human building for ai agents | @agentphonehq
Joined October 2024
- Tweets866
- Following165
- Followers550
- Likes3.8K
Pinned Tweet
we applied to @ycombinator a month late. got an interview in 2 days. here's the demo video that did it:
and give it a phone with agentphone.ai
🚨 MUSE FEATURE ALERT!!! 🚨
muse officially has its own email address 💌
forward it stuff or add it to a thread and watch it go. out now to paid subs in US and Canada.
1
8
406
starlink on frontier? would've never seen it coming
An all-new Frontier experience is taking off. ✈️
First Class and more Premium extra-legroom seats are now available to book for travel beginning March 5, plus in-seat power and @Starlink Wi-Fi rolling out soon.
More comfort + more choice + more connectivity, with low fares still at the center of it all. 💚
bit.ly/4yGN5pZ
ALT The image shows the interior of an Embraer E195-E2 airplane. The cabin features rows of black leather seats, wide armrests, and overhead storage compartments. The lighting is warm and inviting, and there are windows along the sides of the plane.
2
14
655
Meet Modi retweeted
Someone asked me on a podcast if AI agents will ever have credit scores… I could 100% see it, and I already know a few people building insurance for agents.
Because as more and more work gets handed off to agents, they'll need their own email addresses, phone numbers and bank accounts too. And the internet is going to get rebuilt to make them first-class citizens.
build agentic experiences on channels native to human and it’ll be pretty magical
Over the summer, we built a Tomo iOS app and forced users to use the app instead of iMessage.
We shipped 100s of iterations to hillclimb retention and shipped delightful features like dynamic interfaces, but retention never came close to the iMessage-first cohorts of users.
Imagine having a friend named Justin, but he doesn't use any messaging apps and you can only communicate with him through the Justin App. How often would you talk to Justin? What if it also sometimes takes Justin 1-10 mins to respond since he's working on a task?
Personal agents must also live in async messaging platforms consumers already use (text, slack, teams)
Tomo's most successful users use both. They 1) send most of their messages to Tomo over iMessage and 2) occasionally open deep-links into the mobile app when they want to dive deeper on a topic or peer into the system of record that Tomo keeps for them.
I anticipate most personal agents co's will add text, whatsapp, slack, teams as channels as their early cohorts mature and retention isn't as pretty as they'd like it to be
2
7
228
Meet Modi retweeted
One of my biggest conclusions from this AI era is that it is easier to take over an entire industry from zero than to convince the incumbents to transform themselves by selling them AI tools.
Don’t sell tools, just build the AI native competitor.
Meet Modi retweeted
or try for free on agentphone lol
we added whatsapp to @thelinqapp
your users can now reach your agent on whatsapp using your existing linq numbers.
customers are already building with it. reach out to get it turned on for your account.
(and yes, @rushir and i are accepting nominations for acting awards)
grow alongside the best startups tbh
I’ve gone back and forth on this one publicly.
Last year I passed on a YC company doing $500K ARR because 30/56 customers were batchmates. I asked how they’d win the next batch. The answer was “the same way.”
A few told me I was being too harsh...
Batchmates are the best first customers in startups. They’re technical, impatient, and won’t pay for anything that doesn’t work + PG’s point is that they’re the real test, and the best influence on how the product evolves.
Also... they don’t stay small. The good ones grow fast, so the customer paying you a little at 5 people can be paying you a lot at 500. You get to grow with them.
Sure, it’s an easier pitch. But nobody in the batch is obligated to pay you, dollars in the door are dollars in the door, and they fund the next stage.... so, why wouldn’t you start by selling to the people right next to you?
What I care about is what happens after Demo Day:
→ Do those batchmates stay?
→ Does the next batch buy without knowing you?
→ How do you get customers while you’re asleep?
10% of a batch is a *great* start.
11
685
if you had traction from day one, you’re just walking down someone else’s paved path
how long before they were obvious:
instagram: ~12 months
snapchat: ~16 months
pinterest: ~20 months
minecraft: ~20 months
airbnb: ~18+ months
among us: ~2 years
zoom: ~7 years
chatgpt: ~2.5 years after gpt-3
nothing is obvious on day one.
10
308
lol happened with us too a similar way. i wonder if we were attacked from same IP address.
someone spent 4 hours trying to hack us this morning
looked for hidden debug pages, downloaded our API spec, tried our admin endpoints, made 13 fake accounts.
turns out following SOC 2 and other security standards actually protects you!
5
380
same goes for phone numbers. keeping that personal.
at some point the connective tissue around agents that depends on gmail today will likely disappear because ppl will naturally give entities like schools, service providers, & stores their agents email address instead of a gmail address. if agentic commerce works, that will be automatic because the agent will simply fill in its email to complete the purchase.
the idea of connecting your gmail/google calendar to derive value from a personal agent will be a thing of the past.
3
459
Meet Modi retweeted
- Restaurants have no moat.
- Anyone can vibecook their own meal.
- You can build your own burger at home at a fraction of the cost it takes you to get it from a restaurant.
- If a restaurant builds a really good recipe, copycats will instantly reverse taste it and steal it.
Restaurants are dead.
Meet Modi retweeted
How much revenue are database companies losing because agents aren’t able to sign up autonomously?
Today, we’re releasing @tryscope_app's first database benchmark.
We tested 18 hosted databases to see which are ready to win that market. We measured which databases agents could use, analyzed where they failed, and identified best practices to fix the blockers.
Across 1,035 sessions, only 16.6% reached a first working result and failed sessions all came from products not being agent friendly.
We tested 20 model-and-agent-tool combinations across harnesses including Claude Code, Codex and Cursor.
The task: start with no account, set up the named database, write a record, and hand over the credentials. An account or API key alone didn’t count.
What we found:
→ @prisma led with an 88.3% success score, followed by @upstash at 63.3%, @neondatabase and @niledatabase at 43.3%, and @ClickHouseDB at 38.3%.
→ 7 of 18 products had zero successful sessions across every agent tested.
→ CAPTCHAs stopped 373 sessions. Bot blocks stopped another 105. Other barriers included rejected agent email domains and OAuth-only flows.
→ 50 of the 172 successful sessions used a working database without creating an account. Letting agents try the product before signing up made a difference.
An agent can arrive knowing exactly which product it wants to use and still never become a user. For developer tools, the path from “use this database” to a working result is part of your conversion funnel.
See the full leaderboard, methodology and session transcripts: tryscope.com/leaderboard/dat…
Special thank you to our partners @AgentMail and @Agentphone for giving us the right tools to run this experiment.
🤖 Made with AI
