With Agents, I'm 30x More Productive. What Will I Do Now?

My AI agents now make me about 30x more productive than before AI. So what am I doing with my time? I’m working on more projects at once, and looking for social impact organizations that might especially need me. Here’s how I got to this point.

Estimating Productivity

For most of the last half year, I’ve been measuring and estimating how much my AI agents increase my productivity (or not).

To do this, I’ve been doing the analysis below using my weekly review tool:

• Estimated hours of output: how long my work might have taken with no AI help. I estimate this using AI, but spot check and calibrate the estimates.
• Hands-on hours: the number of hours I actually worked
• Productivity multiplier: estimated hours of output divided by hands-on hours

For much of this year, my AI agents have increased my productivity about 10x versus life before LLMs existed. In mid-August, I updated my agent management tool (a tool I built to run many agents at once and review their work), and now my productivity is up about 30x.

As an example of what 30x feels like, in the first 2.5 weeks after I switched to the new agent management tool, I used about 29 hours of hands-on time to build personalized replacements for Asana (task management and goal setting) and Granola (meeting recording and notetaking), and moved most of my work in Claude Desktop over as well.

Before I had access to LLMs, I would have continued using Asana, Granola and Claude Desktop — all good products that took engineering teams months to years to build out. But instead, in a few weeks, I built out workflows that work better for me and my agents, and that are better integrated. And all of this came on top of a foundation where I’d already replaced Notion and Figma a few months earlier.

AI Companies’ Own Engineers Show Bigger Gains (5-8x), But Software Industry Overall Is Changing Gradually

If we compare how things looked before my tool update, published reports from the middle of 2026 showed a wide range of outcomes:

Anthropic: code merged per engineer per day vs. 2024, which they call “almost certainly an overstatement”. OpenAI: lines changed per contributor vs. 2024. DX Survey: median number of finished code changes per engineer across over 500 organizations vs. Q3 2025. DX reports before that showed virtually no gain at all, so we can assume 2024 was a similar baseline as 2025.

Yes, I’m handwaving here a bit — but this much variability is typical given how long technology adoption takes. A few of us are seeing outsized gains (especially for me, since personal projects come with fewer risks and constraints). Yet, industry-wide, change is happening more slowly. Both people and the organizations they’ve built take time to adjust. Plus coding faster does not mean we deliver value to users faster.

History Says Big Changes Arrive Unevenly, and Slowly

Looking at the past when productivity increased, like during the Industrial Revolution, may give us some idea of where we’re going.

The Industrial Revolution didn’t happen everywhere at once. Britain’s Industrial Revolution began in the 1770s; the United States and Western Europe followed through the 1800s, Japan from the 1880s, South Korea from the 1960s and China from 1978. Some countries haven’t started.

Even where a new technology arrives, it takes time to change how work gets done. By 1919, electric motors supplied 53% of the mechanical power in American factories (calculated by David from Richard DuBoff’s 1979 horsepower tables), but factory output per hour of work had grown only about 1.2% a year over the previous decade.

In the 1920s, as factories rebuilt around electricity with new layouts, processes and tools, factory output per hour grew 5.6% a year (Paul David, 1989, Tables 2 and 3).

AI can depend on existing infrastructure where billions of people already use computers and mobile phones with applications served from the Cloud; however, many processes and tools still need to be invented. This will take years to decades (depending on which forecasts you believe).

What’s the Broader View Today?

Even though across the overall economy, AI has made little productivity impact, the rate of change is already uncomfortable for most Americans — in two Pew surveys this year, “Americans largely think AI is moving too fast. About two-thirds say this,” and “71% of adults think AI will lead to fewer jobs in the United States over the next two decades, up from 64% in 2024.”

Part of this is due to widely reported existential risks like AI killing all humans (or, more likely in the near term, humans using AI to kill or hurt many people); however, mitigating unlikely but catastrophic future risk is just one concern.

AI is already impacting people today: companies are hiring fewer entry-level workers, AI is changing people’s livelihoods without their say (e.g. mathematicians), and access to AI is unequal (across genders, education levels, incomes and countries).

What Does This All Mean for Me?

In times of rapid change, one of the articles I often go back to is a Harvard Business Review article from 1997: “Strategy under uncertainty”. The article talks about first characterizing what kind of uncertainty you face and then choosing appropriate tactics in response. Tactics that work best in a time of stability are different than what’s best in a zombie apocalypse. And we’re probably somewhere in between.

So Much Uncertainty!

There are many open questions and scenarios for how the world might play out:

What Can You Do Under Uncertainty?

These are some of the tactics for dealing with the level of uncertainty we have today:

No-Regret Moves

Figure out what’s almost always going to be true, despite all of the uncertainty, and find actions that have no downside.

My example: Extreme inequality exists today, and I haven’t seen evidence that AI is making it better yet. As Bill Gates warned last month, “left to the market, AI will be designed by and for the richest people in the world.” Despite efforts like Anthropic’s Beneficial Deployments Team, AI company strategy largely appears to be to chase where the money is. So I’ll keep looking for ways to help fill in the gaps. See my about page for more.

Identify Opportunities

Look for evidence about what scenario you’re in, and go after the opportunities that open up in that scenario of the world.

My example: When I started consistently seeing my multiplier around 10x a few months ago, but with slower progress everywhere else in the real world, I’m now acting with that reality in mind:

For most of my career, I’ve focused on one startup at a time, because I needed that much time on a problem to feel like I was making progress. Now, at 30x the productivity, I can move many more projects for myself, friends, and collaborators at once.

So I plan to take on more part-time gigs, even as little as 2-4 hours hands-on time a week. But with those few hours, I will be doing as much work as I used to do in weeks.

Shape With Big Bets

Take a bet and lead on how you think the world should work. As Rumi said, “let the beauty we love, be what we do”.

My example: Based on reviewing AI research progress, I believe humans will remain uniquely good at many things for a while (or indefinitely). But AI is rapidly becoming superhuman in many skills. This will be hard for most people — virtually all of us, including me, change slower than the pace this is going (and I’m funemployed and have time to learn). Yet we still want agency to choose how change happens in our lives.

So I do want to find ways to help fellow humans change, learn, and have agency. Cooking up some ideas in this direction :)

In Short

With major shifts already underway from AI, in the last few months, I’ve been looking for ways to make a bigger social impact, taking on more part-time gigs, and thinking about how to help everyone else through these changes. Let me know if you have any interesting leads!

History tells us that when there are major shifts in productivity, the gains don’t reach everyone on their own. People have to help make it happen faster.

What’s your experience been like, and what are you doing with your time?