hedging language or vague phrasing. Do not assign blame to subcontractors or external parties without specifics. Keep the tone factual and professional. Three sentences maximum per section.
Where PRIME works in construction
Any written deliverable your team produces on a recurring basis is a candidate for a PRIME prompt. The examples below are a starting point, not an exhaustive list.
- Budget variance explanations for owner meetings
- Lender draw request cover letters
- Month-end narrative reports
- Accounts payable aging summaries
- Responses to auditor questions on WIP entries
- Job cost report narratives
- Schedule delay justification memos
- Weekly cost report narratives
- Change event log summaries
- Percent-complete justifications
- Risk register narratives
- Owner status updates
- Board updates on project portfolio health
- Go/no-go recommendation memos
- Executive summaries for lessons-learned
- Subcontractor performance summaries
- Kickoff package narratives
- Scope clarification responses to bid queries
- Value engineering recommendation summaries
- Exclusions and clarifications for proposals
- Bid assumption log narratives
- Proposal cover letters
Did you find this useful?
There is a half-day workshop built for your team.
The AI at Work workshop is a hands-on session designed specifically for construction teams. Your people leave with real skills and a reusable prompt library built during the session itself. Virtual or in-person.