👋 Howdy to the 1,965 new legends who joined since our last edition! You are now part of a 562,112-strong tribe outperforming the competition together.

LATEST POSTS 📚

If you’re new, not yet a subscriber, or just plain missed it, here are some of our recent editions.

👾 People First Equity Plans. How we gave away 20% of Athyna and why I'd do it again.
💳 Piloting Mercury Through The AI Era. An interview with Ryan Wiggins, VP of Product at Mercury.
🔥 Startmate: The First Yes A Founder Hears. An interview with Phoebe Pincus, CEO at Startmate.

PARTNERS 💫

Framer helps teams design, build, and launch their marketing sites lightning fast. With the ability to publish hundreds of CMS pages in a single click, operate at a global scale with seamless localization, and even host unified content across multiple domains, teams have never been able to ship faster.

Trusted by companies like Miro, Bilt, and Perplexity.

Wispr Flow spent two years convincing tech to stop typing. Notetaker is their first product beyond dictation.

No bot joins your call. It gets the words and the speakers right, then turns an hour of talking into decisions, dates, and next steps. It plugs into Claude, ChatGPT, or AI app of choice via MCP, too.

Their pitch is fair: run it beside whatever you use today. Use promo code OPEN for 1 month of Pro for unlimited meetings in Notetaker and unlimited words in Wispr Flow dictation. Available on Mac and Windows.

_{Interested in sponsoring these emails? See our partnership options here.}

HOUSEKEEPING 📨

A buddy of mine launched quicker.chat the other day. It’s one inbox for all your DMs: LinkedIn, Slack, WhatsApp, X, iMessage, with more coming soon. You can literally read, reply, tag, search, archive, and set reminders, all in one place. No more context switching between apps.

If you like being organized and only having to check the important messages, instead of all the noise, you should download it. I’ve been using it, and it’s been pretty great. If you want to check it for yourself, join the waitlist at quicker.chat today. Now enjoy today’s piece!

COLLECTION 🏡

How Are You Infusing AI Into Day-To-Day Operations?

A year ago, 'how are you using AI?' was almost a trick question. You'd get a slightly sheepish answer and the sense that everyone was nodding along to something nobody had really wired into their actual work yet. That's changed, and fast. Walk into one of these companies today, and AI is literally the surface where the work happens.

Today’s piece is a summary of what founders have told me recently about how they’re using AI day to day. The answers run all over the map. Data analyst agents that outperform the founder who built them. Whole go-to-market campaigns run start to finish by agents. Claude Code is open on a second monitor at basically every desk. But the thread tying them together is the same—these teams have quietly moved AI from 'a thing we use' to 'the way we work.' Let's get into it.

Keith Peiris, Co-Founder & CEO at Lightfield

The forcing function for AI inside Lightfield is that we're not trying to build a 10,000-person company. We have the aspiration of having more revenue and more growth than Salesforce, but we don't want to be Salesforce-sized. I want this company to be 500 people, not 10k. If you impose those headcount limitations on the team but you enable technology, you start to see whether the incentives can create the outcomes you're looking for. We're almost 40 people, and we still don't have a recruiter. That's one of the sources of pressure. If it's really easy to hire people, you'll just hire people to solve your problems. If it's hard to hire people, you end up coming up with creative solutions.

Inside product and engineering, it's fairly clear. We've moved everything to a mono repository. We put product specs into it so Claude can understand them, we deploy agents everywhere, and our team burns lots of tokens. Outside of product and engineering, it's been more challenging, and really interesting. It turns out we can get pretty far with AI. For example, maybe I don’t really need a head of finance if my chief of staff with a finance background can get really far using spreadsheets and context.

In addition, we record all of our internal meetings and dump them into Lightfield, so it has a knowledge base of everything we've ever discussed and everything we've ever decided. It turns out that if Lightfield has all of that context, it can do some good work for you. We fed a lot of that context into Paraform, and we're getting good candidates from it. Because of all that, we have about five people outside of engineering and product, and we've been able to get really far without any more.

_{*To see our full interview with Keith, head over here.}

Phoebe Pincus, CEO at Startmate

Everyone would have their own individual answer, but as a team, the biggest way we're using it is productizing a lot of what we do on the community side. Building products around things like selection and how we go from a thousand companies down to 20 in three weeks.

We can do that so much more effectively because we've used AI to help us build a platform and run a process where hundreds of people are involved; it all runs smoothly, and everyone stays really engaged with it. The visibility is great. It also helps unearth companies we might otherwise miss. It doesn't make investment decisions for us, but it can flag that it understands our framework and thinks we might be missing something.

So at a team level, it's basically taking the problems we had managing a community at scale and letting us productize some of them.

_{*To see our full interview with Phoebe, head over here.}

Wiley Jones, Co-Founder & CEO at DOSS

It's easy for me to spend a lot of time on it because I'm just so curious. It's consuming. And then what I do is really force people in our company to learn to do things differently.

A good example is our marketing and sales team. I have the whole company using Claude Code. I was going to our sales and marketing team, showing them exactly how to start using it in their jobs. Even with someone on our marketing team who was asking about website edits, I just set it up for them and showed them how to use it. What kinds of questions would you normally ask me? Ask it instead. And now they're going in and making updates on how our CMS connects into the marketing site; something that normally an agency would be doing. Great, we fired our agency because the people on our team can do it now, and they're actually faster.

It's holding people's hands to the stove and saying, "We will do these things, here is how we do them, I will help you, and once I help you, you are enabled to go do it yourself." I think people underrate the amount of energy and aggression required to drive an organization through change. Treating AI not as a tool, but as a way of working has been extremely helpful in driving productivity in areas where it normally isn't there.

_{*To see our full interview with Wiley, head over here.}

Andrew Busse, VP of AI Operations at Hyperagent (by Airtable)

I have two favorite agentic workflows at the moment, and I'll admit the first is a simpler one. The workflow that has saved me the most time is our data analyst agent. I used to spend a lot of time writing SQL. I know it sounds like something you could just use a chatbot for, but the way we've structured this (with skills, a repo of cascading skill retrieval, and specific definitions for things like how we calculate MRR).

The second is the Founding 500 campaign. We gave $10 million in inference credits to 500 founders, and almost the entire end-to-end campaign was run agentically—all the outreach to founders and influencers, building the landing page we went to market with, all the early qualification before a human did final review. A campaign like that, I would have needed multiple people to pull together. Instead, it was me building one agent, and now it's reusable for any campaign we want to run in the future.

_{*To see our full interview with Andrew, head over here.}

Max Marchione, Founder & CEO at Superpower

If you walk by anyone's desk at Superpower, you'll see a Claude Code window on one monitor and their normal work on another. Everyone is using Claude Code. I don't know a single person on the team who doesn't. That can look like building mini widgets and dashboards. It can look like helping with writing. It can look like writing code. It can look like automating things. That's just common for us. We increasingly build internal tools rather than use external tools. We want to get to the point where the AI is autonomous. I don't think we've done that yet. What I mean is: rather than the AI writing an SEO article, how do we have the AI running the SEO function? We're not quite there, but that's what we're working on.

On the mindset piece, the first thing is to hire young, cracked people; they tend to be very AI-forward. The second thing is, we have a guy who runs AI across the team. He champions AI, gets people set up, helps them automate tasks, and shows them how to use Claude Code. The third thing is to try to entrench it within the culture. We did 'AI Hero of the Week' for a couple of months, where at all-hands we'd have whoever the AI hero was get up and share what they built. And then probably the final thing is: let Ajay just cook cool AI stuff, and other people see Ajay cooking cool AI stuff and decide to copy.

_{*To see our full interview with Max, head over here.}

Bryan McCann, Co-Founder at You.com

The first thing is to have the mindset that you should use agents for just about everything. Many of our engineers have told me that they barely write new code anymore, that they're mostly interacting with Claude Code, or Cursor with Claude running, or something to that effect. Their time is already spent much more on thinking about what they should be doing to have impact, rather than on implementing and writing the code themselves. That has been a huge shift. A year and a half ago, I would have said we hardly used any coding tools, and people mostly thought they were great for boilerplate at best. Now, some people are not writing code at all. That has been very interesting to watch.

It has also been interesting to see how this extends beyond engineering. As a founder, I have been actively encouraging our sales and marketing teams to use these tools as well. For example, every time we go into a customer conversation, the notes, the decks, everything should be much more customized and personalized, not only to that customer, but to that specific meeting. There have even been cases where I'm in a meeting with a salesperson and a customer; the customer asks for something, and I'm running Claude Code in the background so that by the time we're done talking, I can already show them a demo.

The mindset I try to instill is similar to something I carried over from my time in research: the goal was always to keep the GPUs running, because the more they're working, the more they're working for you, and the more you can learn and adjust your direction. It's the same idea here. Always have something working in the background, and instill that mindset in people across every department, regardless of which specific agent they use. In terms of structure, it still primarily manifests as one-to-one—individual employees working with an agent to accelerate their own work.

_{*To see our full interview with Bryan, head over here.}

Zach Lloyd, Founder & CEO at Warp

The simplest way to put it is that we use Warp to build Warp. Every software developer at the company starts their work through a prompt in our own product, whether they are writing new code, fixing a bug, or investigating an issue.

The agent inside Warp helps break down the task, generate code, and guide the workflow. It creates a great feedback loop. As the product improves, our engineering speed improves, giving us more insight into how to make the product even better. It is a very tight cycle, and it pushes us to build features that are genuinely useful because we rely on them every day.

_{*To see our full interview with Zach, head over here.}

Scott Chetham, Co-Founder & CEO at Faro Health

It really depends on the team. As an AI company, we're all in on AI, but even internally, we're all over the spectrum. Some teams have been building their own tooling to automate their work, and what they've achieved has been pretty remarkable. Other teams, like our finance team, are not quite as far along.

Personally, I switch between different tools depending on the task. I've built a virtual version of myself for email responses that lives in ChatGPT, which works really well. For other things, I use Anthropic, and for some financial tasks, I've actually found that Gemini works better. I tend to be fairly agnostic and use whatever tool performs best for a given job. The email assistant has been a particular game-changer. It runs as a private instance, so it's secure, and it has a full history of my writing style. I can now feed it a few bullet points and some background context, and with a quick refinement pass, I can produce a polished email in a fraction of the time it would take me to write it from scratch. It's not thinking for me; I'm putting my thoughts down as bullet points and building a narrative around them. That has been a huge time multiplier. I'm constantly pushing my teams to automate more, and we're on that journey, but as I said, we're at different stages depending on the function.

_{*To see our full interview with Scott, head over here.}

Andrew Mok, CMO at HeyGen

We have an agent that lives in Slack. We developed a forked version of OpenClaw so that it's safer and more secure, and our security team has been the pioneer in setting it up. It basically exists as another team member in Slack. We have a bunch of these agents on Slack, and if you look at our Slack instance, it's kind of equal parts conversations from real people and equal parts conversations from AI agents, and everybody's interacting. It's pretty amazing, actually. The one dedicated to the marketing team is called Mara. She does a lot.

I found that as a leader and an executive, you spend a lot of your time doing updates and readouts because it's important to give everybody visibility and connect the dots. I have workflows set up with our agent to do all of that for me, and I just review it.

On a weekly basis, it generates this long report. We put it into our internal knowledge management system, where it creates a little web app. Of course, I use HeyGen, so I generate a talking video of myself. It's 90 seconds, with me explaining what happened, our current priorities and key insights, and it includes motion graphics and visuals. It generates all of this based on input that other team members put together. They also generate it with agents. It essentially creates this whole cascading workflow.

The great thing about it is that it becomes this knowledge base that keeps track of everything happening within the company and the team. If I ever forget something, or the document doesn't have enough detail, and I want to ask, "What was John working on last week?" or "John mentioned something about podcasts, what did he say and what were the numbers?", I can just have a conversation with our agent, and it usually gives me the right answer. I almost think of it as this brain that knows everything. There's no bottleneck anymore in terms of information flow because the agent is all-knowing and connected to everything. You should be able to have context about pretty much everything going on just by having a conversation with the agent.

_{*To see our full interview with Andrew, head over here.}

Adrian Blair, CEO at Trustpilot 

We are using it all over the place. Obviously, we are coding with it—I think everyone is doing that now. We use Claude Code in engineering, and the finance team is all using it as well. We have got a lot of use cases in customer experience now too, but we are really encouraging the whole organization to innovate in their own area. And as I mentioned, with that frontline focus, it is often the folks who are closest to the front line who come up with the best ideas.

Within customer care, being able to take the customer to something helpful, to the right bit of content, much more quickly than was possible in the past. And within finance, some of the analysis we can now do at a far more rapid pace than was ever possible before. I think that kind of thing is interesting. The one I don't appreciate is content where people have basically had a paper written by AI. I think we all have to be a bit careful with that because, of course, it will sound terribly clever and intelligent, but ultimately I want to know what the person who wrote it actually thinks. And I do wonder sometimes, when I get stuff, whether this is your judgment or is this Claude's.

_{*To see our full interview with Adrian, head over here.}

Extra reading

• Arctic Vaults, AI Agents & The Future Of Dev Tools - June, 2025

• Launching Athyna Intelligence - January, 2026

• Neighbors Feeding Neighbors, At Scale - February, 2026

And that's it! You can find all our original interviews with the founders and leaders above here.

BRAIN FOOD 🧠 

TWEETS OF THE WEEK 🐣 

TOOLS WE RECOMMEND 🛠️

Every week, we highlight tools we like and those we actually use inside our business and give them an honest review. Today, we are highlighting ZeroClick—which allows sellers to turn any API or offering into an agent-purchasable service with controls for pricing, identity, analytics, and more.

See the full set of tools we use inside Athyna & Open Source CEO here.

HOW I CAN HELP 🥳

P.S. Want to work together?

1. Hiring global talent: If you’re hiring tech, business or ops talent and want to do it 80% less, check out my startup, Athyna. 🌏

2. See my tech stack: Find our suite of tools & resources for both this newsletter and Athyna here. 🧰 

3. Reach an audience of tech leaders: Advertise with us if you want to get in front of founders, investors and leaders in tech. 👀