Tokens & Tactics #14: Scaling Yourself Horizontally with AI
Philip O'Toole on having GitHub Copilot fix bugs in his open-source database, using GPT to organize cross-team projects, and needing a meta-tool to manage all his AI-generated pull requests.
We’re hosting our next BRXND NYC conference on September 18, 2025 — that’s this Thursday! There are only six tickets left, so grab yours before they're gone!
Welcome back to Tokens & Tactics, our Tuesday series about how people are actually using AI at work.
Each week, we feature one person and their real-world workflow—what tools they use, what they’re building, and what’s working right now. No hype. No vague predictions. Just practical details from the front lines. This week: Philip O’Toole.
Tell us about yourself.
I am an experienced software engineering leader originally from Ireland, now based in the Greater Pittsburgh area. I like to write about software engineering, distributed systems, databases, and the development teams that build them. I am the creator of rqlite, an open-source distributed database built on SQLite. I also work at Google, managing software development teams building large scale Observability systems.
Normally a late adopter when it comes to tech, I’m a big believer in LLMs and coding. I’ve taken up the tech much quicker than I have with any other tech.
ChatGPT, Gemini, or Claude?
ChatGPT web app and GitHub Copilot. Not using any CLI-based tools yet (just haven’t gotten around to it).
What was your last SFW AI conversation?
Non-software-dev prompt:
I am an engineering manager responsible for the delivery of a cross team project. I have a set of priority zero customer user journeys. I also have a bunch of technical tasks that need done for each CUJ. I also have a set of milestones. The CUJ’s are the core functionality we will deliver. The milestones are important progress we make towards delivering those milestones. The technical tasks are a list of tasks that need to be done, many of which are common across all of the CJ’s. What is the right way to organize this work so it can be tracked correctly?
I wanted to check my thinking about some project organization and planning I had to do at Google.
An example of a recent software-development prompt is here: https://github.com/rqlite/rqlite/issues/2273
First "aha!" moment with AI?
When I realised I could scale myself horizontally. The code the agents create is OK, sometimes not great, but I can have it work on many things at once. That makes up for it.
See https://philipotoole.com/rqlite-development-the-agents-are-here/
Also this: https://philipotoole.com/gpt-is-writing-the-github-issues-copilot-is-fixing-them/
Your AI subscriptions and rough monthly spend?
$60 in total. $20 on GPT Plus, and $40 on Copilot. I’d pay more if I had to, the consumer surplus is substantial considering I only have to pay $60.
Who do you read/listen to to stay current on AI?
Hacker News – and then read links from there.
Simon Wilson, the big players’ blogs, occasionally TechMeme.
Exponential View podcast isn’t bad when it talks AI.
Your most-used GPT/Project/Gem?
I don’t have any (assuming I understand the question).
The AI task that would've seemed like magic two years ago but now feels routine?
Agents fixing simple bugs in my open-source software. Just assigning issues to GitHub Copilot and having it write the code. Interacting with machines as easily as I would with people.
Magic wand feature request?
A meta-tool to manage all the Agent-generated PRs. GitHub’s UI may not cut it soon.
If you could only invest in one company to ride the AI wave, who would it be?
Google. Even though I’m employed there (biased?) they’ve got a world-leading presence in the whole stack (TPUs, hardware, DCs, models, cloud, software).
Have you tried full self-driving yet?
Backseat of a Tesla.
What did I think? As a software developer I was nervous since I know how software is written. But I was also aware that it’s probably safer than a human.
Latest AI rabbit hole?
Discussing with GPT the possible behind-the-scenes dynamic of Bachelor in Paradise (which I watch with my wife).
One piece of advice for folks wanting to get deeper into AI?
As it relates to software? Solid software development experience.
And time. The field is moving so fast, it’s hard to keep up.
Who do you want to read a Tokens & Tactics interview from?
If you have any questions, please be in touch.
Thanks for reading,
Noah and Claire