Speaker
Abstract
Most agent architectures stop at the model call. Getting a response to a person on iMessage, WhatsApp, SMS/RCS, Telegram or a phone call is its own distributed systems problem, with its own latency budget, failure modes and state. This talk walks through how Photon built the distribution layer that runs agents natively in those channels for tens of thousands of developers: one conversation model across channels with very different delivery guarantees, cross-channel fallback when a channel degrades, where end-to-end latency actually goes between model serving and delivery, and the production lessons that changed the design. Attendees leave with patterns for shipping agents to the places people already talk.
Sponsored session
Session sponsored by
The distribution layer for AI agents
$3,250, Conference (3 days). Current pricing ends November 20th. All pass options.
QCon San Francisco 2026 is a three day conference for senior software engineers, architects and team leads. An international program committee of working engineers selects every session. Patterns and practices, not products and pitches.
Slides & video
Video Not available
Part of the track
Sponsored Solution Track VFrom the same track
Wednesday 18 November
10:35 Pacific LM Sponsored Simulating the Universe, 100 Times Over: Engineering Highly- Concurrent Agent Testing Andy Liang, Anvit Sinha Testing AI agents properly is surprisingly hard. Production data sources are moving targets, static datasets go stale, and pointing agents at real systems is risky and impossible to reproduce. 11:45 Pacific LM Sponsored The Men in Black Know What's in Your JavaScript and How to Fix it! Chris DeMars Senior Developer Advocate @TuxCare Ever wonder what’s lurking in the dark corners of your node_modules, like unidentified flying dependencies sneaking aboard your project? In an age of escalating software supply chain attacks, knowing exactly what you're shipping is as vital as tracking strange lights in the sky. 13:35 Pacific LM Sponsored The Fast, The Furious and The Fatigued Christian Weber For years, achieving daily releases has been the gold standard of DevOps maturity. But what happens when you get there? According to new data from over 700 engineering leaders, high deployment frequency is increasingly masking a crisis of off-hours rework, instability, and engineer burnout. 14:45 Pacific LM Sponsored Bring AI Agents to Millions of People Where They Already Text Daniel Tian CEO & Co-founder @Photon Most agent architectures stop at the model call. Getting a response to a person on iMessage, WhatsApp, SMS/RCS, Telegram or a phone call is its own distributed systems problem, with its own latency budget, failure modes and state. 15:55 Pacific LM Session Sponsored session Details coming soon!