Noam Yasour retweeted
Today we’ve raised $52M Seed and we are announcing the public launch of S2.1 Pro.
>It can clone a voice from 5 seconds of audio
>2x faster than Cartesia & 1/6th the cost of Eleven Labs
>most expressive model with word level control over emotion, intonation, pacing etc
We support frontier AI companies including HeyGen, LiveKit, Retell, Sanas, and OpenArt all run our model in production.
If you're a business and we can't cut your voice AI costs by 50%, we'll give you 1 year of Fish Audio for free.
Book a demo: s.fish.audio/tmapke
To celebrate our first birthday, we'll give you 1 month of S2.1 Pro for free. Like, retweet, and comment “Fish” to get it.
Introducing Cognee v1.0: a major breakthrough in agentic intelligence.
It is 145% better than Opus 4.8 and GPT 5.5 at long context memory retrieval.
Cognee allows a 100 BILLION token context window 100,000x more than Claude. It's:
- 6.9x cheaper than GPT 5.5 and Opus 4.8
- Cold starts in 350ms & searches in 260ms
Why this matters:
Today agents forget important context, redo tasks, waste tokens, and slow down as workflows get more complex.
Cognee solves this.
It’s not a place to build agents. It connects to the agents you’ve already built, across any platform, and makes them significantly cheaper, faster, and more accurate.
Here's how it works:
Readers added context they thought people might want to know
Cognee's website and GitHub do not provide benchmarks or evidence supporting claims of 145% better long-context memory retrieval than GPT-5.5/Opus 4.8, a 100 billion token context window, or 6.9x cheaper costs. cognee.ai github.com/topoteretes/co…
I would separate “answered” from “handled.” In storm season, the useful thing is: what happened, where are they, is it urgent, and can the team prioritize the call before the homeowner tries the next number.
teapawt.org/onemanifestal/status/2…
This Post is from a suspended account. Learn more
2
3
57
This sounds boring until you look at the call log. A lot of local businesses are not losing because the offer is bad. They are losing because demand arrived and nobody caught it.
teapawt.org/wasco_alex/status/2058…
Here's the simplest business advice I've ever given:
Answer the phone on the first ring.
Not the second. Not voicemail. FIRST ring.
Half your competitors won't pick up at all. Of the half that do, most take 3+ rings.
You just beat 75% of your competition by picking up the phone.
15 years. 50,000 jobs. That one habit is worth more than any marketing tactic.
1
71
This is the part people underweight. The lead does not need “AI magic”; they need a useful response while the crew is still on a job and before the homeowner starts shopping around.
teapawt.org/noahiglerSEO/status/20…
We rolled out our speed to lead agent for a landscaping company we work with about a month ago.
It's already qualified, nurtured, and handled 30 landscaping leads on its own.
This is the part of running a home service business almost no one warns you about. You spend years getting your marketing to a place where the phone actually rings, and the second it does, you're chained to it.
Every missed call is a lost job.
Every voicemail that sits for 4 hours is a lead that already called the next company on the list.
The owner I work with was running his crews, training new hires, going out to do quotes himself, and trying to catch every inbound call in between. The job he was best at, which is closing high-ticket landscape projects in person, was the one he had the least time for. The lead intake was eating HIS time.
The agent took that whole layer off his plate. It picks up the lead immediately, qualifies the lead, gets project scope and budget, books the consultation, and routes the qualified ones to him with full notes. The goal is the owner walks into the quote with everything he needs to close.
Three sizable leads in the last week, all handled by the agent without him touching them until the actual sit-down.
For home service owners who are already drowning in calls during the season, this is the leverage point I'd be paying attention to.
The marketing problem gets solved... then the operations problem of actually catching and qualifying every lead becomes the next bottleneck.
This is the #1 problem I see with fast-growing home service businesses.
1
312
I now let it ring purposely.
One of our first Clara customers said that to me.
He is a plumber.
And honestly, that line is basically the whole product.
1
1
65
He had thought about hiring a receptionist for years.
The problem is obvious: roughly £30k/year for a full-time hire vs around £50/month for Clara.
For owner-operated trades, the question is not "human or AI?"
It is Clara vs a missed call.
1
16
Speed to lead sounds like sales jargon until you watch a homeowner get two quotes before the first contractor calls back. The loss happens before anyone thinks they are in a sales process.
teapawt.org/michaelheredia/status/…
A contractor gets a message about a roof estimate. By the time they check it, they've also got three new messages. They call back third. Prospect already got two quotes. Bid doesn't even get considered. Lost from the start.
1
36
This is the bit I keep coming back to. For service businesses, useful AI is boring: did we answer, did we get the job type/address/urgency, and did someone get booked before they moved on?
teapawt.org/DaineReid8/status/2057…
Most AI advice is useless for service businesses.
A plumbing, HVAC, electrical or cleaning operator does not need “agents”.
They need:
- quotes followed up
- missed calls handled
- jobs scheduled
- reviews requested
- admin removed from the owner’s head
Boring automation prints first.
48
Simple Friday exercise for a local service business:
Open the call log. Count every person who called and never got a useful response.
That is not a phone problem. It is demand that already arrived, then leaked out before it became a job.
1
37
meet @heyclara_ai
We noticed something that bothered us.
Small business owners: plumbers, electricians, contractors, are losing jobs every single day.
Not because of their work. Not because of their price.
Because they couldn't answer the phone. 🧵
1
1
391
Noam Yasour retweeted
Yesterday, I testified in front of the U.S. House Ways and Means Committee about antisemitism and support for terrorism at @Columbia and many other American universities.
The members of Congress asked some really thoughtful questions.
Here were my responses:
(a thread)
Noam Yasour retweeted
We need deterministic LLM outputs.
An information network.
Graphs!
How to build them with LLMs? 🤔
How to search them?🎓
We help with the new cognee 0.1.4 release!
An optimizer to train a knowledge graph generation with #DSPy and @weaviate_io to get better contexts 💥
👇








