Domain-Specific AI for Procurement: The Business Case

Executive Summary

Procurement is the largest cost-control function in most organisations, yet it remains one of the least served by modern technology. Procurement professionals rely on generic tools — spreadsheets, shared drives, and mass-market AI assistants — none of which understand the domain, the vocabulary, or the contextual nuance that makes procurement work.

Procurement Union is building PUC (the Procurement Union Companion): a domain-specific GenAI assistant fine-tuned for procurement professionals. PUC is grounded in the PUKHub — a community-contributed, anonymised procurement knowledge base — and personalised through a structured KYC profiling system that tailors every response to the user's role, industry, and seniority.

The product is delivered as a B2C SaaS subscription in three tiers (Free, Basic, Advanced), with a B2B enterprise on-premises offering on the roadmap. The architecture is fully on-premises by design, ensuring data sovereignty for enterprise customers without dependency on external cloud APIs.

Key metrics: Global procurement workforce estimated at 3–5 million professionals. Enterprise software spend in procurement exceeds $7B annually. Domain-specific AI adoption in procurement is nascent — first-mover advantage is significant.

1. Market Opportunity

1.1 The Size of the Problem

Procurement controls between 50% and 80% of total company expenditure in most industries. A 1% improvement in procurement efficiency translates directly to bottom-line impact. Yet the procurement technology landscape is fragmented, dominated by ERP bolt-ons and process management tools that address workflow — not intelligence.

The emergence of large language models (LLMs) has created a new category: AI-native procurement intelligence. No major player has yet established dominance in this space for the practitioner market. This is the window Procurement Union is entering.

1.2 Why Generic AI Fails Procurement

General-purpose AI tools (ChatGPT, Copilot, Gemini) are increasingly used by procurement professionals — but with poor outcomes. The core problems are structural:

· Domain gap: No domain knowledge of procurement terms, frameworks, or operational logic.

· Data gap: No access to real procurement benchmarks, supplier data, or category intelligence.

· Context gap: No personalisation by role, industry, seniority, or company maturity.

· Trust gap: Outputs cannot be trusted for commercial decisions - hallucination risk is high.

· Compliance gap: Sharing procurement data with public LLM APIs violates data governance policies.

2. The Product -> Kato

2.1 What Kato Does

Kato is an AI companion purpose-built for procurement professionals. It answers questions, drafts documents, retrieves benchmarks, assists with RFx design, evaluates supplier responses, and supports category strategy — all grounded in a curated, community-contributed knowledge base and personalised to the individual user.

Three capabilities define Kato's value proposition:

· Grounded intelligence: Answers are retrieved from the Theca — real procurement documents, anonymised and indexed. Not generic internet content.

· Personalised context: Every session is informed by the user's role, industry, seniority, and company size, captured through a structured KYC onboarding questionnaire.

· Data sovereignty: No data leaves the user's infrastructure. On-premises deployment by design.

2.2 User Journey

New users land on procurementunion.com and select a tier. Free users access basic chat immediately. Paid users complete a Stripe payment followed by a structured KYC questionnaire capturing: role title, industry vertical, seniority level, years of experience, and company size. This profile is stored and injected into PUC's system prompt at every session — ensuring responses are calibrated to the individual, not a generic procurement persona.

Example: "You are talking to a Senior Procurement Manager in Financial Services, 8 years experience, company spend >$500M. Prioritise strategic sourcing and risk management frameworks in your responses."

3. Theca -> The Knowledge Moat

3.1 What Makes Theca Different

Theca is Procurement Union's core defensible asset. It is a community-contributed, anonymised library of real procurement operational knowledge - structured around a five-pillar operating model and continuously growing with every subscriber contribution.

Unlike public procurement content (standards, frameworks, academic papers), Theca captures the institutional intelligence that lives inside organisations: real RFP structures, real category strategies, real supplier evaluation criteria, real savings methodologies. Anonymised. Indexed. Made available to every member.

3.2 The 5-Pillar StructurellarescriptionExample Content

1. Strategy & Governance (S&G) -> Procurement policy, operating models, governance frameworks -> Category charters, procurement policies, delegation matrices

2. Source to Contract (S2C) -> Sourcing events, tendering, negotiation, award -> RFP templates, evaluation scorecards, negotiation playbooks

3. Procure to Pay (P2P) -> Purchase orders, invoicing, three-way match, compliance -> PO workflows, invoice matching rules, maverick spend controls

4. Supplier & Risk Management (S&RM) -> SRM programmes, risk registers, due diligence -> Supplier scorecards, audit frameworks, risk registers

5. Analytics & Technology (APTE) -> Spend analytics, dashboards, transformation -> Spend cube templates, benchmark reports, maturity models

3.3 Community Quality Control

Every uploaded document passes through an automated quality pipeline on ingestion: procurement relevance scoring (TF-IDF, threshold ≥15%), pillar classification, and supplier/IP anonymisation. Documents below relevance threshold are quarantined for manual review. This ensures the PUKHub maintains quality as it scales.

3.4 The Public Library

Separate from the PUKHub, Procurement Union maintains a curated Public Library of hundreds of PDFs of publicly available procurement material: CIPS standards, IACCM frameworks, Deloitte/McKinsey procurement thought leadership, LinkedIn Influencers and global benchmarking reports. Browsable by Basic and Advanced subscribers.

4. Technology & Differentiators

4.1 Architecture Principles

· On-premises first: All inference, storage, and processing runs on dedicated on-premises hardware.

· Data sovereignty: No subscriber data or procurement content is transmitted to external APIs.

· Commercial licensing: Only MIT/Apache 2.0 licensed models and libraries throughout the stack.

· Modularity: Modular service architecture — each component is independently upgradeable.

4.2 LLM Stack

Kato currently runs on Gemma 3 27B (4-bit quantisation) via Ollama on dedicated GPU hardware (NVIDIA RTX 3090, 24GB VRAM). The RAG pipeline uses nomic-embed-text for embeddings (CPU-based), Qdrant as the vector database, and LlamaIndex as the orchestration layer. Post-generation faithfulness scoring validates that responses are grounded in retrieved context.

The architecture is LLM-agnostic. Falcon4 (TII UAE) is planned as a future upgrade, aligning with Procurement Union's UAE incorporation and the region's emerging AI ecosystem.

4.3 The UB2S Reasoning Framework

PUC's reasoning approach is informed by the UB2S Framework — Understanding Business, Suppliers, and Spend — a proprietary methodology developed by Procurement Union's founder, drawing on 20+ years of senior procurement experience. This framework governs how PUC structures its enquiry, challenges assumptions, and calibrates recommendations to the user's procurement maturity level.

About Procurement Union

Procurement Union is a GenAI startup incorporated in the UAE, founded by a procurement professional with 20+ years of senior procurement leadership experience across multiple industries and geographies. The company is building the AI companion that procurement professionals deserve - grounded in real operational knowledge, personalised to the individual, and built with data sovereignty at its core.

The UAE incorporation reflects the founder's alignment with the region's ambition in AI and technology, and positions Procurement Union within one of the world's fastest-growing procurement markets.