built watermelon ui (100k+ monthly users). now shipping 10 products in 90 days, $0 to $25k mrr by dec 31. 100+ mvps for yc & a16z founders @xocketHQ
SF / AMD
Joined July 2024
- Tweets10.5K
- Following2.4K
- Followers6.4K
- Likes11.8K
vansh patel π retweeted
How much revenue are database companies losing because agents arenβt able to sign up autonomously?
Today, weβre 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 didnβt count.
What we found:
β @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%.
β 7 of 18 products had zero successful sessions across every agent tested.
β CAPTCHAs stopped 373 sessions. Bot blocks stopped another 105. Other barriers included rejected agent email domains and OAuth-only flows.
β 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 βuse this databaseβ to a working result is part of your conversion funnel.
See the full leaderboard, methodology and session transcripts: tryscope.com/leaderboard/datβ¦
Special thank you to our partners @AgentMail and @Agentphone for giving us the right tools to run this experiment.
π€ Made with AI
pancake mapped the go-to-market of 40 startups into "GTM brains" you can copy and hand to your ai agent.
agents with a GTM brain scored 88% of the best possible marketing plan. without one, just 48%.
this is what cursor's brain looks like. market profile, personas, ICP, message pillars, competitors, keyword portfolio. all mapped and ready to copy into your agent.
gtmbrain.si
made 40 GTM brains with Pancake
copy a brain and give that context to your agent
gtmbrain.si
our research on how context affects GTM recommendations below
ai agents are already influencing what people buy. muse, instinct, grok bot. $10B+ a year in purchases.
but brands have zero idea what these agents recommend or why they got skipped.
ACO lets you see the actual queries buyers ask, which brands get recommended, and why yours was not chosen. that last part is the most interesting. you get the reason and can fix it.
only 100 brands getting access. agentconversion.ai
Today, we're introducing Agent Conversion Optimization β agentconversion.ai
Muse, Instinct, Grok Bot already process $10B+ a year in GMV via agentic commerce. No GEO tool today tracks who they recommend.
So we built it.
Each agent behaves very differently from typical LLMs, so we track them separately, then Helena deploys the fixes.
1/ Daily tracking of who Muse, Instinct, Dots recommend, and why
2/ Share of recommendations vs your competitors
3/ Agent path to check out audit
4/ Fixes deployed by Helena to improve your discoverability
This is hard to do reliably with little bias given how these agents behave, so we're starting with only 100 initial customers.
Demo:
ai agents are already influencing what people buy. muse, instinct, grok bot. $10B+ a year in purchases.
but brands have zero idea what these agents recommend or why they got skipped.
ACO lets you see the actual queries buyers ask, which brands get recommended, and why yours was not chosen. that last part is the most interesting. you get the reason and can fix it.
only 100 brands getting access. agentconversion.ai
π€ Paid partnership
Today, we're introducing Agent Conversion Optimization β agentconversion.ai
Muse, Instinct, Grok Bot already process $10B+ a year in GMV via agentic commerce. No GEO tool today tracks who they recommend.
So we built it.
Each agent behaves very differently from typical LLMs, so we track them separately, then Helena deploys the fixes.
1/ Daily tracking of who Muse, Instinct, Dots recommend, and why
2/ Share of recommendations vs your competitors
3/ Agent path to check out audit
4/ Fixes deployed by Helena to improve your discoverability
This is hard to do reliably with little bias given how these agents behave, so we're starting with only 100 initial customers.
Demo:
Peaks: working with some fastest growing SF startups π₯±
& ESOPs
Looking for founder office will switch full time within month
Start with remote and move to office (Ahmedabad) within month, let me know if anyone interested
Knows how to handle operation, knows tech, out reaching, hiring all
Comment with your skills and experience
the most honest record of my abandoned ideas is the renewal email from my domain registrar.
bought at 1 am, sure this was the one.
renewed every year out of guilt.
zero commits.
is one of yours renewing this month too?
Please reply fast or DM me if you are Eligible to work in USA, we have crazyy jobs for you
Related to software engineering, dm me quickly
Hiring finance, consulting and tech experts for remote AI training work, 15+ hrs a week.
IB, PE, VC, hedge funds, top consulting, or SWE / data science at a top company. US, UK, Canada, Australia, NZ, Ireland.
Apply: fynolabs.si/jobs/finance-conβ¦
$100 an hour as a freelancer sounds like $208k a year.
that assumes you bill 40 hours a week, 52 weeks a year.
bill 20 hours a week for 48 weeks: 960 hours, $96k.
then pay your own health insurance, taxes and laptop.
the rate is not the salary.
Hiring finance, consulting and tech experts for remote AI training work, 15+ hrs a week.
IB, PE, VC, hedge funds, top consulting, or SWE / data science at a top company. US, UK, Canada, Australia, NZ, Ireland.
Apply: fynolabs.si/jobs/finance-conβ¦
i don't get the "small take home task" in hiring.
it says 2 to 3 hours.
it turns out to be a full app with auth, tests and a readme.
you lose a weekend, then get a one line rejection.
if you hire, how long does your take home really take?
raise your price 20%, lose 10% of customers.
you still make 8% more.
100 x $50 = $5,000
90 x $60 = $5,400
i was today years old when i caught myself saying "please" to a coding agent at 2 am.
then i said sorry to it for a bad prompt.
it did not care. it fixed the bug anyway.
do you talk to yours like a coworker, or like a terminal?
stripe takes 2.9% + 30 cents per card charge.
a $5 monthly plan: about 45 cents a charge, almost 9% of the price.
the same $60 billed once a year: $2.04, about 3.4%.
the 30 cents is what eats small plans alive.
check bupsy.si
