Block Cut 40% of Staff, Boosted Code Reviews 150% With AI
Block cut 40% of its workforce, lifted code review output 150%, and automated 15% of code changes—leaders from four companies explain the AI operating model.
By Grace Kim
3 min read
Updated

What's News
- Block cut its workforce by 40% earlier this year and removed management layers.
- Block engineers improved code review changes by 150% from start of year to midyear; non-engineers improved 60%.
- 15% of Block's code changes are now almost fully automated.
- Cloudflare committed to hiring more than 1,000 interns in 2026 and now hires "both employees and agents."
- Executives spoke at the Fortune Most Powerful Women Summit on Tuesday in Santa Barbara.
By Grace Kim
Block cut its workforce by 40% earlier this year and removed layers of management—and the numbers behind that restructuring are now showing up in its engineering output. Engineers improved their code review changes by 150% between the start of the year and midyear, and "even non-engineers have a 60% improvement," CFO and COO Amrita Ahuja said at the Fortune Most Powerful Women Summit on Tuesday in Santa Barbara.
Ahuja appeared on a panel with leaders from Synopsys, Cloudflare, and Amazon, and her message was blunt about intent: Block's leadership didn't want to sprinkle AI on top of existing work.
"We thought about fundamentally refactoring how we operate as a company," Ahuja said.
What changed at Block?
The parent company of Square and Cash App has been consolidating and simplifying processes for years and using AI in the business since early 2024. The payoff is measurable: 15% of code changes are now almost fully automated, meaning "there's only a human in the loop at the end for the final decision," Ahuja said.
The management layers that disappeared weren't random. Ahuja argued AI shifts what roles a company actually needs.
"We fundamentally believe that in a world of AI, you absolutely need the human at the center," she said. You still need builders, goal-setters, and strategy deciders in every function—but not people whose job is to pass information along or check work before it reaches the reviewer.
Return on all this is still an open question. Much of the industry weighs AI costs against personnel savings, Ahuja said. Block is refining a metric it calls "timed customer value," which traces a code change to a feature, then to a customer interaction, and finally to value.
Where does the AI bill get controlled?
Synopsys, which develops software for designing and verifying semiconductors, hit the cost wall fast. CFO Shelagh Glaser said AI spending "was pretty reasonable until it wasn't," comparing employees' usage habits to snacks in the break room.
The company had to decide who gets premium tiers and who uses internal models. When some providers didn't offer a metering system, Synopsys built its own. The goal is to be "super intentional about where that investment is," Glaser said, with returns measured in customer outcomes, not just product releases.
How do you staff an AI-native company?
Cloudflare now hires "both employees and agents," co-founder and President Michelle Zatlyn said. The web infrastructure and cybersecurity company wants all of its roughly 5,000 people to be builders across the business.
One standout: a finance director who made extensive use of agents and was tapped by the CFO to lead AI transformation for the finance team. Cloudflare also committed to hiring more than 1,000 interns in 2026, and they've helped teams become more AI-native, Zatlyn said.
Amazon measures returns by tracking adoption first, then engagement, then efficiency through outcomes like time saved, said Heather Zorn, the company's VP of customer experience, products, and services.
"The important thing as a leader right now is to make sure that you're giving time and space and permission for your teams to learn," Zorn said.
She framed the payoff as a sequencing problem: "You want to create something that wins for the customer, and then you have to find a way that it wins for the business as well." Leaders, she added, have to decide how long they're willing to invest before seeing both wins.
What comes next for org charts?
Ahuja expects AI agents to move onto organizational charts, with workers responsible for both training and deploying them.
"Where things are headed—and I've already heard about a couple of companies doing this, so we're on the frontier—is you look at your work chart that your company publishes, and it will have not only units, but agents," she said. "Those agents, not too dissimilar to humans, need to be developed over time."
If Block's numbers hold, the companies that refactor first won't just be saving on headcount—they'll be shipping faster with whoever, and whatever, remains.
Original: conferences.fortune.com
More from Grace Kim
Show full bio
Market editor covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.
876 articles