Why RFP Drafting Is Still a 40-Hour Job in Most Organisations
A typical Request for Proposal or tender document takes 40 to 120 hours to draft from scratch. That timeline covers requirements gathering, stakeholder alignment, drafting the scope of work, writing evaluation criteria, building scoring matrices, assembling terms and conditions, internal review cycles, and final revisions before issuance. For a category manager running multiple concurrent sourcing initiatives, that's 1-3 weeks of focused work per RFP—time that competes with supplier management, contract administration, and strategic sourcing activities.
The blank-page problem compounds the issue. A sourcing manager must structure the RFP mentally first: what exactly does the organisation need? How do we articulate technical requirements in a way vendors can respond to? What commercial terms matter most? How do we score responses fairly across technical capability, price, experience, and risk? That structural thinking happens in the first 10-15 hours—before a single word appears in the document.
Then there is the iterative review cycle. Stakeholders (CFO, operations, IT, legal, internal customers) each add requirements or language changes. These cycles can stretch a 40-hour effort to 80 or 120 hours. By the time an RFP is issued, it has passed through 4-6 hands and contains compromises that no single author would have chosen alone.
The consequence is real: organisations issue fewer RFPs, rely on incumbent suppliers longer, and miss competitive tendering opportunities. Or they rush RFPs, creating ambiguous requirements that generate excessive supplier questions and extend tender timelines further.
Where GenAI Genuinely Accelerates RFP Writing
Generative AI (Claude, ChatGPT, Microsoft Copilot) accelerates RFP drafting not by replacing the thinking, but by removing the blank-page paralysis and structure-building overhead. Here's where the time gains actually come from:
First Draft Generation (4-8 hours saved). Instead of beginning with a blank document, you start with a GenAI-generated RFP outline and 60-70% complete sections. The AI understands RFP structure—executive summary, scope of work, timeline, evaluation criteria, terms and conditions—and populates each section with plausible, coherent text based on your category context. A category manager can then edit downward (removing irrelevant sections, refining language) far faster than drafting upward from nothing.
Requirements Structuring (3-5 hours saved). GenAI excels at turning loose requirement conversations into structured, specific deliverables. Provide the AI with your category notes, stakeholder feedback, and a list of 8-10 business needs. The AI will structure those into a detailed Scope of Work with phased deliverables, effort allocation, and success criteria. That structure is 70-80% correct and usually only needs factual adjustments, not conceptual rework.
Evaluation Criteria and Scoring Matrices (2-4 hours saved). Building a fair, weighted evaluation scorecard is often the most tedious part of RFP drafting. GenAI can generate a complete evaluation framework with weighting (technical 40%, price 30%, experience 20%, references 10%), detailed rubrics for each criterion, and a scoring template. Again, you adjust the weightings and criteria language, but the skeleton is complete.
Terms and Conditions Drafting (3-6 hours saved). Standard T&C clauses—confidentiality, intellectual property, liability limits, insurance requirements—can be drafted by GenAI using procurement-standard language. This is not a legal opinion, and internal legal review is mandatory, but it eliminates the need to start from a blank T&C section or hunt for a template.
Parallel Drafting (2-3 hours saved). With GenAI, multiple sections can be drafted in parallel. While you refine the Scope of Work, a colleague can request the AI generate Evaluation Criteria. That parallelism collapses the serial dependency of traditional RFP writing, where one section often blocks another.
In aggregate, organisations moving to GenAI-assisted RFP drafting report 40-65% time reduction. A 120-hour RFP becomes a 42-50-hour effort: 2-4 hours for initial context gathering, 2-4 hours for GenAI output generation and initial edits, 3-5 hours for cross-functional review, 2-3 hours for legal and compliance review, 2-3 hours for final polish. The remaining weeks of traditional RFP timelines are now available for earlier supplier engagement, longer evaluation windows, or other sourcing work.
The RFP Drafting Workflow with GenAI: Step by Step
Phase 1: Preparation and Context Assembly (1-2 hours). Before opening GenAI, gather all input materials: the business case or sourcing strategy document, list of requirements from stakeholders, budget range, timeline expectations, existing contracts (for comparison), and any prior RFPs in related categories. If your organisation uses a sourcing platform like GEP SMART or Jaggaer, export any existing requirement templates or evaluation frameworks. This material becomes the context you provide to the AI—the better your input context, the better the output.
Create a simple briefing document (a few bullets or paragraphs) that answers: What is the category (IT, logistics, facilities management, professional services)? What is the total contract value? What is the timeline for evaluation and award? What are the 5-8 most critical requirements? Who are the internal stakeholders who will review the RFP? This briefing becomes your system prompt or opening context.
Phase 2: GenAI Section-by-Section Generation (2-4 hours). Rather than requesting the AI generate a complete RFP in one go (which often results in generic or overly long output), use section-focused prompts. Request each major section separately:
- Executive Summary. "Draft an executive summary for a professional services RFP. The organisation seeks a consulting firm to lead digital transformation across three business units over 18 months. Budget: $1.2M. Key deliverables: strategy framework, implementation roadmap, change management plan, and staff training. Bidders should have experience in financial services and enterprise process redesign."
- Background and Business Context. "Write a background section that explains: the current state of our order-to-cash process, pain points (30-day DSO, manual invoice reconciliation), and why we're sourcing external expertise. Assume the reader is a consulting firm with big-4 experience but unfamiliar with our specific company."
- Scope of Work. "Create a detailed scope of work for the above engagement broken into four phases: assessment and diagnosis (3 months), roadmap and design (4 months), implementation and training (8 months), post-go-live support (3 months). For each phase, outline key activities, deliverables, and the expected level of effort."
- Evaluation Criteria. "Build an evaluation framework with criteria: technical approach and methodology (35%), proposed team experience and references (30%), pricing and value (20%), implementation timeline and resource commitment (15%). For each criterion, provide a detailed scoring rubric (1-5 scale) explaining what constitutes a score of 1 (poor) through 5 (excellent)."
- Instructions to Bidders. "Draft instructions for bidders on how to respond to this RFP, including format requirements (response length, sections to include), submission deadline, selection process timeline, and contact person for questions."
- Terms and Conditions. "Draft standard T&C clauses for a professional services engagement: confidentiality obligations, intellectual property ownership, limitation of liability, insurance requirements, indemnification, and termination for convenience. Assume this will be reviewed by in-house legal."
For each request, copy-paste the relevant context from your briefing document so the AI has fresh context. Review each section immediately for factual accuracy, clarity, and alignment with your sourcing strategy.
Phase 3: Integration and Reconciliation (1-2 hours). Copy all generated sections into your RFP document template. Ensure consistency across sections: if the Scope of Work lists four deliverables, confirm the Evaluation Criteria address those deliverables, and the Pricing Schedule includes line items for each. Look for repeated language that can be consolidated. Check that timeline mentions (Phase 1 = 3 months, Phase 2 = 4 months) are consistent throughout.
Phase 4: Fact-Check and Legal Review (1-2 hours). Read the complete draft yourself, focusing on accuracy. GenAI is prone to "hallucination"—plausible-sounding but incorrect details. Verify that all timeline and budget numbers match your sourcing brief. Confirm that the Scope of Work reflects actual business needs, not generic consulting platitudes. Remove any vague or overwritten language.
Pass the draft to your internal legal and compliance contacts. They will check that T&C language aligns with your organisation's standard terms, that there are no unintended liabilities, and that the document complies with any regulatory or procurement policy requirements. This step typically reveals 1-3 substantive changes, not structural revisions.
Phase 5: Stakeholder Review and Revision (1-2 hours). Share the draft with your core stakeholder group (usually 3-4 people: the category owner, a budget holder, and a key end-user from the business). Use a shared comment document so stakeholders add input simultaneously, not serially. Most feedback at this stage is refinement: clarifying a requirement, expanding a criterion, or adjusting timeline expectations. Consolidate feedback, make edits, and circulate a revised version for final sign-off.
Phase 6: Polish and Issuance (30-45 minutes). Do a final formatting pass: ensure all headers, numbering, and cross-references are correct. Proofread for typos. Confirm the evaluation scoring template is clear and unambiguous. Then issue the RFP through your sourcing platform (if you have one) or via email, ensuring all suppliers in your target list receive it simultaneously.
Prompt Templates for Different RFP Types
Template 1: IT Services RFP. "We are issuing an RFP for [specific technology project]. Context: our organisation is a [industry, size] company with [current technology stack brief]. The project scope is [problem statement and desired end state]. Budget: $[X]. Timeline: [Y months]. Success criteria: [list 3-5 outcomes]. Key requirements: [list 5-8]. Please draft [Scope of Work / Evaluation Criteria / Executive Summary] that reflects this context, emphasizes [methodology or risk mitigation], and is suitable for bidders with [desired experience profile]."
Template 2: Professional Services RFP (Management Consulting, Legal, Accounting). "We are seeking a [service type] firm to support our [business need]. Our organisation is [size, industry, complexity]. Current state: [problem or capability gap]. Desired outcome: [target state]. Timeline: [project duration]. Budget: $[X]. Key success factors: [list 3-5]. Bidders should have experience in [specific vertical or service type]. Please draft [section name] with language appropriate for [consulting / law / accounting] firms and emphasising [quality, cost-efficiency, or speed]."
Template 3: Logistics and Logistics Services RFP. "We are procuring [logistics service: warehousing, transportation, 3PL, etc.] for [product category]. Service scope: [pick-pack-ship volumes, geography, special requirements]. Current volumes: [units/month or dollars/year]. SLA requirements: [delivery windows, accuracy targets]. Geographic coverage: [regions or locations]. Please draft [section name] with specific KPI and SLA language, including penalty clauses for [specific failures], and emphasising [cost, speed, or compliance]."
Template 4: Facilities Management RFP. "We are procuring [facilities services: cleaning, maintenance, security, catering, etc.] across [number] buildings totalling [square footage] in [geography]. Current contracted costs: $[X/year]. Service standards: [response times, coverage hours, quality levels]. Key requirements: [sustainability standards, reporting, staffing credentials]. Budget: $[X]. Please draft [section name] with detailed SLA language, performance metrics, and staffing/credentialing requirements."
For any template: adjust the bracketed [items] to match your context, then copy-paste the entire prompt into your GenAI tool. The more specific your context, the more relevant the output. Vague prompts (e.g., "draft an RFP for consulting services") generate boilerplate. Detailed prompts (with budget, timeline, specific requirements, and industry context) generate output that requires only light editing.
Generating Evaluation Criteria and Scoring Matrices with AI
Evaluation criteria are the skeleton of a fair RFP process. They define what you're actually scoring suppliers on, in what proportion, and with what definition of "good." GenAI can generate complete evaluation frameworks, though the weighting and priorities should reflect your sourcing strategy, not the AI's defaults.
Sample Prompt for Evaluation Criteria: "Build an evaluation framework for a supply chain consulting engagement. Scoring categories and weights: (1) Technical Approach and Methodology (35%)—does the proposal show deep understanding of our business and a well-structured approach? (2) Team Experience and References (30%)—do the proposed team members have relevant experience, and can they provide strong references? (3) Pricing and Value (20%)—is the price competitive and the deliverable quality justified? (4) Timeline and Resource Commitment (15%)—can they deliver on schedule with adequate team depth? For each category, provide a 1-5 scoring rubric. A score of 1 is unacceptable, 3 is adequate, 5 is excellent."
The GenAI output will typically look like this:
Technical Approach and Methodology (35% weight):
Score 1: Proposal shows minimal understanding of the organisation; approach is generic or misaligned with stated objectives.
Score 3: Proposal demonstrates adequate understanding; approach is sound and includes key project phases, but lacks differentiation or deep industry insight.
Score 5: Proposal shows exceptional understanding of the organisation's context; approach is innovative, well-structured, addresses specific pain points, and includes risk mitigation strategies.
Team Experience and References (30% weight):
Score 1: Proposed team has no relevant experience; references are not provided or are weak.
Score 3: Proposed team has relevant experience; references are provided but are lukewarm or not recent.
Score 5: Proposed team includes recognised experts in the relevant domain; references are exceptional and recent (within 12 months); team has deep experience in your specific industry or use case.
And so on.
Once generated, review for realism: does the rubric distinguish between acceptable and excellent responses? Are the weights aligned with your real priorities? (A common error: procurement weights "price" heavily, but the business cares more about "technical approach." Ensure your weights reflect business priorities, not just financial ones.) Then use this evaluation framework consistently across all supplier scoring.
Terms and Conditions: What GenAI Can and Cannot Draft
GenAI can draft initial versions of standard T&C clauses, including: confidentiality and NDA language, intellectual property ownership and licensing, liability limits and indemnification, insurance and bonding requirements, termination rights and notice periods, and payment and invoicing terms. These are structural, non-negotiable clauses that appear in most commercial contracts.
GenAI struggles with T&C clauses that are organisation-specific, legally complex, or dependent on your jurisdiction. Examples of high-risk GenAI T&C output: liability caps that are too low (creating unacceptable risk), IP ownership language that conflicts with your industry norms, or indemnification clauses that don't address your specific exposures. GenAI will not know your jurisdiction's employment law, data privacy regulations (GDPR, CCPA), or industry-specific compliance requirements (healthcare, financial services, etc.).
Safe Use of GenAI for T&C: Request GenAI generate initial language with the prompt: "Draft standard terms and conditions for a [service type] contract between our organisation and an external [supplier type]. Include clauses on confidentiality, IP ownership, limitation of liability (standard commercial terms, not excessive), insurance requirements ($[X] minimum coverage), and termination. Assume this will be reviewed by in-house legal; draft in formal but accessible language." Then immediately mark those sections for legal review. Your legal contact will adjust or replace as needed based on your organisation's standard terms and jurisdiction.
What Not to Put in Public LLMs: Confidentiality Rules
Before you paste any RFP section (or sourcing context) into ChatGPT, Claude, or any public GenAI tool, check whether it contains confidential information. If it does, do not use a public LLM. Instead, use an on-premises or enterprise-grade AI tool, or redact sensitive details first.
What Not to Include: Strategic sourcing plans or supplier strategies ("we want to consolidate with one global 3PL and exit our current three suppliers"). Pricing or cost data (current contract spend, budget limits). Proprietary product roadmaps or technology specifics if they could identify your competitive advantage. Specific customer names or contract volumes. Financial forecasts or growth plans that are not public. Security, compliance, or operational details (data center locations, system architecture, cybersecurity controls). Employee names or organization charts.
What Is Safe to Include: Vague business context ("we are a mid-market financial services firm"). Generic category descriptions ("professional services consulting for process improvement"). Publicly available business information. Redacted or anonymised requirements ("we need to move 100,000 units per month" rather than "our Q2 forecast is 118,000 units"). General industry benchmarks or market research.
Practical Approach: If you have confidential details essential to the RFP (e.g., specific customer SLA requirements), redact them before pasting. Replace "Customer Acme Corp requires 99.99% uptime" with "One major customer requires 99.99% uptime." That change preserves the requirement without exposing competitive information. If redaction isn't practical, use an enterprise GenAI tool (Claude for enterprise, GPT-4 with enterprise access, or on-premises alternatives) where data is not retained or used to train models.
For details on GenAI governance and security policies for procurement teams, refer to our governance guide.
AI-Native Sourcing Platforms: GEP, Jaggaer, and Others
If your organisation uses a sourcing platform, it may have built-in GenAI capabilities designed specifically for RFP drafting. These tools offer advantages over public ChatGPT or Claude:
GEP SMART AI. GEP's platform includes an AI assistant that can generate RFP sections, evaluation criteria, and scoring templates directly within the SMART interface. The advantage: your RFP context stays within GEP's system, not transmitted to a public LLM. The AI has been trained on thousands of GEP RFPs, so it understands procurement structure and sourcing best practices. Learn more about GEP SMART AI.
Jaggaer. Jaggaer's sourcing module includes AI-assisted RFP generation, proposal evaluation, and scoring. Like GEP, it keeps sensitive sourcing data on-premises. Learn more about Jaggaer sourcing AI.
Coupa, Ariba, and Others. Most major sourcing platforms have added GenAI features in 2025-2026. If you use one of these, explore whether your platform includes GenAI-assisted drafting. If so, use it—the confidentiality and integration benefits outweigh using a public LLM.
If you do not have a sourcing platform or your platform does not offer GenAI assistance, use a public tool (Claude, ChatGPT, Copilot) but treat all output as a first draft, always subject to review and editing.
Quality Control Checklist Before Issuing an AI-Drafted RFP
Before you issue any RFP that involved GenAI drafting, work through this checklist. It typically takes 30-45 minutes and prevents common problems:
- Factual Accuracy: All numbers (budget, timeline, volumes, SLA targets) match your sourcing brief and internal forecasts. No "hallucinated" metrics or assumptions.
- Scope Clarity: The Scope of Work is specific enough that a qualified supplier can estimate effort and provide a firm price. Avoid vague language ("support the transformation") in favor of concrete deliverables ("deliver implementation roadmap, configuration guide, and staff training plan").
- Evaluation Criteria Alignment: The evaluation criteria and weighting reflect your actual priorities. Do the weights align with what the business really cares about? (E.g., if cost is secondary to implementation speed, is your weighting 20% price and 30% timeline? Or did you leave price at 40%?)
- No Contradictions: Scope of Work, timeline, and budget are internally consistent. If Phase 2 is listed as 4 months in the timeline but 8 months in the project plan, that's a contradiction that will trigger supplier questions or RFP addenda.
- Instructions to Bidders Are Complete: Suppliers know exactly how to respond: page limits, section requirements, format (PDF or online portal), submission deadline, and contact for questions.
- T&C Language Is Appropriate for Your Jurisdiction: Have your legal contact review all T&C clauses, especially liability, indemnification, and termination language. GenAI will not know your jurisdiction's norms.
- No Jargon or Overwritten Sections: GenAI sometimes generates verbose or buzzword-heavy language. Tighten any section that uses 5 words where 2 would do. Suppliers should not have to decode your RFP.
- Tone and Professionalism: The RFP reads as a professional, polished document from your organisation. No awkward GenAI phrasing or tone shifts between sections.
- Pricing Schedule Is Clear: If you want suppliers to submit cost proposals, your pricing template or schedule is explicit. Specify whether you want line-item costs, a fixed fee, time-and-materials, or cost per unit. Avoid ambiguity that leads to apples-to-oranges bids.
- Evaluation Process and Timeline Are Transparent: Suppliers know when you'll evaluate proposals, when you'll request clarifications, and when you'll announce the award. This transparency reduces post-issuance questions and improves supplier experience.
- Governance and Approvals Are Complete: The RFP has been reviewed and approved by all required stakeholders: procurement, legal, finance, end-users. No surprises after issuance.
This checklist is the final gate. If you check all boxes, you're ready to issue. If you find gaps, use your remaining time to correct them before suppliers see the document. Issuing a sloppy RFP costs far more in clarification cycles and extended timelines than an extra day of drafting.
Frequently Asked Questions
How do I know if GenAI output is hallucinating?
GenAI "hallucination" is plausible-sounding but incorrect information—a made-up metric, a feature that doesn't exist, or a timeline that's unrealistic. The most reliable detector is your own domain knowledge. Read GenAI output with a critical eye: does this match your business context? Do these numbers sound right? If something seems off, ask the AI follow-up questions or consult your own notes. For any claim in the RFP that a supplier might challenge (e.g., "we anticipate 20% growth in volumes"), verify it against your sourcing brief before including it.
Should we use the same GenAI prompt for multiple RFPs in similar categories?
Yes, with customisation. If you have a successful RFP for IT services, you can reuse that prompt structure for another IT project, but always update the specific details: budget, timeline, technology platform, and business context. Reusing a prompt without customisation produces generic output that may not fit your actual needs. Treat prompts as templates, not copy-paste jobs.
Can we issue a fully GenAI-drafted RFP without human review?
No. Always have a human (preferably the category manager or sourcing lead) review and edit GenAI output before issuance. The purpose of GenAI in RFP drafting is to accelerate the first draft and eliminate blank-page paralysis, not to automate away human judgment. Your role is to ensure the RFP accurately reflects your sourcing strategy, your actual business needs, and your organisation's procurement standards.
What if our legal team rejects most of the GenAI-drafted T&C language?
That's normal. GenAI drafts generic T&C language; your legal team will adjust for jurisdiction, industry norms, and your organisation's risk tolerance. Don't view rejected language as a failure of GenAI—view it as time saved by legal (they're not drafting from scratch) and as learning input for future RFPs. After a few cycles, your legal team will have adjusted GenAI-generated T&C to your standards, and subsequent RFPs will require fewer revisions.
Summary: From 3 Weeks to 3 Days
RFP drafting at 40-120 hours is a significant bottleneck in procurement. GenAI reduces that timeline to 15-30 hours of concentrated work, with the remaining time spent on review, legal approval, and stakeholder alignment—activities that add value and cannot be accelerated away. The shift from blank-page drafting to editing a GenAI-generated first draft is a meaningful change in how sourcing teams work and where they direct their attention.
Start with a moderate-complexity RFP in a familiar category. Use the prompt templates and workflow outlined above. Measure your actual time against prior RFPs. Iterate on your prompts based on output quality. Over time, your prompts will be more precise, your edits will be lighter, and your RFPs will be ready for issuance in days, not weeks. That time savings translates directly to faster sourcing cycles, earlier supplier engagement, and more competitive categories.
For deeper guidance on AI governance, security, and policy for procurement teams, see our procurement AI governance guide. For broader context on GenAI's impact on procurement, read our strategic sourcing AI articles.