Skai

So proud of my team for their incredible work on this. The next generation of agentic media planning, activation, optimization, and measurement is here! Just announced at #unBoxed2026: Amazon’s new AMC SQL Generation expert tool lets agents turn natural-language business questions into expert-level AMC queries for custom insights and audiences. During the Amazon Ads keynote, my colleague Courtney Crossley and our great partner Benoit Vatere from Liquid Death explained how Skai’s agents used this to determine where to spend the brand’s next best dollar to efficiently reach new customers. Skai’s agents and agent-native environment: - Brought Amazon’s AMC intelligence together with Liquid Death’s brand context - Identified the audiences and campaigns showing the strongest opportunities - Allocated spend behind new-to-brand audiences and areas with the strongest growth potential - Turned those insights into a clear plan, activated it across Amazon Ads products, seamlessly optimized post-launch, and leveraged Skai’s best-in-class measurement methodologies to create a flywheel that delivers significant impact. The big opportunity that Amazon and Skai make possible: connecting business questions to insight, decision, approval, and action, with marketers remaining in control. Find out more about our latest innovations: https://okt.to/hYcLiz

For agencies, going agentic isn’t the hard part anymore. Scaling it across dozens of teams and brands is. Teams are already building agents, workflows, and new ways of working at incredible speed. But when everyone can build, how do you make sure everyone is building to the same standard? Benefiting from each other's learning? That’s the real challenge. In Skai’s latest “Ask the Experts,” Courtney Crossley, EVP of Strategic Client Advisory at Skai, and I dig into what it takes to scale agentic AI without sacrificing quality, consistency, or speed, and why this is as much an operating model shift as a technology one. If your agency is navigating an agentic transformation, this one’s for you: https://lnkd.in/g_M_gg7z

Why do pilots fail to scale? Well, have you ever watched 90 Day Fiancé, Love Is Blind, or Age of Attraction? Then you've seen a failure to scale a pilot, in real time. Two people fall for each other under conditions built for falling in love. Good lighting, a set schedule, producers smoothing every rough edge. Then the cameras stop and real life shows up: rent, in-laws, a bad commute, a bad day. The connection that felt inevitable on screen doesn't survive contact with an actual apartment. AI pilots run on the same formula. Clean data someone spent three weeks scrubbing by hand. A team of believers who'd troubleshoot at 11pm because they wanted to be right. Governance that looked the other way so the demo could ship on time. Take away the produced conditions and the "success" doesn't scale. It was never built to. My latest byline with Skai is live. In it, I break down why fewer than 1 in 5 companies get an AI pilot past the pilot stage, and what building for real conditions looks like from day one. Read it here: https://lnkd.in/eTeXCxyJ

It's been a while since we ran a test and were so pleasantly surprised by the results... For over a decade, Skai has helped marketers (humans) take action efficiently and gain insights across hundreds of ad accounts and any number of publishers. A single view eliminated the need to stitch things together or repeat the same actions again and again. So we thought, with our MCP, we should be able to provide similar value to their agents, right? We expected an agent working with Skai's MCP to be somewhat faster and cheaper, since it wouldn't need to use a publisher's MCP account by account, action by action, report by report. The results were far more decisive: - At a realistic, medium-advertiser scale (e.g., three publishers, 90 accounts, 300 questions a month), using the Skai MCP is about 10 times cheaper, and the difference grows with the number of accounts. - Skai came back with a complete answer about 2.5 times more often, while publisher MCPs did not give the agent enough to work with, forcing it to revert to the user for clarification. If you're only using agents for reporting today, this is the floor, not the ceiling. As more real work (budgets, bidding, optimization, trafficking) shifts from people to agents, these same savings apply to a much bigger number. Read the full findings: https://lnkd.in/ev3qXgXr

I just built an omnichannel live dashboard with Claude CoWork and Skai's new MCP in minutes! With Claude’s new Live Artifacts, you can easily build a dashboard that will continue to update every time you view it, directly from Skai’s MCP, supporting all your publishers in one place. Digging into the data, identifying what's important, and visualizing it for ongoing tracking and insights is just a short conversation away. The demo in the gif is the result of a 10-minute chat, 8 of which I spent simply watching Claude do its thing… Writing this post took me longer…. Connected to our MCP? Want the prompt? Just ask!

Manage, optimize, and report on all your commerce channels from one platform. See how it works. Watch the product tour.