Qasar Younis, CEO of @AppliedInt, has been a mentor and friend to @AshwinSreenivas since the earliest days of Decagon. He came to our SF office to share what he’s learned across two startups, Google, YC, and 8+ years building Applied Intuition. Thanks for joining us, @qasar!

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:

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

Introducing Campaign Composer, a new way to build AI agents for proactive outreach. Agents can coordinate voice and SMS across days or weeks, carry context between interactions, and follow campaign rules while working toward a customer outcome.

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.

Conversational voice agents need to know when a different person is speaking, but not every background voice should count. We combined speaker embeddings with a post-trained audio-language model to determine when a speaker change matters.

multimodal inference feels pretty underexplored, especially when you move beyond standard transformer architectures. fun to share what we learned when scaling our text to speech model to 10x throughput

multimodal inference doesn't fit the mold of standard text autoregressive generation. text-to-speech models feature different architectures, different states, and different batch shapes. we rebuilt our tts serving around that and simultaneously reduced our time to first audio while improving throughput by several fold.

our cloud agent usage at decagon is up 6x over the past 3 months. cloud agents are quickly becoming the default way our team works because: - they scale beyond a single laptop: each agent gets its own isolated sandbox (no worktrees) so there’s no local resource contention - you can spawn swarms from anywhere (web, slack, linear): designers and PMs are shipping - they can run for hours, spin up all services, and put up PRs with screenshots/videos but if verification isn't trustworthy, you just multiply slop. the bottleneck isn’t model intelligence, it’s the environment setup and verification. to verify effectively, you need to: - give it the same tools an engineer has locally (same stack, MCPs, context) - create a framework (like @poteto's pstack) that encodes how the agent should create/test its changes and ship artifacts you can review without reconstructing the session the horizon of work you can give agents is much longer - cloud agents can pretty much ship while you sleep. probably the highest leverage thing an eng team can invest in right now.

We’re proud to welcome Snap Inc., the company behind @Snapchat, to the Decagon customer family. Snapchat has become one of the defining consumer products of its generation, giving hundreds of millions of people a more visual and personal way to communicate with friends and family every day. Decagon helped power millions of support interactions last quarter as part of an initiative that has reduced Snap's support ticket volume by 62% since the beginning of 2026. Excited to keep working with the Snap team.

Modern TTS models can sound great — and still fail badly on pacing, pauses, and prosody. We adapted DPO + GRPO to flow-matching models to tackle the tail end of TTS behavior:

We're proud to welcome American Airlines as a Decagon customer. @AmericanAir is the world's largest airline by fleet size and passengers carried, operating more than 6,000 flights every day to over 350 destinations across 60 countries, serving hundreds of millions of customers each year. We're excited to partner with the AA team as they explore new ways to enhance the travel experience and deliver more seamless support for customers when they need it most.