Gautham Seshadri retweeted
𝗪𝗲 𝗵𝗮𝘃𝗲 𝗮 𝗻𝗲𝘄 𝗶𝗱𝗲𝗻𝘁𝗶𝘁𝘆.
Heres why. When we started Snezzi, we put something scrappy together. A name, a logo, a website. It was enough, because our focus was somewhere else: our customers. Whatever they needed next, we built and shipped.
Since then weve ss many industries. That work taught us who we are, who we serve and what value we add.
So our identity is growing up with us.
Heres where were heading.
Showing up in AI answers and in Google takes consistent work. New pages, site fixes, mentions on the sites AI reads. Every week, not once.
Thats hard to keep up at any size. Launches and campaigns pull teams away, and the weekly work slips.
So were building an agent that does the work. It finds where AI recommends a competitor, writes the pages, fixes the site and earns the mentions. Your team approves what goes public and sees every step.
Our team and our agents already do this for our customers. Now were making that agent available to many more of them, from growing teams to large enterprises.
New look. Same team. Same focus on the customer.
Every brand starts in the valley. Were here to help them climb.
Heres why. When we started Snezzi, we put something scrappy together. A name, a logo, a website. It was enough, because our focus was somewhere else: our customers. Whatever they needed next, we built and shipped.
Since then weve ss many industries. That work taught us who we are, who we serve and what value we add.
So our identity is growing up with us.
Heres where were heading.
Showing up in AI answers and in Google takes consistent work. New pages, site fixes, mentions on the sites AI reads. Every week, not once.
Thats hard to keep up at any size. Launches and campaigns pull teams away, and the weekly work slips.
So were building an agent that does the work. It finds where AI recommends a competitor, writes the pages, fixes the site and earns the mentions. Your team approves what goes public and sees every step.
Our team and our agents already do this for our customers. Now were making that agent available to many more of them, from growing teams to large enterprises.
New look. Same team. Same focus on the customer.
Every brand starts in the valley. Were here to help them climb.
2
4
7
272
If the team can only explain the product from the chat history, the AI rollout already failed.
Keep one person who can still say what “done” means with the laptop closed.
Keep one person who can still say what “done” means with the laptop closed.
0
0
0
31
Cheap decision models are the aluminum. Jev isn’t a faster chat, it’s intelligence cheap enough to live inside the product, not next to it.
If you only bolt a chat model onto last year’s process, you swapped the case.
Ask before you buy: what handoffs and reviews die if a decision costs almost nothing?
x.com/paulg/status/2…
If you only bolt a chat model onto last year’s process, you swapped the case.
Ask before you buy: what handoffs and reviews die if a decision costs almost nothing?
x.com/paulg/status/2…
In the 1950s, an aluminum company commissioned Vacheron & Constantin to make them some aluminum pocket watches. The case, dial, hands, numerals, and even the plates in the movement are aluminum. The resulting watch is 44 mm in diameter but weighs only 22 g.
0
0
0
43
Smart phones put computers in pockets. Internet put libraries on desks. When intelligence gets cheap, it doesn’t stay in a chat window. It goes into every small choice software makes all day.
x.com/completeskepti…
x.com/completeskepti…
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x ns free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
0
0
1
25
If the model gets better every month, the edge that lasts is the data you refuse to upload.
Pharma already figured that out. Most product teams still paste first and regret later.
Pharma already figured that out. Most product teams still paste first and regret later.
0
0
0
7
Talent will chase the best model on its own.
What they try, break, and fix is the part worth owning.
x.com/naval/status/2…
What they try, break, and fix is the part worth owning.
x.com/naval/status/2…
The Frontier Lab flywheel is to get the smartest people to use the leading models.
Then distill their inputs, your own model’s outputs, and resulting synthetic data and environments.
And the people have to use the leading models because they’re locked in mutual competition.
0
0
0
12
Google's Search Console AI reports finished global rollout Aug 31. Impressions, pages, countries, devices. No clicks. No queries.
1
0
0
11
Impressions tell you that you were in the room. Grounding queries tell you which question put you there.
1
0
0
2
Verify in Bing Webmaster Tools even if Bing sends you no traffic. You're not there for traffic. You're there for the query data Google is withholding.
0
0
0
3
if I get ___ done in three months, people will clearly see I did a great job.
if you can’t finish that line, the busy work already owns you.
x.com/zaygranet/stat…
if you can’t finish that line, the busy work already owns you.
x.com/zaygranet/stat…
Thoughts on finding your "golden snitch" x.com/i/article/2017…
0
0
0
20
Most “we need better tools” problems are actually “nobody owns the next step” problems.
New software won’t invent the owner.
New software won’t invent the owner.
0
0
0
6
Krishna didn’t tell Arjuna to switch off his heart.
He told him to think clearly and do the hard thing anyway.
Courage without that is just noise.
Happy Janmashtami.
He told him to think clearly and do the hard thing anyway.
Courage without that is just noise.
Happy Janmashtami.
0
0
0
36
The expensive part isn’t the thinking anymore. It’s deciding what “done” looks like.
If you’re still hiring for the middle of the job, you’re buying last decade’s skill. x.com/Konstantine/st…
If you’re still hiring for the middle of the job, you’re buying last decade’s skill. x.com/Konstantine/st…
0
0
2
46
The stuff you touch every morning is the wrong place to be cheap. x.com/signulll/statu…
my rule of thumb for buying anything is simple. if i use it at least twice a day, i buy the best version i can reasonably afford.
e.g. the best toothbrush, the best sheets, the best shoes, the best chair, the best headphones, the best phone, etc.
maybe more specifically the formula is something like frequency x duration x intimacy with your body/attention. if it touches you, supports you, or mediates reality for hours every day, buy the best version you can afford.
this also applies to product development, you should build products that touch ppl’s lives at least twice a day (when i was at g larry page would often say this rule a lot at tgif).
0
0
0
19
Don't buy AI for work that still lives in someone's head.
If nobody wrote it down, you'll just ask them again next month.
If nobody wrote it down, you'll just ask them again next month.
0
0
1
19
If a personal agent only exists as a loss-leader, it won’t stay personal. Ads, lock-in, or your data will pay the rest.
x.com/signulll/statu…
x.com/signulll/statu…
personal agents will be insanely expensive to run. they’ll need a level of subsidy that makes uber, food delivery, & the entire gig economy look incredibly disciplined.
the opportunity exists cuz frontier labs want enterprise margins & public market narratives but personal agents require consumer pricing (zero), endless inference, & possibly years of loss leading behavior.
i have posted about personal agents for over a year now & the trajectory is pretty clear now.
0
0
0
31