DAM is constantly evolving, and the latest shift is significant: moving from a static repository to an intelligent teammate. While many of us recognize that modern DAM systems do more than simply store assets, they’re still often perceived, and function as, relatively static tools. The next evolution changes that fundamentally.
In October, Dovecot had the pleasure of attending and presenting at Henry Stewart DAM NY 2025. Throughout the conference, a clear theme emerged: we’re at an exciting inflection point for digital asset management. AI is finally mature enough to handle the tedious and difficult tasks that have challenged humans involved with DAM since the beginning. Fortunately for us, DAM is an ideal domain for practical AI applications – not just AI for AI’s sake.
To support this evolution, today’s DAM needs to communicate in new ways: with AI agents and other machines, and with humans through AI-supported channels, not just through direct logins or integrations. Sarah Iskander’s keynote, “The Evolution of DAM: From Static Library to Intelligent Engine,” explored exactly this transformation.
What this means for your DAM journey:
If you’re just starting out: The fundamentals haven’t changed—getting your information architecture (including taxonomy and metadata) right remains critical. But you have a distinct advantage: you can build IA for AI from day one, rather than retrofitting existing structures. You’ll also need a DAM librarian or administrator who brings a unique skill set: AI literacy, including the ability to train AI systems, plus strong communication and storytelling skills. This person needs to be a brand steward who can collaborate effectively with both humans and machines. Sarah James highlighted this skillset in her talk “The Metadata Culture Shift: Evolving DAM Librarians in the Age of AI.”
If you’re upgrading or migrating: Ask your vendor about agentic workflows. In my view, these are the areas where AI agents can deliver the most value in DAM. Inquire about image recognition and text generation capabilities (even though you will probably be shown these without asking.) The key requirement is that these AI features must be trainable. Generic AI won’t understand your specific organizational context, so customization through human-led training is essential.
What excites me most:
The potential for AI agents to handle contract and rights management is genuinely exciting. This has always been incredibly challenging for Dovecot’s clients when managed by humans alone. One vendor (FADEL) presented their AI-driven brand compliance/ rights & royalty software solution that seemed to be able to do exactly this, which I’d never seen before.
I’m also dreaming of an AI agent that can perform metadata quality assurance checks for DAM administrators, another persistent pain point that’s ripe for automation.