Everyone is talking about AI. And Generative AI adoption, especially, has practically become a corporate mandate - everywhere you look, it is being deployed.
Across industries, executives are urging teams to integrate AI into their workflows. Engineers want access to AI coding assistants, and consumers are constantly being introduced to new AI products whether they want them or not.
But amid the rush to deploy new tools, many organizations are overlooking a simple question. Is it doing its job and making anything any better?
“We are helping engineering leadership to govern AI and adopt it,” said Liad Elidan, co-founder and CEO of Milestone. The platform provides engineering leadership with something deceptively simple: an honest account of what AI is actually doing inside their organization. Not what the AI vendors tell them it’s doing — what’s really happening, measured against business outcomes that matter.
Elidan described it as a situation where the whole system is pushing forward at once: “The world’s adopting AI. Every person is using AI, either in their personal life or in their professional life.” But the result is that many organizations deploy AI tools before they fully understand what those tools are actually doing. And the metrics provided by those tools do not necessarily answer the questions executives actually care about.
This is giving rise to an emerging category of technology: systems designed specifically to measure the interaction between humans and AI tools inside enterprise environments. Milestone sits at the center of that category — sitting above the vendors, correlating usage data with actual engineering outcomes like code quality, review times, and delivery speed.
Elidan described this challenge as a shift in management thinking. “Now you add another animal into the play, which is AI itself.” Looking further ahead, he is optimistic about where this leads — for engineers willing to adapt. The profession isn’t contracting, he argues. It’s mutating into something far more powerful.
“It’s very controversial,” he said, “but you can be much more independent now. And that is amazing.” The engineer who learns to work with AI tools effectively, who treats adaptability as a core professional skill rather than an occasional inconvenience, is not being replaced. They are being amplified.
Five years from now, Elidan predicts the best engineers won’t be defined by what they can write, but by what they can direct. “I expect that engineer to be almost a Superman engineer compared to the engineer of today,” he said. “He’ll have a hundred or more agents under his belt — agents that he can run, configure, and push forward.”
The future of software development, in other words, may be defined by the ability to manage how humans and AI work together and the wisdom to actually measure whether it’s going well.
