We’re in the midst of a massive pivot in the entire line of work of software development. At least in some places, like where I work, everything is different.

Breakneck change

A year ago, I thought agents were vaporware and LLMs were not to be trusted with anything where truth is paramount.

If you asked me 2 months ago, I would have told you that communication skills with the LLM were now paramount.

Now, in some cases, that’s barely necessary. I’m sometimes able to one-shot changes just by pointing Claude code at a ticket with just a description and no content.

People are unleashing fleets of agents on backlogs. Coding is no longer the bottleneck, and gradually, we’re working on eliminating subsequent bottlenecks, like testing and code review. I believe it’s completely possible.

For a glimpse of what people on the bleeding edge are doing, read Welcome To Gas Town. This could soon be how we all work.

The one caveat is that my company is spending oodles of money on tooling and AI credits, and not every org has the capacity or desire to do this.

So, what do we do?

A colleague in /dev/color posed the question of how us devs can future-proof ourselves. My perspective right now is that there is no future-proofing, there is only the now. And now, my mandate is to figure out how to create value, by any means necessary.

Our CEO said in All-Hands that knowledge work is essentially capability * taste * agency. In software:

• Capability is dev skills
• Taste is knowing what should be built
• Agency is opportunity, will and execution

AI is leveling capability, so value is coming from taste and agency going forward. If this is true, this completely pivots the value structure. As devs, we made huge investments of time and energy into cultivating capability, and in a lot of cases, our role has not been to bring taste and agency. That was for designers and PMs, at least in product engineering. (Maybe one could say there’s still space for engineering taste in system architecture.)

I’m thankful that I think taste is a strong point for me. What I’m not used to is it being it as my job. Put another way, I have to rethink how I exercise agency. That’s what I’m working to figure out: rewiring my instincts from what I’m used to doing and how I’m used to doing it, internalizing a new sense of possibility.

For instance, I used to think a lot about how to get buy-in for my ideas. This is not just because it’s good to have support and advocacy, but also because I am asking my company to allocate some of its scarce labor and capital to something I think is best, in the face of thousands of competing priorities. That needs to be carefully justified. But now, I have the power in many more cases to just create the thing, and then the question is whether it is the right thing for the company to adopt.

This is a significant mindset change. It’s not easy, but it’s what needs to be done. On the bright side, the opportunity for impact is much greater than ever before.