Jeremy Olson

17.7K posts

Jeremy Olson banner
Jeremy Olson

Jeremy Olson

@jerols

PM Director at @superhuman but actually product builder (more fun!) @maxtokenai building my side hustles. Past: @coda_hq, founder of @tapityapps & @usehours.

Charlotte, NC Katılım Nisan 2007
1.4K Takip Edilen6.7K Takipçiler
Jeremy Olson
Jeremy Olson@jerols·
It's quite important to train your vagus nerve. Big unlock to be able to regulate emotions and return to calm after stress. This is obviously why I am playing Resident Evil Requiem. Great workout. I've also been exercising my skill of rationalization. That's a good one too.
English
2
0
3
165
Jeremy Olson
Jeremy Olson@jerols·
@DanGrover Big caveat though: I think this only works with tiny tiny (2-4 people) high trust, super IC teams where every role stretches into the other, but all bring their own unique skill set and quality bar from each discipline to bear on the thing itself.
English
0
0
1
74
Jeremy Olson
Jeremy Olson@jerols·
Last year we did a lot of these prototypes and we definitely felt this pain. Designers often would have to go back to Figma for "final" handoff to Eng. But this year with the latest models I think we have a solution: designers just work directly in prod. Skip the prototype and the handoff. Prototype in prod, refine in prod, and engineers review and adjust any architecture issues as needed to ship. All roles are huddled around the same artifact: the actual thing itself.
English
1
0
16
715
Dan Grover
Dan Grover@DanGrover·
Biggest gap I've seen with "prototypes as PRDs" idea in practice so far (vs figma mocks) is that I haven't seen any teammate make any entirely slopless prototype. If you are giving an artifact to engineers, it needs to be perfect, because things will only degrade from there.
English
24
16
224
18.9K
Jeremy Olson
Jeremy Olson@jerols·
When super-IC teams of 1-3 people are building products and features end to end, the work will increasingly become a sharper reflection of the individuals. This is why talent density is becoming far, far more important than headcount. It's why, in going after a very ambitious project, I handpicked one very talented AI forward engineer to start when way more resources were available.
English
1
0
1
199
Jeremy Olson
Jeremy Olson@jerols·
this looks really cool
Joan Rodriguez@joanrod_ai

Introducing @QuiverAI, a new AI lab and product company focused on frontier vector design. We’ve raised an $8.3M seed round led by @a16z, with support from amazing angels and investors. Our first model, Arrow-1.0, generates SVGs from images and text. It’s available now in public beta at app.quiver.ai

English
0
0
0
369
Jeremy Olson
Jeremy Olson@jerols·
Well said
Andrej Karpathy@karpathy

It is hard to communicate how much programming has changed due to AI in the last 2 months: not gradually and over time in the "progress as usual" way, but specifically this last December. There are a number of asterisks but imo coding agents basically didn’t work before December and basically work since - the models have significantly higher quality, long-term coherence and tenacity and they can power through large and long tasks, well past enough that it is extremely disruptive to the default programming workflow. Just to give an example, over the weekend I was building a local video analysis dashboard for the cameras of my home so I wrote: “Here is the local IP and username/password of my DGX Spark. Log in, set up ssh keys, set up vLLM, download and bench Qwen3-VL, set up a server endpoint to inference videos, a basic web ui dashboard, test everything, set it up with systemd, record memory notes for yourself and write up a markdown report for me”. The agent went off for ~30 minutes, ran into multiple issues, researched solutions online, resolved them one by one, wrote the code, tested it, debugged it, set up the services, and came back with the report and it was just done. I didn’t touch anything. All of this could easily have been a weekend project just 3 months ago but today it’s something you kick off and forget about for 30 minutes. As a result, programming is becoming unrecognizable. You’re not typing computer code into an editor like the way things were since computers were invented, that era is over. You're spinning up AI agents, giving them tasks *in English* and managing and reviewing their work in parallel. The biggest prize is in figuring out how you can keep ascending the layers of abstraction to set up long-running orchestrator Claws with all of the right tools, memory and instructions that productively manage multiple parallel Code instances for you. The leverage achievable via top tier "agentic engineering" feels very high right now. It’s not perfect, it needs high-level direction, judgement, taste, oversight, iteration and hints and ideas. It works a lot better in some scenarios than others (e.g. especially for tasks that are well-specified and where you can verify/test functionality). The key is to build intuition to decompose the task just right to hand off the parts that work and help out around the edges. But imo, this is nowhere near "business as usual" time in software.

English
0
0
1
562
Jeremy Olson
Jeremy Olson@jerols·
Last year using these tools to 10x prototyping output was obvious. Now I'm merging over a dozen significant PRs per day to a mature prod environment running usually 5-8 agents at a time to build a project that we would otherwise slot a large team many quarters to design and build. It's hard to even fathom the gains of going from rough idea to prod. Not just in speed but in truth finding and, ultimately, quality. I have no task list. If I have a big idea, a bug, an improvement as I use it - I just spin up a new Claude Code worktree in @conductor_build and the swarm goes after it. Hard to even relate to assigning tickets to engineers in weekly planning. It's definitely not 10x better. It's far, far more.
Omar Pera@ompemi

x.com/i/article/2026…

English
0
0
3
847
Jeremy Olson
Jeremy Olson@jerols·
Has anyone solved the problem where Claude Code can only have one connection to Chrome at a time? I want all my Claude Code instances to be able to spin up agents to check their own work in the browser.
English
2
0
0
245