AI performs seven distinct supplier-relationship roles that firms report as boosting supply-chain performance (~49%), resilience (~66%) and risk-detection (~85%), and cutting procurement times (~85%); however, the most valuable functions remain under-adopted because legacy IT, bad data, and low digital skills make them hard to use.

Main Finding

The systematic review (35 peer‑reviewed studies, 2016–2026) finds that AI is reshaping Supplier Relationship Management (SRM) into a proactive, data‑driven function. Seven distinct AI roles are identified—centered on real‑time monitoring, decision support, and predictive risk—which together deliver large reported gains in supply‑chain performance and resilience but face adoption limits driven primarily by Perceived Ease of Use (legacy systems, data quality, workforce digital literacy) under the Technology Acceptance Model (TAM).

Key Points

• Seven AI roles in SRM (typology validated across sectors: manufacturing, construction, banking, enterprise procurement):
• Real‑time supplier performance monitoring
• Data‑driven decision‑making (analytics/forecasting, spend analytics)
• Predictive risk assessment and mitigation (including federated learning)
• Procurement cost optimization (including RPA and spend control)
• Buyer–supplier collaboration platforms
• Supplier selection and segmentation (ML matching)
• Contract management and optimization
• Reported empirical outcomes (from reviewed studies):
  • Supply‑chain performance improvement: ~49%
  • Supply‑chain resilience improvement: ~66%
  • Early risk detection accuracy: ~85%
  • Supplier financial‑distress prediction: ~88% (up to 8 months ahead)
  • Procurement processing time reductions: reported up to ~85% (e.g., 14 days → 2–3 days)
  • Supplier selection accuracy improvement: ~42%
  • Contract processing time reduction ~60%; spend reductions ~15–20%; compliance issues down ~30%
  • Supplier quality improvement example: ~70% → ~94% within 12 months after AI integration (reported)
• Adoption framing via TAM:
  • All seven roles generate measurable Perceived Usefulness (PU).
  • Perceived Ease of Use (PEOU) barriers—legacy/ERP incompatibility, poor data quality, limited digital skills—suppress uptake of high‑impact functions.
• Additional findings:
  • Generative AI and NLP extend risk assessment by analyzing unstructured data (news, communications).
  • Federated learning cited as a privacy‑aware route for cross‑firm risk modeling.
  • Research gaps highlighted: algorithmic bias, longitudinal deployment effects, evidence from developing economies, and integration of AI with ESG/compliance objectives.

Data & Methods

• Method: Systematic literature review following PRISMA 2020.
• Search (February 2025): Google Scholar, Semantic Scholar, CORE; five search strings → 29,900 initial records.
• Screening flow (author report):
  • Deduplicated → ~14,950 records screened by title/abstract.
  • Full texts assessed: 133; excluded 98 for reasons (blockchain only; insufficient AI specificity; non‑peer reviewed; predatory outlets; duplicates).
  • Final sample: 35 peer‑reviewed articles (2016–2026), English, full text accessible.
• Thematic analysis: two‑round coding protocol adapted from Braun & Clarke (2006). Articles could be assigned to multiple themes if central to each.
• Theoretical lens: Technology Acceptance Model (PU and PEOU), used to interpret adoption dynamics in the reviewed literature.

Implications for AI Economics

• Cost structure and productivity
  • Reported reductions in procurement processing time and spend translate into lower transaction and administrative costs, faster working‑capital cycles, and higher procurement productivity. Aggregate firm‑level cost savings (15–20% on spend categories, large reductions in process time) imply potential improvements in margins for adopters.
• Value capture and market structure
  • AI that improves supplier monitoring and selection can shift bargaining power: buyers gain more precise leverage (through better information and easier switching), potentially compressing supplier margins in concentrated industries; conversely, suppliers who adopt AI or integrate well may capture more demand by proving reliability.
• Risk pricing and externalities
  • Higher accuracy in early risk detection (≈85%) should reduce realized disruption costs and change the premium pricing of supply‑risk insurance or contingent sourcing strategies. Federated approaches that share risk intelligence may create positive externalities but also raise coordination/antitrust and privacy considerations.
• Labor and task reallocation
  • Automation of transactional SRM tasks (RPA, contract processing) shifts procurement roles toward strategic decision‑making and relationship management, implying upskilling demand and short‑term displacement of routine roles. The PEOU barriers (digital literacy gaps) will affect the pace and distribution of these labor adjustments.
• Investment, diffusion, and inequality
  • Adoption frictions (legacy systems, data quality, skills) imply uneven diffusion: large firms with modern ERPs and data capabilities will likely realize gains faster, potentially increasing competitive divergence across firms and countries. Developing economies may under‑capture benefits unless capacity building and data infrastructure investments occur.
• Policy, governance, and ESG
  • Integration of AI with ESG and compliance monitoring presents opportunities to internalize non‑price supplier attributes (labor, environmental compliance) into sourcing decisions. However, risks of algorithmic bias, opaque decision rules, and regulatory gaps mean policymakers need to consider transparency, auditability, and fairness standards for procurement AI.
• Research and measurement recommendations for economists
  • Need for causal, firm‑level studies (RCTs, difference‑in‑differences) to estimate net effects on prices, quantities, supplier welfare, and market structure.
  • Longitudinal evidence to track persistence of gains, second‑order effects on supplier investments, and labor market adjustments.
  • Cross‑country work to assess how infrastructure and institutional differences mediate AI returns in SRM.
  • Better measurement of externalities (systemic risk reduction, ESG outcomes) and distributional impacts across suppliers.

Concise conclusion: the review provides a clear typology and promising empirical signals that AI materially improves SRM outcomes and reduces costs, but adoption bottlenecks and open questions about causal impacts, distributional effects, and governance mean AI’s full economic consequences in supplier markets remain an important area for targeted economic research and policy design.