Hardest Part

Consistently extracting and correlating fragmented claims across 100-plus page unstructured PDFs against strict, granular compliance matrices without hallucinating compliance where none exists.

Min Viable Scope

Focus exclusively on binary compliance matrix checking for standard commercial RFPs, identifying whether specific required sections, clauses, and certifications are present. Deliberately leave out qualitative grading, pricing analysis, and multi-vendor rank-ordering.

Cold Start Problem

You need complex, real-world RFPs alongside historically scored proposals to calibrate the evaluation logic. Break this by partnering with one mid-market prime contractor to ingest their archived bids and baseline the AI against their actual past human-graded scoring sheets.

Time To First Value

1 week to map a specific organization's scoring rubrics and ingest the first active RFP, followed by minutes to process the incoming proposals.

Data Moat Available

true

Technical Difficulty

High

Wedge

The initial wedge is state-level IT procurement contracts where RFPs are highly technical and vendor responses contain massive boilerplate. This niche provides acute pain and fast proof by automating the initial pass-fail compliance checks for mandatory requirements. Expansion occurs by introducing qualitative scoring for technical approaches and subsequently moving horizontally into municipal and federal procurement markets.

Timing

Million-token context windows now allow entire vendor proposal packages, including annexes and technical specifications, to be processed simultaneously for cross-referencing against strict grading rubrics without the data loss previously caused by document chunking.

Why This ICP

State and local government procurement teams face strict statutory deadlines for RFP evaluations, chronically understaffed departments, and highly standardized rubric-driven evaluation criteria, making them ideal and motivated early adopters.

Size Of Prize

Approximately 30,000 enterprise procurement teams and government purchasing agencies globally spend an average of $50,000 annually on internal labor and external consultants specifically for proposal evaluation, representing a $1.5B addressable market.

Gap Narrative

Procurement teams and grant officers spend hundreds of manual hours scoring dense vendor responses against complex compliance matrices and evaluation rubrics. Existing software only manages the intake and routing workflow but leaves the cognitive burden of reading, comparing, and scoring unstructured technical documents entirely to human evaluators.

Defensibility

The product builds defensibility through deep workflow lock-in and the accumulation of agency-specific scoring data. As the system processes an organization's evaluations, it calibrates to their specific grading nuances and risk tolerances, creating a high switching cost for procurement teams who become dependent on the calibrated institutional memory.

Why This Thesis

A Service-as-Software approach fits perfectly because procurement officers require a completed audit-ready evaluation matrix with cited evidence rather than a conversational copilot, requiring the system to autonomously map vendor claims to rubric requirements.

SAM

~$1B-2B focusing on North American and European enterprises with high-frequency RFP cycles

SOM

~$20M-50M

TAM

~100,000 global enterprise procurement departments × ~$50,000/yr ≈ ~$5B

Growth Rate

~12-18%/yr, driven by rising vendor compliance complexity and enterprise mandates to reduce procurement cycle times

Paid Comparable Spend

~$100,000-250,000/yr per enterprise spent on manual procurement analyst hours, external sourcing consultants, and basic RFP intake software

Kill Thresholds

• Score override rate > 25% after 30 days of usage
• Information security rejection rate > 40% in the pipeline
• Time spent editing > 2 hours per proposal
• Zero conversions to $50,000 paid tier within 90 days

Leading Metrics

• Time-to-first-scorecard from document upload
• Score override percentage per vendor proposal
• Number of RFPs processed per account per month
• Time spent editing system-generated rationales

What Proves Right

Procurement teams upload vendor proposals and accept the system-generated scoring matrix for at least 85% of criteria without manual correction. Pilot customers transition to $50,000 annual contracts within 90 days of deployment. Analysts evaluate 3 or more RFP cycles per month exclusively using the platform.

What Proves Wrong

Users spend more time editing the extracted compliance tables than they previously spent reading the raw PDF submissions. Information security teams reject the architecture during the sales process due to concerns about ingesting confidential vendor pricing data. Evaluators abandon the platform after the first RFP because the generated rationales lack specific page citations.