AI's Impact on Product Data: A Conversation with Viamedici

In this live PIMvendors webinar, co-founder Stephan Spijkers spoke with Al Bissmeyer, Global Vice President at Viamedici, about where AI delivers measurable value in product data and where it wastes budget. Viamedici has worked with machine learning in MDM and PIM for more than 20 years, which gave the discussion a practical, B2B-heavy perspective covering onboarding, enrichment, translation, classification and publication.

The central message for PIM buyers is clear: AI multiplies the quality of the data and the expertise behind it. Companies with structured product data and skilled data owners cut administration costs and project timelines sharply. Companies that point AI at fragmented spreadsheets get fast output of poor quality and rising token bills. Stephan added a warning for selection projects: for years most PIM shortlists contained platforms of roughly equal fit, and AI readiness of the underlying architecture now creates clear winners and laggards.

Takeaways

• Fix the data foundation first. AI cannot structure a dataset without common terminology, so cleansing, classification and a dedicated system of record come before any AI use case.
• Keep humans in the loop. Al compares AI to a nail gun: a skilled carpenter roofs a house in a fraction of the time, while an unskilled user causes damage. Approval workflows and subject matter experts stay essential.
• Prioritise by revenue. Enrich the 20% of products that generate 80% of revenue first, store results in reusable dictionaries, then scale to the full catalog at lower cost.
• Treat translation as a context problem. Al estimates machine translation at roughly 80 to 90% accuracy. Product context (a car spring translated as the season) and local variants (Paris French versus Montreal French) require expert review.
• Protect data and cost. Small language models behind the firewall, bridged to external tools, reduce both data exposure and API spend.
• Favour headless, API-rich platforms. Coding agents and MCP let teams build custom interfaces for marketing, engineering, sales and quality on top of a stable data layer, which also eases change management.
• Re-examine your PIM shortlist. Partners and platforms must now deliver end-to-end business value and AI readiness, and a PIM chosen on yesterday's criteria may slow you down for the next five to ten years.

👉 Find the PIM solution that fits your AI roadmap at pimvendors.com