AI agents for concierge customer experiences
San Francisco
Joined January 2024
- Tweets717
- Following30
- Followers7.4K
- Likes965
Decagon retweeted
Devin can do more than ship PRs. @DecagonAI used it to run an inference research lab.
One Devin orchestrates, subagents chase competing hypotheses in parallel, and every idea is tested against replayed production traffic on @modal GPUs. The result was up to 2.5× more throughput per GPU.
Decagon retweeted
Great seeing Devin used to improve inference efficiency, this is something that normally takes many engineers but using Devin to keep looping for improvements is likely how things will be done going forward
Decagon retweeted
I'm hosting a voice/audio research night with @DecagonAI and @Accel on Nov 9. @nishnik from Meta Superintelligence Labs, @chongz from Cartesia, and I are each giving a quick talk then it’s just hanging out with other speech people
if you work on voice in SF come through!
luma.com/j3ddfdsw
Decagon retweeted
Next up on First Pass: A conversation with @EveraertDante, a voice AI researcher at @DecagonAI on Duplex Models
LLMs are turn based. Duplex Models are constantly "listening" streaming audio in and out simultaneously.
Decagon retweeted
Had a blast talking with @jaminball and @palak_go on First Pass about why transcription shouldn’t be on the critical path for voice agents, and what it actually takes to go audio-native. Thanks for having me!
Next up on First Pass: A conversation with @EveraertDante, a voice AI researcher at @DecagonAI on Duplex Models
LLMs are turn based. Duplex Models are constantly "listening" streaming audio in and out simultaneously.
Decagon retweeted
marketing team advised that I shitpost less and try to provide some real value to the world, so here's what I've been working on recently!
Decagon retweeted
i'll be speaking at sglang summit in SF, nov 12-13!
going to walk through the post-training and inference work we’ve been doing at Decagon, and how we scaled out our RL training pipelines to ~1T param models
if you’re going, let’s say hi! sglang.io/summit @sgl_project
Decagon retweeted
Super excited to announce @DecagonAI’s personal agent gateway!
Personal AI assistants like Muse, Instinct, Dots, and Grok Bot are taking off. They’re already starting to buy, book, cancel, and negotiate with businesses on behalf of consumers.
This creates a new challenge for businesses… but blocking them risks turning away customers.
Everyone will need an equally smart agentic front door that can handle interactions with both agents and humans.
Personal Agent Gateway helps enterprises identify personal AI assistants and customize those requests so that you can give a great customer experience while controlling all your logic.
We’re also announcing our PACT protocol that allow personal AI assistants to prove who they represent and what they are empowered to do.
Read more:
decagon.ai/blog/personal-age…
Decagon retweeted
Decagon CEO @thejessezhang predicts that within 10 years, AI agents will become the new websites for every business:
"In a decade, everyone's gonna be primarily interacting through personal agents. Every business is gonna have an AI agent that is the front door. It becomes like a new website or a new user interface."
"To start, those will naturally be natural language interactions 'cause they have to be compatible with both sides. But in the future, agent reaches out, agent receives the call, recognizes that both sides are agents, and there may be some more efficient ways to conduct business."
"Every business is gonna have this, just like how every business has a website now."
@DecagonAI
Decagon retweeted
Decagon CEO @thejessezhang on how enterprise AI went from experiments to real production in a year, with Decagon Dialogues now centered on what the biggest companies actually learned deploying it:
"The focus is not even on Decagon specifically. It's more about how are the biggest brands in the world actually deploying AI. When we have these top-down transformations, how do we measure if they're successful or not?"
"Last year, the conversation was more around, here are the experiments we're running right now, here's what we want to do."
"Traditionally you would think that the bigger they are, the slower they move. But that actually hasn't really been our experience. A lot of them now will have actual learnings and lessons from real production deployments."
@DecagonAI
We're hosting our annual conference, Decagon Dialogues, in SF on 10/1!
@ScottWu46 from Cognition is joining us for a keynote.
We'll have folks from American Airlines, Verizon, Paypal, Chipotle, etc discussing how they're deploying agents at scale.
Come by: decagon.ai/decagon-dialogues…
Decagon retweeted
How do you keep up when you go from 50 to 500 employees in a year?
@thejessezhang shares how @DecagonAI scaled that quickly, from hiring ahead of demand to keeping the company connected as the team grows.
He also explains why the zero-to-one phase is so different from scaling, and how Decagon found its way out of the idea maze in just three weeks.
(0:00) Trailer
(0:46) Introduction
(1:13) Decagon's rapid scaling
(14:26) Why Jesse doesn't like being a solo founder
(22:38) Finding product-market fit with six customers
(31:03) Building resilience for the next downturn
(35:43) Product differentiation from Sierra and Bret Taylor
(44:49) The most misunderstood function: marketing
(1:03:01) Building an agent that can't afford to get it wrong
Decagon retweeted
I'm building a new team at @DecagonAI: Engineering Lead, Strategic Deals 🚀
After founding & selling my last company, I was worried about how I'd replicate the same autonomy and agency in a job that I felt running a company of my own.
That line of thinking shaped the team I'm building now, and I'm positive it's what every former technical founder is looking for. It has astonishing breadth (you sit across infrastructure, model research, agent development, security, etc.) and even more astonishing autonomy: you make decisions that affect millions of dollars of revenue every single day.
If you're a former technical founder, or any engineer looking for a high agency gig, take a look at the blog post below!
Decagon retweeted
We're hosting our annual conference, Decagon Dialogues, in SF on 10/1!
@ScottWu46 from Cognition is joining us for a keynote.
We'll have folks from American Airlines, Verizon, Paypal, Chipotle, etc discussing how they're deploying agents at scale.
Come by: decagon.ai/decagon-dialogues…
Decagon retweeted
One cool technique to improve latency in LLMs is to have a small, fast model generate tokens and a larger, slower model review the output in one pass.
It'd be like having a junior employee write a report, and then having a senior employee review it in one go before it gets sent to a client.
This is called "speculative decoding" and lets you generate faster while maintaining the output quality.
Great write-up from the @DecagonAI team on how we've been doing this:
We were very excited by @inco_ai's Dflash2 release and started building with it immediately.
A few tweaks to the training recipe further improved acceptance length by another 33%. Full write-up below.
Decagon retweeted
Internal AI moved beyond eng, and changed how every function works.
We're hosting builders / former operators who've built internal AI system to talk learnings, and what's next.
- @DecagonAI
- Former builders at @tryramp @cursor_ai
- @cognition
Link below! @Accel @a16z
Decagon retweeted
at decagon, we’ve been building around a simple idea: the failures of your current model checkpoint should shape the training data for the next one.
Decagon retweeted
Decagon is powering customer support for some of the biggest brands in the industry: Delta Airlines, Ticketmaster, Deutsche Telekom, American Airlines. A lot of support related questions require accurate online information and we’re happy to power that with Perplexity’s search.
We’re partnering with @Perplexity_AI to bring live web search to Decagon agents.
Customer questions often depend on information that changes by the hour.
Agents can now search the live web mid conversation, pull in current information, and respond with cited sources.
