Just thought this was good knowledge for
leaders thinking about getting their AI game more systemized across orgs.
Dennis Woodside
Reskilling used to be framed as an individual project. Nights, weekends, your career, your problem, your dime. Plenty of people are still doing it that way —… more
Reskilling used to be framed as an individual project. Nights, weekends, your career, your problem, your dime. Plenty of people are still doing it that way — nobody is going to be more invested in your fluency than you. But the pace of AI has pushed the load past what individual effort can carry on its own. We've ended up splitting it three ways. The first part stays with the individual, and I don't want to soften that. The second sits with the company and it costs money. The third sits with whoever is running the place and it costs something harder to give; time. Companies that leave all three on the employee will get the fluency of the people who had time. On Freshworks’ share: we run a program called AI in Motion. We recently wrapped the third installment — three regional events across 11 cities, thousands of employees. Those events produced 25 innovation showcases: working prototypes from SecOps, IT, CX, Sales, Sales Ops, Engineering and Product, each one built against real customer or business problems. And 15 sessions where employees taught other employees how they're using AI in their own jobs — financial workflows, building pipeline, prototyping. Two things about that. It happened on work time, not on people's weekends. And it wasn't built for engineers. The prototypes that surprised me came out of functions that don't write code for a living. On the leadership share: I plan my whole training year in Claude — I run four races a summer and the plan is built and adjusted there. I also build prototypes in Lovable and send them to the product team. They're usually terrible ideas, but you can't set expectations from a distance. If the people running a company aren't in the tools, everyone below them reads that accurately, and no training program closes that gap. Here's what it looks like when fluency lands. Our product development lifecycle has changed. Designers work in Figma, push through Figma Make directly to code, and we've built the hooks into our production environment so what comes out already meets our internal coding standards. QA is largely automated. Cycle times are ~30% faster and AI products ship on a two-week cycle, which we weren't doing before. We've also turned our own AI Email agent on our own billing questions. It handled about 30% of them from the day we switched it on. Neither of those came from buying something. They came from people who already worked here figuring out what the tools could do in their part of the business. Most of our customers aren't from the tech industry. They're not benchmarking models on the weekend. They're looking to us to be conversant enough in this to help them get from experimentation to real use, and you can't outsource that to your product. Most of the AI debate at the CEO level is about which model, which stack, how much budget. That debate matters. What determines whether AI shows up in the P&L is whether the people you already employ know how to use what you've already bought19 reactions · 5 comments · – repostsOpen on LinkedIn
