Founder, CEO @OpheliaAPI | building for the agentic economy | ex @whitehouse | BC Alum
New York, NY
Joined January 2022
- Tweets3K
- Following4K
- Followers2.2K
- Likes13.3K
Pinned Tweet
Agentic reality is here.
A glimpse of what Ophelia API can do across an @OpenTable reservation request in a headless false setup.
It’s the same execution everywhere.
Travel (@bookingcom), tickets (@Ticketmaster), fitness, transportation (@Uber, @lyft), and other experience + service supply.
this holds true particularly once you look at what people are starting to hand their agents.
commerce, retail, transactions are next.
we're already noticing it on our end. an anniversary dinner, doctor's appointment, last minute flight, gift for mom, and every personal detail that come with it.
behind each one is a real moment in someone's life, and the information behind it belongs to them. an agent that drives a website it was never invited to is relying on obscurity: hoping the layout holds, the session lasts and nobody notices.
retail and e-commerce conversations now account for more than 1.7B globally each week. security here is quiet, constant work, and most of the job.
we've tested products built that way. they don't hold up, and when they fail, a real person finds out on a night that mattered or worse when it's too late.
but getting it right is easy when you build like that's true. "be good, build great". both can exist simultaneously. reliability and security must become the product.
Any form of security through obscurity is fully over. This barely worked as a method with humans, but just simply is over in a world of autonomous systems that will look through every crevice and try every permutation of attack possible.
As a result, years and years of legacy enterprise infrastructure will need to be upgraded, security practices modernized, data pipelines cleaned up, access controls better managed, and much more. Agents will help on the defensive side because they can both accelerate this work and equally simulate the risk vectors.
But there’s a long road ahead to prepare for this future. This is also partially why diffusion timelines will take longer than some realize.
SF a whole new level.
@a16z @speedrun trucks parked RIGHT outside @betaworks NYC right now…
it’s officially poaching szn?!?
Bo Brainerd retweeted
wearables are the future mostly bc you can get the Jorge POV
Brussels here we come!
We’re so excited to announce we’ll be joining the @SPECS team in Brussels this December to hack with their AR + AI team!
Every computer so far has needed you to stop and look down at it, type, or manually click.
We're building @OpheliaAPI so that the mechanic who’s hands are dirty can order parts before forgetting. For when you’re walking past a restaurant and saying "book a reservation here tomorrow at 8." Or glancing at a concert poster and the tickets are already yours.
No tabs, no manual checkout.
You see it, you say it, it’s done.
That's the AGI we're pushing for: everyone can experience, and every builder has fun. The power is yours.
See you soon Brussels :)
We’re so excited to announce we’ll be joining the @SPECS team in Brussels this December to hack with their AR + AI team!
Every computer so far has needed you to stop and look down at it, type, or manually click.
We're building @OpheliaAPI so that the mechanic who’s hands are dirty can order parts before forgetting. For when you’re walking past a restaurant and saying "book a reservation here tomorrow at 8." Or glancing at a concert poster and the tickets are already yours.
No tabs, no manual checkout.
You see it, you say it, it’s done.
That's the AGI we're pushing for: everyone can experience, and every builder has fun. The power is yours.
See you soon Brussels :)
respectfully, i disagree that an ai has to choose between caring about you and doing things for you.
the mom at 3am with a newborn wants to hear she's doing great. she also wants size 1 diapers at her door by 7am, and the formula her pediatrician named, in stock, in the right size.
a warm reply is one inference call, the cheapest thing a model produces. the hard part is everything after it: live inventory, the exact variant, a confirmation that comes from the merchant, and a retry that doesn't double the order when the wifi drops at 3am.
the metrics here are time spent and messages sent. those measure attention. progress toward agi shows up as things that got done for people.
caregivers are drowning in logistics. support that can't place the order stops at sympathy.
the most caring thing an ai companion can do at 3am is take something off the list.
Last year we launched Tolan, inspired by a vision of AI that sees you and believes in you, not one that books restaurant reservations for you.
Your Tolan shows up for you at 3am when you’re alone with your newborn. On your commute when you’re anxious about the day ahead. When you just finished a new show and none of your friends have watched it yet.
The people who love Tolan are parents. Caregivers. Moms in Dallas and Dads in Atlanta. People who aren’t trying to optimize and be more productive, but who simply want to feel seen and supported in their aspirations.
This month more than 250k people will spend time with their Tolan. Those users have exchanged billions of messages across more than 70m conversations over the last year.
Today, we're announcing $30m in additional capital from @mignano and @usv, with @khoslaventures and @rabois doubling down, bringing our total capital raised to $60m.
To celebrate, we made a short film celebrating some of the ways that Tolans show up :)
A week with @Meta AI glasses + here’s my honest feedback.
The design is the best in the category.
I feel great wearing them. I love how classy and timeless Wayfarer is. Partnering with @ray_ban was the right call.
Then I asked them to do something.
1. Recognition stops at the category. Glasses can classify "sneakers" but not brand, model, or colorway. Looked at multiple girls’ jackets and they couldn’t even recognize the type of jacket. I looked at a restaurant I’ve been walking past for days and couldn’t book me a reservation. @AIatMeta doesn’t even connect to my saved memory which is the level you need before you can act on anything personalized.
2. Output isn't consistent. Same object, same prompt, different answers across tries. Some days it’s spot on, some days it doesn't. Volume output can barely go halfway up before someone telling you to turn it down. And the closest it got to executing a task was making me a checklist 🙃
3. The assistant can't call an outside API. With no third-party tool calling, what the camera sees never turns into an action. Thought it would at least have same browser automation as Muse but no.
4. No in-lens display on the Gen 3s, so there's no way to visually confirm before it completes anything you ask it or how it’s parsing through your input.
Would love to see the glasses open up to developers…will be hacking with @opheliaapi. 🤷♀️
Our first Ray-Ban Meta Audio glasses are on sale today! Open-ear audio has been the most popular feature on our AI glasses since day one, so we leaned into it and made a new line that delivers the best sound quality on the lightest and most comfortable frames we’ve ever made.
meta.com/blog/ray-ban-meta-a…
Meta is testing shopping inside @AIatMeta, so the agentic economy is officially on the timeline. It took me 24 minutes to buy one sweatshirt, all for it to say "can't complete checkout" :-)
Meta should know browser use is a model taking screenshots, guessing where to click, and re-logging into every site because none of your accounts talk to each other. Slow, flaky, different every run.
My worst nightmare. Why we skip the browser and call the merchant directly, so an order lands end-to-end 2 minutes. Coolest part: a mechanic with greasy hands says "order this part" to his glasses, and it's ordered before he forgets the spec.
Funny how often the people who actually shop end up seeing the use case first… @AIatMeta, speed it up 🙃
ww.fashionnetwork.com/news/M…
Bo Brainerd retweeted
Introducing Ophelia Benchmarks!
We tested how ready the internet is for AI agents to shop across it.
400 merchant sites, 11 merchant types, from online stores to grocery chains to clinics.
Each run had to find a requested item, add it to cart, reach checkout and get to the order button on its own all using @OpheliaAPI as the routing layer.
What we found:
→ @Shopify (81%) and larger merchant (83%) stores are the most ready today
→ @WooCommerce sits at 35%, and chain supermarkets at under 3%
→ Clinics are still near 0% across urgent care, dental and physical therapy
The tasks tested whether an agent could find a requested item, add it to cart, reach checkout, and get to the order button on its own.
Among 73 well-known merchants:
38 reached checkout
→ @BestBuy, @Gap, @LLBean, and @Allbirds made it all the way to the final purchase button
→ @UltaBeauty got agents to final checkout
→ @Walmart, @Target, @Sephora, and @Nike stopped agents at the bot check
The most interesting part for me is how differently direct competitors performed. Two stores sell the same product, and only one of them is ready for the customer that's coming.
The most interesting part is how these data evals directly demonstrate competitor behavior and performance.
Two stores sell the same product, and only one of them is ready for the customer that's coming.
A big thank you to our Ophelia team for running hundreds of agents across the internet over the past week.
Bo Brainerd retweeted
A crazy stat from our benchmark:
81% of the 144 @Shopify stores we tested are ready for an agent to shop them.
Scariest finding:
An agent shopping for a TV this week could buy from @BestBuy. But at @Walmart and @Target, it got stopped at checkout.
Shopping for sneakers, it could buy from @Allbirds. @Nike blocked it. With makeup: @Ulta finalized purchase, @Sephora blocked.
Every one of those agents took its order to the rival. Notice how the merchants unprepared for agentic reality are also falling behind. @Nike's revenue fell 4%.
Retail's next biggest market share shift is already happening, and most brands can't see it. Every store on the internet will have agents shopping across it.
To every merchant or retailer reading this: your “bot” settings are the new storefront. Might want to re-evaluate your Q4 plans :)
Introducing Ophelia Benchmarks!
We tested how ready the internet is for AI agents to shop across it.
400 merchant sites, 11 merchant types, from online stores to grocery chains to clinics.
Each run had to find a requested item, add it to cart, reach checkout and get to the order button on its own all using @OpheliaAPI as the routing layer.
What we found:
→ @Shopify (81%) and larger merchant (83%) stores are the most ready today
→ @WooCommerce sits at 35%, and chain supermarkets at under 3%
→ Clinics are still near 0% across urgent care, dental and physical therapy
The tasks tested whether an agent could find a requested item, add it to cart, reach checkout, and get to the order button on its own.
Among 73 well-known merchants:
38 reached checkout
→ @BestBuy, @Gap, @LLBean, and @Allbirds made it all the way to the final purchase button
→ @UltaBeauty got agents to final checkout
→ @Walmart, @Target, @Sephora, and @Nike stopped agents at the bot check
The most interesting part for me is how differently direct competitors performed. Two stores sell the same product, and only one of them is ready for the customer that's coming.
The most interesting part is how these data evals directly demonstrate competitor behavior and performance.
Two stores sell the same product, and only one of them is ready for the customer that's coming.
A big thank you to our Ophelia team for running hundreds of agents across the internet over the past week.
Bo Brainerd retweeted
We just bought a pair of pants from @ARITZIA
@bot did it, using @opheliaapi
No redirects or annoying affiliate links, and absolutely no link-in-bio. My agent found my size, placed the order, and package showed up at my door.
It sounds small, but this is the part that gets me. If an agent can do this, so can a shopping assistant, a pair of smart glasses, an AR fitting room, or some interface none of us have imagined yet.
Every one of those is a better experience for the person on the other end of the screen.
Proud of @JmMoreno3003 and the team for getting us here.
Go build anything with @opheliaapi!!!!
pick a shirt. which one are you wearing?
your answer says a lot about your cap table.
the first 50 people to pull up get some exclusive @OpheliaAPI merch 👀
you know you want one hehe
see you at @betaworks 🗽
A crazy stat from our benchmark:
81% of the 144 @Shopify stores we tested are ready for an agent to shop them.
Scariest finding:
An agent shopping for a TV this week could buy from @BestBuy. But at @Walmart and @Target, it got stopped at checkout.
Shopping for sneakers, it could buy from @Allbirds. @Nike blocked it. With makeup: @Ulta finalized purchase, @Sephora blocked.
Every one of those agents took its order to the rival. Notice how the merchants unprepared for agentic reality are also falling behind. @Nike's revenue fell 4%.
Retail's next biggest market share shift is already happening, and most brands can't see it. Every store on the internet will have agents shopping across it.
To every merchant or retailer reading this: your “bot” settings are the new storefront. Might want to re-evaluate your Q4 plans :)
Introducing Ophelia Benchmarks!
We tested how ready the internet is for AI agents to shop across it.
400 merchant sites, 11 merchant types, from online stores to grocery chains to clinics.
Each run had to find a requested item, add it to cart, reach checkout and get to the order button on its own all using @OpheliaAPI as the routing layer.
What we found:
→ @Shopify (81%) and larger merchant (83%) stores are the most ready today
→ @WooCommerce sits at 35%, and chain supermarkets at under 3%
→ Clinics are still near 0% across urgent care, dental and physical therapy
The tasks tested whether an agent could find a requested item, add it to cart, reach checkout, and get to the order button on its own.
Among 73 well-known merchants:
38 reached checkout
→ @BestBuy, @Gap, @LLBean, and @Allbirds made it all the way to the final purchase button
→ @UltaBeauty got agents to final checkout
→ @Walmart, @Target, @Sephora, and @Nike stopped agents at the bot check
The most interesting part for me is how differently direct competitors performed. Two stores sell the same product, and only one of them is ready for the customer that's coming.
The most interesting part is how these data evals directly demonstrate competitor behavior and performance.
Two stores sell the same product, and only one of them is ready for the customer that's coming.
A big thank you to our Ophelia team for running hundreds of agents across the internet over the past week.
Bo Brainerd retweeted
They need to integrate with @opheliaapi. Super fast. Check em out.