Another Crazy Day in AI: What 12X Productivity Actually Looks Like in Practice
- Mar 31
- 3 min read
Hello, AI Enthusiasts.
Here's another crazy day in AI:
How autonomous agents are changing engineering
California tightens oversight on AI use
Apple shifts to a flexible AI strategy
Some AI tools to try out
🎧 Listen to a quick breakdown of today’s stories.
TODAY'S FEATURED ITEM: How AI Agents Are Delivering 12X More Output
Image Credit: Wowza (created with Ideogram)
How do teams work differently when execution is no longer the bottleneck?
Joe Lonsdale sits down with Scott Wu and Russell Kaplan, co-founders of Cognition, on his American Optimist podcast. Cognition is the company behind Devin, the first autonomous AI software agent. Scott is a three-time global gold medalist in competitive programming and a former world champion at 17. Russell previously worked on Tesla's Autopilot team and sold his startup to Scale AI. They founded Cognition in 2023, shipped Devin a year later, and have been growing fast since. The conversation gets into how their team actually works now, where Devin is being deployed, and what happens when AI agents take on software problems that organizations have been sitting on for years.
What's worth noting from this one:
By early March 2026, Devin had already logged more usage sessions than in all of 2025 combined.
One hour of a human overseeing Devin is estimated to be worth 6 to 12 hours of that same person doing the work themselves.
Cognition's own engineers no longer write code by hand — they describe what they need in plain English and the agent handles the rest.
Legacy codebases at banks, insurers, and retailers turned out to be some of the strongest early use cases for Devin, not the hardest ones.
Cognition has moved into government IT, targeting public systems still running on decades-old languages like COBOL.
AI model capability in software engineering is reportedly doubling roughly every two to three months.
Younger engineers tend to adapt faster, mostly because they have less prior workflow to unlearn.
There are still real questions worth asking — around reliability, oversight, and what accountability looks like when something goes wrong without a human directly in the loop. Those concerns are not going away, and any honest look at this space has to hold them alongside the progress.
What is worth sitting with is how much of this extends beyond large tech companies. A lot of organizations — small businesses, public agencies, local governments — have had ideas they simply could not act on because building and maintaining software was too slow or too expensive. Whether that gap actually narrows, and for who, is still very much to be seen.
Watch the full conversation here.
OTHER INTERESTING AI HIGHLIGHTS:
California Tightens Oversight On AI Use
/Cecilia Kang, Reporter, on New York Times
California is taking a more active role in shaping how AI is used—especially when it comes to companies working with the state. A new executive order introduces stricter requirements around safety, privacy, and transparency, signaling a push for clearer accountability. It also shows a willingness to diverge from federal positions, particularly in how companies are evaluated for risk. On top of that, measures like watermarking AI-generated content aim to address growing concerns around misinformation.
Read more here.
Apple Shifts To A Flexible AI Strategy
/Bloomberg Technology
Apple is rethinking its AI strategy by allowing multiple chatbots to operate on its devices, rather than relying on a single system. This “agnostic” approach could give users more flexibility while opening up new revenue streams through an AI-focused app marketplace. At the same time, Apple continues improving its own tools like Siri, even as competition for top AI talent intensifies. The shift reflects a broader industry trend where hardware, software, and AI ecosystems are becoming more interconnected—and more competitive.
Check it out here.
SOME AI TOOLS TO TRY OUT:
Diagrimo – Turn text or ideas into clear, structured diagrams and infographics.
Podcast Transcript – Generate transcripts and summaries from podcast episodes.
Lightning Rod – Turn raw data into training datasets for domain-specific AI.
That’s a wrap on today’s Almost Daily craziness.
Catch us almost every day—almost! 😉
EXCITING NEWS:
The Another Crazy Day in AI newsletter is on LinkedIn!!!
Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.
Comments