Anthony Scodary retweeted
Today we are announcing @goguavaai's $12.5M seed financing, led by Acrew Capital, with Flourish Ventures, WillowTree Ventures, Clocktower Ventures and Commerce Ventures.
I've been asked why I went back to operating and co-founded Guava. I had been an investor in Gridspace, the lab that preceded Guava, so I knew the team and the technology. A frontier voice lab with a full team of top ML engineers, in a market as big as voice, is rare. This team is extraordinary, and uniquely qualified to deliver state-of-the-art voice.
At Marqeta, I saw firsthand what it takes to ensure reliability at scale in regulated environments and to take real accountability for performance. Guava is built in that same spirit. Voice AI that sounds good in a demo is the easy part. Voice AI that holds up on real calls is the hard part, and it is the job Guava is built for.
Thank you to my co-founders @scodary, Nico Benitez for building the technology, to @MPiechna31 our co-founder and Chief Revenue Officer, and to the whole Guava team. Thank you to the investors who see the future we see: @LaurenKolodny, @EmmalynShaw, @rohit_bodas, Ben Savage, and our pre-seed evangelist @venturedan.
We are growing the team to take Guava to every enterprise and BPO ready to move voice AI from pilot to production. If that's you, come talk to us.
This is true of all statistical models. When you read a weather forecast, you're reading a hallucinated inference to fill a gap in the weather record for days that have not yet occurred.
Language models are a classifier of singular next tokens.
1
5
Everything is a hallucination.
Anil Seth: perception is "a kind of controlled hallucination, in which the brain's predictions are being reined in by sensory information from the world."
1
29
Recently, universities have started flirting with a return to oral exams to LLM-proof the evaluation process in classes. This prevents outright cheating. But the undergrad experience has always made large committed projects an optional part of research.
1
6
If I interviewed someone who had translated an epic poem, I'd trust them more to build a large software system than a computer science major whose experience was limited to problem sets and class projects.
1
1
If your question is who will use modern tools well to do high-quality deep work, the answer is anyone who is going to have the Sitzfleisch to specify, iterate on, and polish complex systems.
1
3
Oftentimes broad world knowledge, a memory for complexity, and endurance are bigger force multipliers. Regardless of your major, being well-rounded and worldly helps a lot.
1
2
Financial Times journalist @gilliantett reported that the work coming out of many "AI native" STEM majors is so shallow that financial firms are increasingly seeking out humanities grads instead.
1
5
One rapidly evolving question in the world of vibe coding is: who do you hire into your engineering organization?
1
6
He worked on a wide array of topics including cosmic rays, atomic physics, quantum field theory, and neutron stars. However, according to Gell-Mann, he never committed long enough to build depth, polish, and work out all the edge cases and details.
1
13
His criticism of Oppenheimer was that many of his papers were short letters investigating the surface of phenomena like black holes, leaving detailed calculations to future work.
1
6




