AI that actually matters in 2026, Andrew Kumar | EvolveDigital Toronto 2026
Summary
Andrew Kumar, Global VP of Technology at Uniform, delivered a presentation examining the reality versus hype of AI applications in digital experience platforms as of 2026. Kumar positioned himself as someone with credibility to discuss both sides, having worked at major technology companies including Contentful, Mulesoft, and Telus, while acknowledging his role at a vendor that contributes to AI hype. He opened by explaining MCP (Model Context Protocol) servers, which allow AI agents to work across different systems, and provided context about Uniform's role as a headless CMS unifier that helps organizations with complex multi-system architectures.
Kumar presented compelling industry predictions, including Gartner's forecast that 80% of customer interactions will shift from traditional touchpoints to agentic experiences by 2028. He discussed the MACH Alliance's research showing that composable, API-first organizations are adopting AI faster than those using legacy systems. The presentation highlighted how brands need composable infrastructure to maintain brand guidelines across various AI-powered interfaces, including chat experiences where AI agents can retrieve content definitions and render them appropriately.
However, Kumar emphasized that much current AI discussion remains predictive rather than proven. When examining actual AI adoption data from Uniform's customer base over the previous 90-120 days, he revealed that the most popular feature was 'AI guidance' - essentially providing AI tools with brand voice, tone, and editorial guidelines to improve output quality and reduce token usage. Other heavily adopted features included content creation, content previews across multiple channels, translations (with one Danish customer using translation services 104 times in a month), and metadata creation for SEO purposes.
Kumar proposed a key hypothesis: there's a strong correlation between how boring and painful a task is and the likelihood of AI adoption for that capability. He advocated for starting AI implementation with unglamorous but necessary tasks like SEO metadata and translations before moving to more strategic applications. The presentation concluded with practical principles for AI adoption, including prototyping before replacing enterprise tools, maintaining strategic alignment with business objectives, managing escalating token costs, keeping humans in the loop for quality control, and understanding when processes break down.
During the Q&A, Kumar addressed organizational readiness for AI adoption, emphasizing the need for proper infrastructure, governance guardrails, and starting with meaningful small-scale implementations rather than attempting enterprise-wide AI deployment.
Key Takeaways
- •The most adopted AI feature is 'AI guidance' - providing brand voice and editorial guidelines to improve output quality and reduce expensive token usage
- •There's a strong correlation between how boring/painful a task is and the likelihood of successful AI adoption for that capability
- •Organizations with composable, API-first architectures (MACH Alliance principles) are adopting AI significantly faster than those with legacy systems
- •AI token costs are escalating rapidly across organizations, making cost management a critical consideration in implementation
- •Start AI implementation with unglamorous but necessary tasks like SEO metadata and translations before moving to strategic applications
- •Proper infrastructure and governance guardrails must be in place before scaling AI adoption across an enterprise
- •One customer increased product launch velocity from quarterly to 3-5 launches per month using AI translation capabilities
- •Content preview capabilities across multiple channels become increasingly important when AI is generating the content
- •Prototype and experiment with AI tools before replacing existing enterprise solutions to avoid scaling issues
- •Human oversight remains critical - know when to keep humans in the loop and when processes break down
- •MCP servers enable AI agents to work across different systems, opening new possibilities for cross-platform content orchestration
- •The shift from human-operated to agent-operated content consumption requires rethinking how content is structured and delivered
Transcript
I think we're ready to get started here. Um, so thank you for coming after meeting me after lunch. Um, we are starting now with Andrew here from Uniform who's going to be talking about AI, everyone's favorite subject. Um, a lot about the hype and and and what is and what isn't hype. I'm pretty excited to see this to hear about this and hear what they're doing. Um, if you don't know about Uniform, they're a sort of a headless CMS uh, Unifier or Uniform. Uh, bringing together a bunch of different ...
