AI for Proposal Development

Shredding the RFP: extracting every requirement

Shredding an RFP means systematically reading the entire solicitation and extracting every requirement — every shall, will, must, is required to, is expected to — and cataloguing it with its location, the evaluation factor it maps to, and the type of response it requires.

On a 200-page RFP, a thorough shred can identify 150-400 individual requirements. Manually, this takes a skilled proposal manager 8-16 hours. The risk of human error increases with document length and complexity — requirements buried in attachments, cross-referenced in multiple sections, or stated differently in the SOW versus Section L are the ones most commonly missed.

AI performs the shred in 15-30 minutes and catches requirements that humans routinely miss on the first pass. Here is the prompt template:

You are an expert proposal manager at a government contractor.
Shred the following RFP and extract every requirement.

[Paste the full RFP text — SOW/PWS, Section L, Section C, all
attachments]

For each requirement, provide:
1. Requirement ID (sequential numbering)
2. Exact requirement text (quoted from the RFP)
3. Source section and page reference (e.g., "Section C, para 3.2.1")
4. Requirement type: SHALL (mandatory), SHOULD (desired), MAY
   (optional)
5. Evaluation factor it maps to (from Section M)
6. Response type needed: narrative, staffing, process description,
   past performance example, or deliverable/CDRL
7. Cross-references to other requirements in the RFP that relate to
   this one

After extracting all requirements, provide:
- Total count of SHALL vs SHOULD vs MAY requirements
- Requirements that appear to conflict or are ambiguous (flag for
  questions to the Contracting Officer)
- Requirements that are referenced in Section M evaluation criteria
  but not detailed in Section L (potential risk areas)

The output is not your final compliance matrix. Your proposal manager reviews it, corrects any misclassifications, adds judgment about how requirements cluster into response sections, and makes decisions about emphasis. But they are refining a comprehensive first pass, not building from scratch.

Building the compliance matrix

The compliance matrix is the backbone of a winning proposal. It maps every Section L instruction to the corresponding Section M evaluation criterion, identifies where each requirement will be addressed in your proposal, and ensures that no requirement falls through the cracks.

A compliance matrix serves two purposes. First, it is a management tool: it tells your writers exactly what they need to address in each section and how much weight the evaluators place on it. Second, it is a quality control tool: during colour team reviews, the matrix is how you verify that every requirement has been addressed.

AI builds the first-draft compliance matrix directly from the RFP shred:

Using the requirements extracted from the RFP shred, build a
compliance matrix with the following columns:

1. Requirement ID
2. Section L Instruction (exact text)
3. Section M Evaluation Factor and Subfactor
4. Evaluation weight or priority (if stated in Section M)
5. Proposed response location (Volume, Section, Paragraph)
6. Page allocation (suggested based on evaluation weight)
7. Assigned writer (leave blank for PM to assign)
8. Status (leave as "Not Started")
9. Compliance notes (any specific formatting, content, or page
   limit requirements)

Organise the matrix by evaluation factor (highest weight first).
Flag any Section L requirements that do not have a clear mapping
to Section M criteria — these are compliance risks.
Flag any Section M criteria that are not addressed by specific
Section L instructions — these need proposal manager judgment on
where to address them.

The critical value of AI here is not just speed — it is completeness. AI identifies the cross-references and gaps that even experienced proposal managers sometimes miss. When Section M says "The Government will evaluate the offeror's approach to quality control" but Section L does not explicitly ask for a quality control section, AI flags that gap. Your proposal manager then decides where quality control is addressed — but they know about the gap on Day 1 instead of discovering it during Red Team.

AI-assisted draft generation

First-draft generation is where AI saves the most raw hours — and where the most discipline is required. An AI-generated first draft that is not grounded in your specific solution, your actual past performance, and the RFP's evaluation criteria will produce the kind of generic proposal language that evaluators see in every losing bid.

The key is what you feed AI before asking it to draft. The quality of the output is directly proportional to the specificity of the input.

Before AI drafts any section, it needs:

1. The specific Section L instructions for that section (from your compliance matrix)
2. The Section M evaluation criteria and their weights
3. Your win themes for this opportunity (from your capture plan)
4. Your actual technical approach — what your team will actually do, what tools you will use, what methodology you follow
5. Relevant past performance examples that demonstrate you have done this before
6. Page limits and formatting requirements

The prompt template for a technical approach section:

Draft a technical approach section for a government proposal.

SECTION L INSTRUCTIONS:
[Paste the specific Section L instructions for this section]

EVALUATION CRITERIA (Section M):
[Paste the relevant Section M evaluation factors and subfactors]

OUR TECHNICAL APPROACH:
[Describe your actual approach: methodology, tools, processes,
staffing model, transition plan — be specific about what makes
your approach different]

WIN THEMES TO EMPHASISE:
[List your 3-5 win themes with supporting evidence]

RELEVANT PAST PERFORMANCE:
[Summarise 2-3 relevant past performance examples that demonstrate
your approach works]

PAGE LIMIT: [X pages]

Write in active voice, third person. Address every element in the
Section L instructions. Lead each major section with a win theme.
Include specific references to past performance that demonstrate
capability. Use concrete metrics where available (e.g., "reduced
processing time by 40%" not "significantly improved efficiency").
Do not use generic phrases like "proven methodology" or
"industry-leading" without specific substantiation.

The AI produces a structured first draft that addresses every Section L requirement, incorporates your win themes, references your past performance, and stays within the page limit. Your SMEs and writers then refine the technical content, add specifics that were not in the input, and sharpen the language. The draft-to-final revision takes 40-60% less time than writing from scratch.

Past performance narratives

The past performance volume is simultaneously the easiest and the most frustrating part of a proposal. It is easy because you already have the data — contract numbers, period of performance, dollar values, scope descriptions, and CPARS ratings. It is frustrating because every RFP asks for the same information in a slightly different format, and your team ends up rewriting the same past performance narratives from scratch for every proposal.

AI eliminates this rewrite cycle. With your past performance library loaded as context, AI can generate a past performance narrative tailored to any RFP's specific format requirements in minutes.

The prompt template for a past performance narrative:

Draft a past performance narrative for a government proposal.

RFP FORMAT REQUIREMENTS:
[Paste the past performance instructions from Section L — what
information is required, format, page limits]

EVALUATION CRITERIA:
[Paste Section M criteria for past performance evaluation]

CONTRACT DATA:
- Contract number: [number]
- Agency: [name]
- Period of performance: [dates]
- Contract type: [FFP/T&M/CPFF]
- Contract value: [total]
- Key personnel: [names and roles]

SCOPE OF WORK:
[Describe the work performed — be specific about services delivered,
systems supported, outcomes achieved]

PERFORMANCE METRICS:
[List quantifiable results: uptime percentages, tickets resolved,
cost savings delivered, schedule performance]

CPARS RATINGS:
[List ratings by category if available: Quality, Schedule, Cost
Control, Management, Regulatory Compliance]

RELEVANCE TO CURRENT OPPORTUNITY:
[Explain specifically how this past performance demonstrates
capability for the target contract]

Write the narrative in the exact format specified by Section L.
Emphasise the aspects of this past performance that are most
relevant to the evaluation criteria. Include specific metrics
and outcomes, not generic claims of success.

When you maintain a past performance library in your Capture Brain, generating a tailored narrative for any new proposal becomes a 15-minute task instead of a 4-hour rewrite.

Colour team review preparation

Colour team reviews — Pink Team, Red Team, and Gold Team — are the quality gates that separate winning proposals from losing ones. AI does not replace these reviews. It makes them more effective by doing the preparation work that reviewers rarely have time to do.

Pink Team (outline and strategy review):

AI contribution: generate a crosswalk document that maps every Section L requirement to the proposed response outline. Reviewers can verify completeness by checking the crosswalk rather than reading the entire RFP alongside the outline. AI can also flag where the outline deviates from the evaluation weighting — if Section M assigns 40% weight to technical approach but your outline allocates 25% of page count to it, that is a problem worth catching at Pink Team.

Red Team (full draft review):

AI contribution: run the draft proposal against the compliance matrix and produce a gap analysis. For each requirement in the matrix, AI checks whether the draft addresses it and flags any requirement that is not clearly addressed with a specific section and page reference. This gives Red Team reviewers a head start — instead of hunting for compliance gaps, they can focus on evaluating the quality, persuasiveness, and technical credibility of the response.

Additionally, AI can score the draft against Section M evaluation criteria and provide an assessment of strengths and weaknesses from an evaluator's perspective. This is not a replacement for experienced Red Team reviewers, but it is a useful pre-screen that helps reviewers focus their limited time on the most critical issues.

Gold Team (final review and pricing):

AI contribution: generate an executive summary of how the proposal addresses each evaluation factor, with a risk assessment of areas that may score lower. For the pricing volume, AI can draft pricing narratives and basis of estimate descriptions (the narrative components — actual pricing remains in your pricing tool and spreadsheets). AI can also cross-check that the technical approach and the pricing volume are consistent — that every position in the staffing plan appears in the pricing, and every tool or system referenced in the technical approach is reflected in the cost estimate.

Price volume narrative support

To be clear: AI does not build your pricing. Your cost model, labour rates, indirect rates, fee structure, and price-to-win calculations belong in your pricing tool (whether that is a spreadsheet, PROPRICER, or a custom model). Pricing data is sensitive and should not be entered into AI tools without careful data handling consideration.

What AI does handle is the narrative components of the pricing volume.

Most best-value procurements require a cost/price narrative that explains your pricing approach: why your rates are fair and reasonable, how your cost model supports the technical approach, what basis of estimate supports your proposed level of effort, and how your pricing strategy delivers value to the government.

These narratives follow predictable patterns and can be drafted by AI when given the right inputs:

• Your pricing strategy (competitive, value-based, LPTA-driven)
• The basis of estimate methodology (historical data, engineering estimates, analogous contracts)
• Key cost assumptions (labour categories, travel estimates, ODC projections)
• The relationship between your technical approach and your cost structure

AI drafts the narrative framework. Your pricing team and contracts staff fill in the specific numbers and rate justifications. The result is a cohesive cost narrative that ties directly to your technical volume — a connection that many proposals fail to make clearly.