Bid evidence graph for infrastructure contractors that assembles compliant tender answers from past projects, teams, and partner docs.

1. Win-rate improvement is becoming a budget line item for bid teams, not just a soft productivity promise, which makes new software easier to justify.
2. Tender operations are still stuck in spreadsheets and shared drives, so the workflow remains structurally open for a purpose-built system of record.
3. Construction and engineering already appear as named early buyer segments, giving a sharp wedge where tender complexity is high and outcomes are measurable.
4. The category can start in public-sector tenders and expand into private-contract bidding while using the same evidence graph and workflow engine.

Catalyst. BidScript's reported up-to-50% win-rate lift and Tech.eu's note that teams still run bids through spreadsheets and shared drives make 2026 the moment when tender ops can shift from artisanal process work to software.

Bid Evidence Graph ingests prior submissions, project case studies, CV packs, HSEQ certificates, and partner documentation into a requirement-tagged evidence graph. When a new tender arrives, it parses evaluation criteria, maps each question to reusable proof, and flags missing evidence before the team burns estimator hours. The product gives a bid lead a go/no-go score, owner-by-owner workflow, and red-team checklist, then exports a compliant response package with citations back to validated source material. After submission, it records win/loss feedback so each tender improves the next one.

What's different. Most AI bid products will draft prose on top of messy files. This company wins by owning the evidence layer: which project, person, certification, and partner can answer each scorer criterion and how often that proof correlates with wins. Over time the product turns every submission and evaluator outcome into proprietary win-loss data that improves templates, partner recommendations, and go/no-go decisions.

Jobs to be done

flowchart LR
  Buyer[Bid Director] --> Pain[Manual tender evidence scramble]
  Pain --> Product[Bid Evidence Graph]
  Product --> Outcome[Higher win rates and lower bid cost]

Executive takeaways

• The UK beachhead is real but not huge: a few hundred tender-heavy civil/infrastructure contractors support a modeled SAM around $19.5M at ~$65k ACV, so the product must win on evidence reuse, qualification, and compliance rather than prose alone. [3][4][30][73]
• Procurement reform is a tailwind for structured evidence, not just writing speed: suppliers now face reusable core data, clearer participation criteria, assessment summaries, and persistent social-value burden. [10][19][20][21][22][25]
• The hardest problem is cross-functional evidence assembly under deadline; that is why vendors are moving from simple drafting into go/no-go, compliance, and knowledge-grounding workflows. [30][34][46][77][83]
• Generic RFP automation is crowded, but a contractor-native evidence layer tied to NEC/JCT, frameworks, and subcontractor proof still looks differentiated in UK infrastructure. [31][35][36][45][58][92]
• Buyers will pay meaningful annual contracts when the tool cuts non-compliance and SME time, but archive cleanup and process change can still delay ROI. [27][41][73][90]

Market definition

This is UK infrastructure tender evidence management: a layer between document storage and proposal writing that handles go/no-go, requirement parsing, evidence reuse, compliance checking, and portal-ready submission. [30][31][34][58][77]

Customer and buyer

The daily user is a bid manager or preconstruction lead coordinating SMEs, project evidence, CVs, certifications, and partner documents; the buyer is commercial/preconstruction leadership that cares about win rate, bid cost, and compliance failures. [30][34][46][111]

Buying triggers

• A new framework tender or annual rebid forces the team to refresh PQQs, method statements, social-value narratives, and supporting evidence on a fixed deadline. [10][19][25][31]
• Higher tender volume or a fuller public-infrastructure pipeline makes manual coordination and document hunting visibly more expensive. [4][30][46]
• Recent clarifications, near-misses, or disqualifications around attachments, portal rules, and non-compliance create an urgent budget case. [21][22][34]
• Growth or multi-office expansion fragments knowledge ownership and makes stale content libraries more painful. [46][50][90]

Willingness to pay

Category pricing already spans $5,000/year starter packages and $899-$1,299/month AI-native plans, while BidScript sells custom per-seat enterprise scopes. Budget exists when software replaces manual labor and protects win probability, but buyers still ask for measured ROI before signing larger annual contracts. [27][41][73][83]

Category dynamics

Growth signal 5.6% YoY growth in GB construction new orders in 2024

Validation signals

• BidScript already markets UK construction/public-sector workflows and independent coverage reports early customers seeing win-rate improvements of up to 50%.
• Responsive/APMP benchmarking says teams with dedicated RFP software submit materially more proposals and otherwise juggle too many disconnected tools.
• AutoRFP’s public pricing, go/no-go, and gap-analysis modules show buyers will pay for decision quality and compliance analytics—not just answer drafting.
• Constructionline’s social-value positioning confirms non-price evidence burdens remain active in public-sector supply chains.

Regulatory & technical constraints

• Conditions of participation can only test proportionate legal/financial capacity and technical ability, so supplier-level evidence has to be organized separately from answer prose.
• Assessment summaries make it easier for suppliers to understand how their tenders scored, which increases the value of requirement-level traceability and audit-ready notes.
• The central digital platform reduces repeated data entry but raises the bar for clean, reusable core supplier information across procurements.
• Social-value expectations expand the amount of supporting evidence required beyond price and technical method statements.
• High-stakes tender AI needs grounded retrieval and human review to align with trustworthy-AI risk practices.

The field has split into legacy/scale SRM suites (Responsive, Qvidian), AI-native proposal writers/extractors (AutogenAI, Arphie, AutoRFP), and vertical tender platforms like BidScript. The open gap is a contractor-native evidence layer tied to frameworks, NEC/JCT, subcontractor packs, and requirement-level proof rather than generic questionnaire automation. [35][36][45][58][73][90][92][109]

Why incumbents do not win by default

• Generic cloud/document suites. Repositories store files but do not interpret tender requirements, map evidence to scoring criteria, or produce assessment-ready compliance trails; teams still chase SMEs and stale content.
• Proposal-management incumbents. Responsive and Qvidian are strong on governed libraries and workflows, but their value is highest where teams can maintain centralized content and heavier process discipline; that weakens fit for mid-market contractors with messy archives.
• AI drafting specialists. AutogenAI speeds writing and extraction, but its positioning still centers on proposal writing, qualification, and capture rather than a contractor-specific evidence graph spanning frameworks, CV packs, and partner credentials.
• AI-native questionnaire tools. AutoRFP and Arphie prove AI can automate go/no-go, gap analysis, and auditable answers, but they remain broad cross-industry response platforms rather than UK infrastructure tender systems of record.

UK civil and infrastructure contractors still run tender evidence assembly through shared drives, spreadsheets, Word libraries, and consultants, even as procurement reform increases the need for reusable supplier data, traceable assessment summaries, and social-value evidence. The proposed company sells a contractor-native bid evidence graph that ingests prior submissions, project case studies, CV packs, HSEQ certificates, and subcontractor documents, then maps them to requirement-level answers, gap flags, and compliance workflows for one public-sector tender type at a time. The first customer is a 150-person UK civil contractor bidding 40-60 council, utilities, and framework tenders per year, where a live rebid or new framework deadline makes document hunting and non-compliance risk immediately budgetable. Go-to-market starts with a paid evidence audit and archive onboarding, then converts into a £30k-£75k annual subscription priced by bid desk and tender volume if the product cuts SME search time, submission defects, and time to compliant first draft. The wedge is deliberately narrower than generic RFP automation because the researched SAM is only about $19.5M and the company needs proof that one contractor ontology, one rule pack, and one implementation motion repeat before expanding into adjacent sectors or geographies. The defensible angle is not better prose generation but a reusable mapping among tender clauses, past proof, subcontractor credentials, and eventual assessment outcomes that generic content libraries do not capture cleanly. The biggest disconfirming risk is that archive cleanup and change management remain too services-heavy, which would let BidScript, Responsive, consultants, or internal SharePoint workflows stay good enough. Market size and some ROI claims remain modeled or vendor-reported rather than independently benchmarked, so this pre-seed plan should be judged on whether the first five pilots convert into repeatable onboarding, measurable compliance gains, and subscriptions above point-tool price levels.

Problem

• Bid teams lose days hunting for case studies, CV packs, HSEQ certificates, and subcontractor proof across shared drives, Word libraries, and email every time a live framework or project tender lands.
• Public-sector construction bids now require reusable supplier data, social-value evidence, and audit-ready compliance, so manual reuse creates higher disqualification risk and inconsistent scoring.

Solution

• Ingest prior submissions, project records, CV packs, certificates, and partner documents into a requirement-tagged evidence graph for one UK civil tender workflow.
• Parse each new tender into go/no-go signals, source-linked draft answers, evidence gap flags, and reviewer workflows that export a compliant submission package without replacing existing storage.

Why we win

• The product is built around contractor evidence objects and UK infrastructure rule packs rather than generic proposal text, which makes it harder for horizontal RFP tools to match proof depth.
• Each live bid adds requirement-level mappings, assessment feedback, and win/loss data that improve qualification, reuse, and partner-document coverage over time.

Milestones

flowchart LR
  Wedge[UK civil tender wedge] --> MVP[Source linked evidence graph]
  MVP --> Proof[Faster compliant drafts fewer defects better shortlist rate]
  Proof --> Expansion[More bid desks partner workspaces and adjacent tender modules]

Founding team

Experiment roadmap

Risk assessment

What must be true

• At least 3 of the first 5 design partners must share a narrow evidence ontology that can be onboarded in 30 days or less.
• Live pilots must reduce SME search and drafting time by at least 25% on one real public-sector tender.
• Commercial buyers must convert at least half of paid pilots into annual contracts above £30k ARR within 6 months.
• Assessment outcomes or internal QA must show fewer compliance misses than the customer's prior manual process.
• Expansion into extra bid desks, partner workspaces, or adjacent tender modules must lift ACV beyond roughly $65k without custom services exceeding 30% of delivery effort.

Open diligence questions

• How many UK 50-500 employee civil contractors truly run 20-150 relevant tenders per year and own a software budget centrally?
• What percent of archive-ingestion work can be standardized across CV packs, HSEQ certificates, case studies, and subcontractor documents?
• Will buyers pay for fewer compliance errors and lower SME hours before they see independently measured win-rate lift?
• Can the product coexist with SharePoint, consultants, and current bid libraries, or must it replace them to deliver ROI?
• How different are rule packs and evidence needs across highways, utilities, local-authority, and framework tenders?

Model sanity

• Revenue engine. Base-case revenue is driven by subscription ACV stepping from $45K (Year 1 pilot pricing) to $65K (Year 3 blended) across a customer base that scales from 3 to 30 logos, plus one-time onboarding fees for each newly signed design partner.
• Must go right. The company must convert at least 2 of the first 3 paid design partners into annual subscriptions in Year 1 and keep onboarding partner/template-led so gross margin can climb from 55% toward the BP's 70% target instead of getting stuck in services-heavy delivery.
• Model breaks if. The Q4Y2-to-Q4Y3 jump from 8 to 30 customers slips or gross margin stalls near 64%, which the sensitivity table shows could cut Year 3 revenue by roughly $200K+ and pull the cash trough down toward the tighter $1.0-1.3M minimum-burn estimate.
• Next-round proof. The seed case is strongest once the company reaches 5-7 subscription customers with at least one partner-sourced pilot by Q4Y2 and begins posting EBITDA-positive quarters in Year 3, as shown in Q3Y3-Q4Y3.

Scenarios

Sensitivity

flowchart LR
  Leads[Bid director and preconstruction leads] --> Audits[Paid evidence audits]
  Audits --> Pilots[Design-partner onboarding, 8-12 weeks]
  Pilots --> Subscriptions[Annual bid-desk subscriptions]
  Subscriptions --> Revenue[Blended ARPU x subscription customers]
  Revenue --> GrossProfit[Gross profit after delivery cost]
  GrossProfit --> Cash[Cash runway toward next round]

Flags: Year 3 customer count jumps from 8 (Q4Y2) to 30 (Q4Y3), a roughly 3.75x logo increase in a single year that mirrors the BP's own 24-36 month milestone but is far more aggressive than the hiring-constrained Year 1-Year 2 pace; this is the single biggest execution risk in the model. · Rule-of-40 and YoY growth % are distorted by a very small Year 2 revenue base ($503.8K) and should not be benchmarked against mature SaaS comparables. · Gross margin recovery from 55% (Y1) to 70% (Y3) assumes onboarding standardizes largely as hoped; the BP's own risk section warns archive ingestion could 'remain too bespoke,' in which case both gross margin and the Year 3 customer ramp are optimistic. · Revenue per FTE ($150.2K) stays below the $200-400K mature SaaS benchmark through Year 3, so headcount efficiency at scale is not yet proven. · The $2.5M funding ask leaves a modeled cash trough of ~$1.42M, comfortably above zero; this is sized to the BP's own $2-4M/18-month guidance rather than the tighter ~$1.0-1.3M minimum this cost model implies is strictly required, with the gap acting as a buffer against slower sales cycles or heavier services costs than modeled here.

• ROI proof gap. Buyers may discount the up-to-50% win-rate lift because the sources attribute it to company-reported customer results rather than independent benchmarking. Mitigation: Launch with baseline imports and tender-by-tender outcome analytics so early customers can measure win-rate and bid-cost improvement against their own history.
• Archive ingestion drag. Past bids, certifications, and partner materials are messy and inconsistent, which can make onboarding slow and delay time to value. Mitigation: Narrow the initial product to one tender type, offer concierge archive cleanup, and prioritize connectors for the folders and templates bid teams already use.
• Workflow adjacency. SharePoint consultants, bid-writing agencies, and construction software vendors can slow adoption by positioning the product as another layer on top of existing tools. Mitigation: Sell as an evidence and decision layer that plugs into current storage and makes consultants faster, with proof of lower bid cost per submission rather than a rip-and-replace story.