COO at @linear. previously building at @stripe, @NotionHQ & @firstround. angel investor to ~80 early-stage startups
San Francisco, CA
- Tweets 4,307
- Following 977
- Followers 38,152
- Likes 23,010
Linear Montreal Offsite Day 1 Agenda
Four walls.
85°C.
And everyone agrees on the product roadmap.
7
65
4,074
AI agents may eventually transform shopping, but we still have a long way to go.
Shopping is visual. I want to compare products, browse photos, check sizing, read reviews, and notice things I wasn’t specifically searching for.
That’s why I usually ask the agent for a product link instead of buying through it. I need more data and interrogating a chatbot one question at a time is rarely the fastest way to get it.
What is the agent-equivalent of the storefront?
38
5
85
8,741
I use agents a lot. I do not trust them with my credit card or edit access to my personal inbox or texts.
We’re still in the “wow, it booked a restaurant reservation” phase, not the “here are the keys to my entire life” phase.
I continue to be amazed how the tech industry doesn't realize that the majority of people won't hand over a digital wallet for AI agents to go and spend on stuff, because buying socks + groceries is not a chore to outsource w/o oversight, but an expense to manage...
27
3
182
20,906
@daniel_levine said I got it wrong. He didn’t eat peanut butter sandwiches. He’d just spoon the peanut butter right out of the jar. 😆
4
71
12,756
I find all this dating discourse funny, but I’ll weigh in.
If you’re dating in your 20s, be an angel investor, not a public markets investor.
Look for the guy eating peanut butter sandwiches for lunch who considers Chipotle a fancy dinner because he’s making $30k a year and living in the Bay Area. You’re investing in the *person* and not the current valuation.
At least, that’s what worked for me.
52
15
927
120,658
The future of software development is approving your agent’s code from the school pickup line.
Linear Diffs are coming to mobile.
Sign up for early access:
linear.app/mobile-diffs
22
3,114
Cristina Cordova retweeted
Join our first conf in SF this Dec!
Orbit
by
Linear
for
Linear by Linear Orbit Inc.
7
7
149
24,637
Linear's first-ever conference. One day. The future of building software.
Orbit lands in San Francisco on December 9 🪐
3
4
88
8,942
Dial-up got dramatically cheaper too. But I’m not buying dial-up in 2026. I’m paying more for fiber because my expectations for what the internet should do went up with the technology.
My bet is intelligence works the same way: yesterday’s frontier becomes cheap, while we keep paying a premium for tomorrow’s frontier.
I agree the cost of today’s intelligence will keep plummeting. I just don’t think we’ll want today’s intelligence tomorrow.
Future generations will hear stories of a brief period when $200/month bought you seemingly unlimited access to the world’s most expensive intelligence, the way we talk about $2 Ubers now.
8
1
53
6,794
A $7B healthcare AI company left a certain legacy issue tracker for Linear. They’ve since logged 80,000+ customer requests and built a system to weigh them by frequency and revenue. linear.app/customers/commure
2
1
48
3,951
Future generations will hear stories of a brief period when $200/month bought you seemingly unlimited access to the world’s most expensive intelligence, the way we talk about $2 Ubers now.
117
47
1,526
235,199
If Miro had raised $50M total, last at a $20B valuation, and sold for $1.79B in equity value, would we tell a different story about the outcome? They're profitable and still have substantial cash ($435M). The capital wasn’t simply lit on fire.
I don’t love overcapitalizing companies, and we’ve chosen not to at Linear when we could have. But "they raised too much" is sometimes a convenient morality tale, not an explanation for why a business with ~$600M in ARR sold for $1.79B.
Darling of 2021, ~20B valuation, 500m raised
They don’t tell you this one but “too much capital” is equally likely, if not more, to kill a company than too little
You get drunk assuming the party never ends and then boom, you get wiped out
9
7
185
76,486
At the most product-forward companies, the whole company is starting to work like an engineering team. Marketing, recruiting, finance, ops. The work is visible, owned, and increasingly done with agents.
I run six teams at Warp. Not one of them is engineering. All of them run like engineering teams.
Marketing, benefits, people, talent, partnerships, ops. All six on @linear, all six in @claudeai Code.
I was a mechanical engineer at Apple, a software engineer at Facebook, then a PM at Google. The most useful thing I took out of all three was watching how those engineering teams worked. Five things they all did:
1/ Everything is written down. Every task has an owner and a state and anyone can go look at it. That sounds like bureaucracy but it does the opposite. Writing work down is what lets ten people work in parallel without a layer of humans in the middle asking each other for updates.
2/ Doing the same thing by hand three times is a bug. An engineer hits the third repetition and writes a script. In ops you can hit the three hundredth and still call it your job.
3/ Deadlines come from breaking the work apart into timed pieces. You cannot put a real date on something you have not decomposed, and the decomposition is the estimate. Every optimistic timeline I have ever been handed was someone skipping that step and hoping.
4/ You ask for the outcome and stay out of the method. People stop coming for permission once the request is clear enough that there is nothing left to ask.
5/ When something breaks, the conversation is about the system that let it break. In engineering that is called a post-mortem.
Because of these 5 things, I have changed how my teams operate.
Every week, I run a standup with each team around their Linear board. What happened last week, what we are focused on this week, feedback to each other, what is coming. Every engineering org runs this. But almost no operating team does.
The usual answer for a high-volume team is to hire an ops person to track the work.
I think my people should just be their own project manager. They have ten times the context on what actually matters this week. You lose an entire layer of admin and the prioritization gets better, not worse.
None of this would have worked three years ago. The tooling was not there. Now the person who needs the thing can just build the thing, and the work lands in Linear from whoever did it rather than from someone tracking them.
The surprise was how fast my team started getting addicted to Linear. Not because they love adding process, but because for most of them this was the first time their work was visible to everyone else.
Most operating problems are engineering problems that nobody bothered to engineer.
10
7
191
38,680
Now that @daniel_levine is deploying personal capital into startups, he’s just live-texting me every check he cuts at YC alumni demo day.
Very lean fund structure. One GP, one involuntary LP.
20
311
45,502
Looking forward to joining some great product, ops and marketing leaders to discuss how to make your brand inevitable on Oct 7th.
For Linear, it's meant becoming the default choice for every new company and the eventual destination for everyone else.
Join us for an SF Techweek fireside chat hosted by Warp and @linear with a few of our favorite leaders on what it means and feels like to build an inevitable brand.
Speaker list:
- Ben Shen (Product Lead @ Coinbase)
- Cristina Cordova (COO @ Linear)
- Gaby Garcia (COO @ Warp)
- Hannah Hughes (CMO @ Plaid)
- Michael Duane (Product Lead @ Cash App)
Join us for a discussion on what it takes to build an inevitable brand for both humans and agents, as well as how to scale technical innovation while still maintaining design simplicity.
Register here: partiful.com/e/2RDOLZR1FZrXr…
4
2
38
5,986
It’s one of those rare 48 hour periods where you don’t need to bring a jacket with you in San Francisco.
8
46
3,185
Cristina Cordova retweeted
Probably nothing
Linear’s share of Jira vs. Linear at @jamdotdev went from ~25% to ~75% in two years. @linear is clearly winning.
1
4
47
8,930
