AI to Boost SME Participation in Public Procurement

Why focus AI on SME participation

Public procurement must balance efficiency, competition and compliance with Law 9/2017 on Public Sector Contracts. In many municipalities and public bodies, formal requirements, vague criteria or excessive documentation reduce participation — especially from SMEs — and can increase long‑term costs. AI is not a cure‑all, but when applied with technical and legal criteria it can detect and reduce barriers to entry, improve the clarity of tender documents, and make it easier for small companies to submit bids.

Below I outline concrete uses, compliance requirements and a practical roadmap for municipal pilots.

Concrete AI use cases to help SMEs

• Automatic analysis of tender documents to detect unnecessary or overly restrictive requirements

  • An NLP engine reviews criteria, flags requirements that limit competition (for example, highly specific certifications without technical justification) and highlights sections for legal review.
  • Useful output: a prioritized list of “potential barriers” with references to Law 9/2017 and a technical memo.

• Simplification and standardization of requirements

  • AI‑assisted text generators that propose alternative wordings that are clearer and reduce ambiguity while preserving legal obligations.
  • Includes clause templates compatible with publicity and transparency duties.

• Market impact assessment and competition simulation

  • Models that, using public data (past contracts, business activity registries), estimate how many market participants could meet certain requirements.
  • Result: a “likely competition” indicator for each criterion, helping to calibrate technical thresholds.

• Guidance and support for SME bidders

  • Chatbots and assistants that explain an procurement file in plain language, generate personalized checklists and flag documents that are commonly missing.
  • Important: support should be backed by human assistance for legal questions and final decisions.

• Automatic prevalidation of bids

  • Scanners that check whether submitted documentation meets formal requirements (annexes, signatures, deadlines), reducing correction requests and administrative costs for SMEs.

Legal, security and transparency requirements

• Law 9/2017: any change to tender documents must respect the principles of free competition and proportionality. AI‑generated recommendations should be reviewed and signed off by the responsible authority.
• GDPR: company and personal data processed (e.g., participation histories) must be handled in accordance with the GDPR: lawful basis, data minimization, data subject rights (access, rectification, erasure and objection), and impact assessments where applicable.
• EU AI Act: assess whether the tool falls within the scope of the Regulation (for example, systems that automate decisions with legal effects or that are used for classification). If it applies, obligations on transparency, technical documentation and risk management must be met.
• ENS (Royal Decree 311/2022): systems that handle sensitive information or procurement data should align with the security measures of the National Security Scheme (ENS) if they are hosted or managed in public administration environments.

Operational and governance best practices

• Keep human oversight: AI suggests; the final decision on requirements and awards is made by the administration.
• Traceability and logs: record inputs, outputs and justifications for AI recommendations. Preserve logs for auditability and transparency.
• Applicable explainability: produce interpretable summaries (for example, “this requirement reduces estimated competition by X reason”) so legal and technical staff can understand the recommendation.
• Pilot with non‑sensitive data: start with anonymized past files to validate models and mitigate GDPR risks.
• Involve SMEs: include business representatives in validations to avoid adjustments that favor a few operators.

Metrics and KPIs to monitor

• Number of bids per file (before/after the pilot).
• Rate of formal correction requests from SMEs.
• Average time SMEs take to prepare a bid.
• Number of criteria reviewed/adjusted based on AI recommendations.
• User satisfaction (technical staff and bidders) with the clarity of the tender.

Do not invent percentages: measure and compare locally in your municipality.

Practical roadmap (90‑day pilot)

1. Select 2–3 representative contract types (works, supplies, services).
2. Assemble the team: procurement, legal, IT and an SME representative.
3. Prepare anonymized historical data and sample tender documents.
4. Define clear pilot objectives (e.g., reduce correction requests in X type of file).
5. Deploy a tender analysis tool in a controlled environment; generate recommendations, not automatic changes.
6. Evaluate results with metrics and adjust the model confidence thresholds.
7. Formalize a process for incorporating recommendations: legal review, political approval, publication.

Modular platforms like OptimGov can be integrated into these workflows to combine analysis, traceability and decision support.

Risks and how to mitigate them

• Overreliance on AI: mitigate with business rules and human review.
• Competition filtering due to data bias: audit models and avoid discriminatory criteria.
• Handling of personal data: apply GDPR principles and anonymize data during training phases.

Conclusion and recommended action

AI can be a practical lever to increase SME participation in public tenders through tender analysis, market simulations and bidder support. Immediate action: organize a 90‑day pilot focused on a single contract type, with measurable goals and legal oversight from the start. This will allow you to validate real benefits without compromising compliance or security.