# AI Engineer World's Fair 2026 — the field guide > 37 talks from AIE 2026 (June 29 – July 2, Moscone West, San Francisco), decoded into the topics, the tools, and the momentum that actually moved. Recorded on the floor, distilled for the record by ANALOG. Interactive version with charts: https://analoghq.ai/aie-2026 ## At a glance - 37 talks captured on the floor - 35 speakers across 11 tracks - 29 distinct topics surfaced — most-discussed: Agents - 148 products & frameworks named — most-named: Claude Code ## Top topics by talks - Agents — 33 talks - Developer experience — 16 talks - Context engineering — 12 talks - Enterprise adoption — 12 talks - Agent harnesses — 10 talks - Observability — 10 talks - Code generation — 9 talks - Evals — 9 talks - Leadership — 9 talks - Infra & Inference — 8 talks ## Momentum & tools (members' cut) The year-over-year trend read and the tool landscape (every product named on stage, plotted by utility × momentum with mention counts) unlock with one email on the page: https://analoghq.ai/aie-2026#unlock ## Speakers on the record - Elizabeth Fuentes Leone — Developer Advocate — Amazon Web Services - Sandhya Subramani — Senior Developer Advocate, GenAI — Amazon Web Services - Ahmad Osman — Founder & CEO — Osmantic - Alex Cheema — CEO — EXO Labs - Annabell Schäfer — Growth Engineer — Clickhouse - Arjun Singh — Co-founder and CEO — Superconductor - Barr Yaron — Partner — Amplify Partners - Christopher Manning — Distinguished Member of Technical Staff — Moonlake AI - Clare Liguori — Senior Principal Engineer — Amazon Web Services - Dex Horthy — Co-Founder — HumanLayer - Gagan Bhat — Member of Technical Staff — Anthropic - Hassan El Mghari — Director of Developer Experience — Together AI - Hursh Agrawal — Co-Founder & CTO — The Browser Company - Ignacio Martinez — AI Developer Advocate — Oracle - Isabella Kai He — Member of Technical Staff — Anthropic - Jetashree Ravi — Tech Lead Manager — Fireworks AI - John Ousterhout — Professor Emeritus — Stanford University - Joseph Nelson — Cofounder, CEO — Roboflow - Kent C. Dodds — Software Engineer and Educator — EpicProduct.engineer - Kenton Varda — Principal Engineer — Cloudflare - Laurie Voss — Head of Developer Relations — Arize AI - Leo Mehr — Director of Engineering — Ramp - Lotte Verheyden — AI engineer and developer educator, Langfuse — Clickhouse - Louis-François Bouchard — CTO & Co-Founder — Towards AI - Matthew Berman — Founder — Forward Future - Nader Khalil — Director of Developer Technology — NVIDIA - Natalie Meurer — Head of Agent Engineering — Sierra - Omar Solano — AI Engineer — Towards AI - Pauline Brunet — VP, Forward Deployed Engineering — Cursor - Samridhi Vaid — Senior Machine Learning Engineer — Towards AI - Sarah Sanders — Context Engineer — PostHog - Sunny Rekhi — FDE CTO — Decagon - Thariq Shihipar — Claude Code — Anthropic - Theo Browne — Founder/YouTuber — T3 Tools & YouTuber - Vinoo Ganesh — CEO & Co-Founder — Kepler ## The talks Titles, speakers, and one-line teasers are open. The full per-talk breakdowns, key takeaways, and quotes unlock with an email on the page: https://analoghq.ai/aie-2026#the-talks ### Total Recall: Agent Memory and Harness Engineering - Speakers: Ignacio Martinez, AI Developer Advocate, Oracle - Track: Harness & Context · Day 1 (2026-06-29) An agent is a rented, frozen model plus the harness you actually control. Oracle's Ignacio Martinez maps the harness layers and makes memory engineering a first-class discipline. [Recording](https://web.plaud.ai/nshare/pub_2c13ef4d-a8de-474f-8280-8f42bc97b85c::ylE7Q8bVWTJmop3mp8yqKfoy0o9jv7V-kWryjuzSaTbWoNDsV6lEZMZYjhReZ_P4TCmX1TFZBxRXPLkC) ### Agent Speedrun: Idea → Code → Deploy → Observe, Fix → Ship - Speakers: Elizabeth Fuentes Leone, Developer Advocate, Amazon Web Services · Sandhya Subramani, Senior Developer Advocate, GenAI, Amazon Web Services - Track: Agents · Day 1 (2026-06-29) AWS's hands-on speedrun: build a customer-service agent on the open-source Strands harness, bolt on hooks, skills, and steering, then ship it to Bedrock AgentCore. [Recording](https://web.plaud.ai/nshare/pub_90bb57fe-7db3-410f-bdd2-fc94151148a1::53hAAL0ZK0P3DITobxQA052kDL64NgVm48axBSbAPxsQWHYVAlI-q1Q8iLZP8AbjYdi-GKQlaqOe1rwC) ### Continuously improving agents with Langfuse - Speakers: Lotte Verheyden, AI engineer and developer educator, Langfuse, Clickhouse · Annabell Schäfer, Growth Engineer, Clickhouse - Track: Evals · Day 1 (2026-06-29) Langfuse's workshop turns agent reliability into a loop: trace everything, monitor with targeted evaluators, then let a coding agent mine production traces for silent failures. [Recording](https://web.plaud.ai/nshare/pub_a90cf913-9245-49d6-89ca-3701afc09617::kqAjnLhHzHHRcPh27hMJoQltqEVvgolRk6oRruGmcMK9Fg2-Lhhu-7PxuW0QrFCUw9JZtUTpONBNY0AC) ### Evolution of agentic surfaces - Speakers: Gagan Bhat, Member of Technical Staff, Anthropic · Isabella Kai He, Member of Technical Staff, Anthropic - Track: Harness & Context · Day 1 (2026-06-29) Anthropic traces the road from the Messages API to Claude Managed Agents, a cloud harness that splits an agent's brain from its hands, and demos a production SRE investigator. [Recording](https://web.plaud.ai/nshare/pub_23184e8c-8599-40f0-91b4-4d83a275edb2::aabx-kTCqDymGUBS-nO1OjGoZ6xmQEvTC50aks9PszdXCxPn3Dzs0myA5q2AK6l_kZHlEcJUqwh7Ro8C) ### From Vibes to Production: Evaluating and Shipping AI Agents That Work 101 - Speakers: Laurie Voss, Head of Developer Relations, Arize AI - Track: Evals · Day 1 (2026-06-29) Arize's Laurie Voss replaces 'ship it if it looks good' with a full eval loop: trace every agent step, read the failures, stack evals, let Claude Code fix what judges flag. [Recording](https://web.plaud.ai/nshare/pub_8406e308-baf4-42a7-b378-b87e9a883e89::44COTWPRpgzeXHnI8DDGfzcgmEbI_o9KPJWVlYRX_3GSHWTsnVzsQ-t1oBTsGuNcrpQR8O748kbcBm0C) ### Agent Speedrun: Idea → Code → Deploy → Observe, Fix → Ship - Speakers: Elizabeth Fuentes Leone, Developer Advocate, Amazon Web Services · Sandhya Subramani, Senior Developer Advocate, GenAI, Amazon Web Services - Track: Agents · Day 1 (2026-06-29) Two AWS advocates speedrun an agent from idea to production: Strands Agents as the harness, hooks and steering for determinism, Bedrock AgentCore for one-command deploys. [Recording](https://web.plaud.ai/nshare/pub_8d8b7d58-4c16-444b-b563-ff2d9e86ee11::Vo3xep7HjMVgA5YAcKw6ErYEtW6R-TGIrjJ6jeA48asBO99A2nyJIIxSxr6sRvxmoEECICn5osoNpz0C) ### Build the Right Thing: Product Engineering for Software Developers (Part 1) - Speakers: Kent C. Dodds, Software Engineer and Educator, EpicProduct.engineer - Track: Product & Design · Day 1 (2026-06-29) Kent C. Dodds argues AI agents have commoditized implementation, so the engineer's one durable skill is judgment: knowing what's worth building at all. [Recording](https://web.plaud.ai/nshare/pub_be6c0d8f-7106-42b4-ad1d-88a3b067da35::TgJmfRdrViQduMSLyZjyaQcrsgEiMnGGXmPc6VjuApM6rojyXyoxPaR3TpmXtyfJRv85NIH_B1kpewkC) ### Context Engineering in 2026: Compaction, Memory & Cost - Speakers: Louis-François Bouchard, CTO & Co-Founder, Towards AI · Samridhi Vaid, Senior Machine Learning Engineer, Towards AI · Omar Solano, AI Engineer, Towards AI - Track: Harness & Context · Day 1 (2026-06-29) Towards AI ran 11 context strategies through a live AI tutor and found prompt caching flips the playbook: full history beat summarization on recall, cost, and speed. [Recording](https://web.plaud.ai/nshare/pub_448033f0-5e64-49cc-8851-fcfdb1614bdc::B8w8Wof4_MNWvCAzQ9nElsFFtBCHkwMSUc9GqNIO1Sn4XXNh558fynPGATLtIELuzpYGpj9ZcHtx-KcC) ### Gadgets: Personal app vibe coding that is actually safe - Speakers: Kenton Varda, Principal Engineer, Cloudflare - Track: Code & SWE · Day 2 (2026-06-30) Cloudflare Workers creator Kenton Varda argues personal AI code gen breaks cloud infrastructure, and demos Fungy, a platform where sandboxing makes vibe-coded apps actually safe. [Recording](https://web.plaud.ai/nshare/pub_fce65f5c-2294-492e-abf5-3d706dcabf8e::1PSC6kCswPFYsYQrQiIdWKHF_x6l2FbFOtHbb3ENWOJeRpOFY12kGkEaS0RHruLo0JVXGClNBa4UPe0C) ### Harness Engineering is not Enough: Why Software Factories Fail - Speakers: Dex Horthy, Co-Founder, HumanLayer - Track: Code & SWE · Day 2 (2026-06-30) Dex Horthy's case against lights-off software factories: RL-trained coding agents optimize for passing tests, not maintainability, so humans must keep reading the code. [Recording](https://web.plaud.ai/nshare/pub_bae3164a-234d-403f-9b6b-32448985b03e::cXNYg_Qd-kY2lAYIM1a6acx7b05RjZb0-kuBWSi0KjLNcqDiRcq0qzJvSx0IJtVM0XqXThsGZLbZ9tgC) ### Enterprise AI Agent Adoption and Infrastructure Challenges - Speakers: — - Track: Agents · Day 2 (2026-06-30) A blunt panel on why agent traffic melts human-built infrastructure, and why trust, taste, and an agent-native, text-first internet decide who wins the agent era. [Recording](https://web.plaud.ai/nshare/pub_67259a68-c59b-440a-989a-1bd3bc7a3828::zHovO1RpKNLKBcCTYyUy8yNdYm6dSm8dOYk5cdbhhF7He_KfiYXZR2TcDndXItUcQurXqjl4jF8XVWYC) ### Future of Software Development with AI - Speakers: — - Track: Code & SWE · Day 2 (2026-06-30) GitHub, LaunchDarkly, Mintlify, VS Code, Greptile, and GitHub Next leaders on what's left when code costs nothing: alignment, validation, and taste. [Recording](https://web.plaud.ai/nshare/pub_fdfbf645-a643-4cdd-9b18-6c66f8a52ffd::svCNwTPLdjtdnPafqRVf0vXwxnhMljD5dOSGWtXqbqcZL0vtdS0vgks-kFyNrCZJ3lNKxB4i592kqDgC) ### How Forward Deployed Engineering is done at Cursor - Speakers: Pauline Brunet, VP, Forward Deployed Engineering, Cursor - Track: Leadership · Day 2 (2026-06-30) Cursor's VP of forward-deployed engineering shares a decade-honed playbook for building an embedded FDE team that drives enterprise AI transformation, not staff augmentation. [Recording](https://web.plaud.ai/nshare/pub_272ce3cb-2b62-4bd8-a39d-6eade946fb7a::ZwXwmnpwg0v0hYnuPr2M5y-QreWPVSKvjD-NzOigtEMXk_jia6nzdgbbPX6FelyTLLNuRssrn14U_zoC) ### How Forward Deployed Engineering is done at Kepler - Speakers: Vinoo Ganesh, CEO & Co-Founder, Kepler - Track: Agents · Day 2 (2026-06-30) Kepler's CEO says AI execution is solved; the real bottleneck is going deep into each customer. His fix: forward-deployed engineers augmented by an internal FDE agent. [Recording](https://web.plaud.ai/nshare/pub_656d53a5-ae97-407e-8804-e80c51e40901::rgx4QqBPgVxy_iMNfEzTSdPaK4Naz63WdvjQA7bCSJgNEgn1zOozOmNoNI_nwFyaml-1QQC9TEIBA1gC) ### Prototyping as Leadership: How a CTO Ships with AI Agents - Speakers: Hursh Agrawal, Co-Founder & CTO, The Browser Company - Track: Leadership · Day 2 (2026-06-30) A CTO with 15+ weekly meetings and seven direct reports still ships 2-10 PRs a week - by turning the fractured manager calendar into overnight agent runs. [Recording](https://web.plaud.ai/nshare/pub_66c14034-0077-4511-9c22-a796d44933ce::yI0SCFgYQieecNrCcLiCgSfEwy_xov8ubsoElogWqUnNF1gAJCC8OVzJ0wbmlbZ6zjzuzIzwFMjFlWsC) ### The Dirty Secret of Forward Deployed Engineering - Speakers: Natalie Meurer, Head of Agent Engineering, Sierra - Track: Leadership · Day 2 (2026-06-30) The unicorn FDE job posting doesn't exist, and doesn't need to. Sierra's Natalie Meurer argues cheap code and outcome pricing are making every engineer forward deployed. [Recording](https://web.plaud.ai/nshare/pub_cd21ef6f-4572-4303-bb5c-99a54a6561ff::4IcXTitCfCSSqUOghicZ3XbxIEFajljtb0LfXm-QCT85GCqMgnIJqKlxXQKKdIB0joYslGZw87n7ezEC) ### How Forward Deployed Engineering is done at Decagon - Speakers: Sunny Rekhi, FDE CTO, Decagon - Track: Other · Day 2 (2026-06-30) Decagon's Sunny Rekhi on why forward-deployed engineering is product engineering: solve one customer's ask so the next five never have to make it. [Recording](https://web.plaud.ai/nshare/pub_c98817e3-83dd-463b-8926-3b54a18c86e0::B4DweykdKpJ-tPWJG9jlaDixZYeoO-dAFB6ZlmFcWw_lUN5HiMRLgY2j0vYYFofGwNImtiF2zrz8VrkC) ### From AI-Assisted to AI-Native: Building a Frontier Development Team - Speakers: Clare Liguori, Senior Principal Engineer, Amazon Web Services - Track: Leadership · Day 2 (2026-06-30) Amazon watched 50 teams use Kiro; the 4.5x-plus teams didn't have better tools: they changed how they worked. Clare Liguori distills the five habits of frontier engineering. [Recording](https://web.plaud.ai/nshare/pub_2cf156e6-17b1-4a06-86f5-42499ab01113::crbWo1gY2Clqq_zlH-aLqa1zeYM-LIGLTodvs1z_BrP4kr2iaUvimTG9m7LccQoCib4yE42CBy0mVOMC) ### How Forward Deployed Engineering is done at Ramp - Speakers: Leo Mehr, Director of Engineering, Ramp - Track: Other · Day 2 (2026-06-30) Ramp's FDE playbook in two rules: interrogate every 'urgent' enterprise request before building, then hand the scoping-to-shipping pipeline to agents, starting with intake. [Recording](https://web.plaud.ai/nshare/pub_b83f565a-277c-417e-b82a-e98e55f0bd3b::pdGEgFkyQByq7Fkgt4-7qbU0_6Ff3QZTxwixjrha6DKIe1eM58P9yQn2sjpQYISU_UxHPWtTPJ2ZXnkC) ### Field Guide to Fable - Speakers: Thariq Shihipar, Claude Code, Anthropic - Track: Harness & Context · Day 3 (2026-07-01) Anthropic's Thariq Shihipar launches Fable with a field guide: unhobble the model with tools, hunt your unknowns, grieve the old craft, and get unreasonable about ambition. [Recording](https://web.plaud.ai/nshare/pub_b49c9d28-4918-4004-aa85-bf9cc4cecfec::vniUQqouAgtDPfULMowfGzGaST9j2m3SU2bTAAzQFA-dGWYT2Gbf9BB7ligGTLhj_S8FcuYov27fuqcC) ### Building the simulation infrastructure for practical world model use - Speakers: Christopher Manning, Distinguished Member of Technical Staff, Moonlake AI - Track: Robotics · Day 3 (2026-07-01) Christopher Manning walks 70 years of AI history to one conclusion: embodied AGI needs verifiable, code-powered world models, not pixel-deep generative video. [Recording](https://web.plaud.ai/nshare/pub_184c6e24-13b9-45c1-bbff-d30057223b42::Zvrub7LOrWuegbctxmW1I-sTqDr1pu5odA9wRTrlqmYnZkZfNVW0vOYNSe5-Y4U2YS-nqm7qGmLRhnMC) ### The Missing Layer: Design Taste in AI Agents // Stop Letting Your Agents Ship Ugly UIs - Speakers: Hassan El Mghari, Director of Developer Experience, Together AI - Track: Product & Design · Day 3 (2026-07-01) Together AI's Hassan El Mghari on purging AI slop from generated UIs: codify the tells, feed agents screenshots, and iterate with fast open models like GLM 5.2. [Recording](https://web.plaud.ai/nshare/pub_4e0b881e-98dc-4e36-a937-51b5c3fcc98c::YjryYVaTyX1qyKQ7NcT653d8JRRqg4mw5QFgxrrKNe4cc0vk3kQlBM7ZGFcEMAuHrZapDVk7j3L35HEC) ### AI Agents in Software Engineering — Measurement, Quality, and Agent Readiness - Speakers: — - Track: Code & SWE · Day 3 (2026-07-01) A Vercel-heavy panel on why validation loops (not lines of code) predict AI delivery gains, how to fight slop, and what actually makes an agent feel tasteful. [Recording](https://web.plaud.ai/nshare/pub_5b23fbdf-e87c-4869-b81e-f1059903e590::ybdPhZkYU_aydY3zUrvUeZg8ts_v79dCKnldE8Ff6_m1vhYpCF00Sig8_N4t9DSSOZLbJ4fFRkX7y3kC) ### Browser-Based Agents for Knowledge Work - Speakers: — - Track: Agents · Day 3 (2026-07-01) Browserbase's live demo argues the browser agent is a coding agent in disguise, ordering dinner and solving LeetCode to show agents doing real knowledge work on the web. [Recording](https://web.plaud.ai/nshare/pub_18511631-2c1f-4409-b79c-7db31b172fd0::VrWgOCDn7gqOm2jOATn57R5x8W-KmOdW3HkP2u3NOggtOT6HYxm-BKuctJLQnXGCGVNQLGrxqQddLIoC) ### Infra as Code for Agent-Driven TypeScript SaaS - Speakers: — - Track: Infra & Inference · Day 3 (2026-07-01) Agents will provision your infra, not just your code, so hand them a typed TypeScript config. Neon's neon.ts plus a two-command CLI stands up Postgres, auth, and storage. [Recording](https://web.plaud.ai/nshare/pub_711c5e79-71ed-4702-93ee-54872c2570b0::CU2zbSzylaYK60h8nKHheiBRUG9l40ENdpW55o4mcmyBv5CAKK6KRO-1Q_k3XQ93pyO2ZGNFSUfbjqIC) ### Stop Renting Intelligence: The Train-to-Deploy Loop for Specialized AI - Speakers: Jetashree Ravi, Tech Lead Manager, Fireworks AI - Track: Infra & Inference · Day 3 (2026-07-01) Fireworks AI's pitch for 'owned intelligence': ditch pricey closed models, fine-tune open ones like GLM 5.2, and run a continuous train-to-deploy loop on custom CUDA kernels. [Recording](https://web.plaud.ai/nshare/pub_d6231581-1397-4f85-abda-451913dab805::KswnDFQTiYuohsWoPGsTZYSb82WczJGUGcnovtjXVg7m5rLPWVbJSGSJ7HajhFL_FCLD94n_AyUdJRsC) ### Mapping Human Memory into Agent Systems with Spectron on SurrealDB - Speakers: — - Track: Memory · Day 3 (2026-07-01) SurrealDB's founder pitches Spectron, a memory layer that mimics human cognition (beliefs, dream-like consolidation, temporal facts) so agents stop forgetting. [Recording](https://web.plaud.ai/nshare/pub_f50bf1a2-50b0-4983-a7e8-5798a1efeb50::vqjSTzF87ig-8Yb99NYgPqeoqIZDxw2-4Wii7xTfDuzkeFSN-_lWRLd-fmuKsF7U23HUvuNXnhfAeNIC) ### Agent Optimizer — Autonomous AI Agent Cost and Performance Governance - Speakers: — - Track: Agents · Day 3 (2026-07-01) RunLayer runs 482 agents with just 40 humans, so they built another agent to audit the fleet, right-sizing models and pruning tools for roughly $65K a year in savings. [Recording](https://web.plaud.ai/nshare/pub_5812440c-0a58-4c46-a560-1792249de777::KM-L3TJ2elAPRtbI-IPtnu_n3M6gON4h48trem69vhhfNSrTUrXbVGY6lseX82l4-87w14GraUsd0EUC) ### Exa AI Search Paradigm, Token-Efficient Context, and Agentic Workflows - Speakers: — - Track: RAG & Search · Day 3 (2026-07-01) Exa's lightning-talk pitch for search built for AI: token-efficient Highlights instead of whole pages, and an agent that traced Marc Andreessen to Kevin Bacon in 42 seconds. [Recording](https://web.plaud.ai/nshare/pub_887cdd5a-54d8-43a2-a463-1a7308e16092::NBy56Lg3FmFiYK03yTfBNruKbKgla2lnMuHliPwYMmEgY6nhyXq8qRY7vQrEsHlVFBhjyXXGDx4wUFYC) ### Natively Multimodal from Step Zero - Speakers: — - Track: Infra & Inference · Day 3 (2026-07-01) MiniMax explains how M3 got a 1M-token context and native vision: a two-branch sparse attention design plus multimodal training from the very first pretraining step. [Recording](https://web.plaud.ai/nshare/pub_a4dd0f85-21d0-41a7-832d-738d0309719b::wW5F_Z0UNcJWcbuFjAvd9rYEkez1j6Wa9WXULycwcQa7cMR5eJETKNA5VLUR2qa9yggsNlO5wEEim9AC) ### Sentry for Fixing Broken Code - Speakers: — - Track: Code & SWE · Day 3 (2026-07-01) Sentry's Dorian Crutcher compresses the debugging loop into five minutes: trace-connected issues, session replay, and an AI called SEER that root-causes the bug and drafts the PR. [Recording](https://web.plaud.ai/nshare/pub_d7d2ee46-b658-4867-a6b5-c3323d2fd24a::aVIOg919czX70oz5KMMewScW8FOuvRpVQISlO-WtxFRMxcZNnpXAMD2f6QXVfzw8vhiQ_huPiW0VIdgC) ### We let an AI agent execute Bash and lived to talk about it - Speakers: Sarah Sanders, Context Engineer, PostHog - Track: Harness & Context · Day 4 (2026-07-02) PostHog's setup agent hit 8,000 weekly runs. An agent that executes commands is a 'malware starter pack'. Here's the deterministic, layered defense that made it shippable. [Recording](https://web.plaud.ai/nshare/pub_0cfc1606-bf3a-4ee9-9a36-d92443856cea::WWzN_LfHwdtIzxKHqrh-97GmxWfcAJbOd1Cevl7IAd7XPvyFyo9e4clC213cnJA8Lpypkf9sdlIP9icC) ### Closing Keynote — Theo Browne - Speakers: Theo Browne, Founder/YouTuber, T3 Tools & YouTuber - Track: Product & Design · Day 4 (2026-07-02) Models are now improving faster than the people using them. Theo Browne's closing keynote: drop your skeuomorphic dev habits and chase ideas that feel stupidly big. [Recording](https://web.plaud.ai/nshare/pub_84b45a31-e424-4305-ada9-5276da9c2a42::cBtEOdoeq66NQnXdB0UD8yGPBMkVWpTcYnfvJr2b08_qljTfPbiTFJ6_nUq6v15B7CYsfR_QftigbioC) ### The 2026 State of AI Engineering - Speakers: Barr Yaron, Partner, Amplify Partners - Track: Harness & Context · Day 4 (2026-07-02) Amplify's AI Engineer survey: 97% call AI a net positive, 90%+ feel the downside. Cheap failure means more experiments, more review burden, and a looming maintenance bill. [Recording](https://web.plaud.ai/nshare/pub_7cb7b745-f1f8-4639-b19c-249fa3c6f7bc::_xKqHZ37jqMooSRrd6Aw8piLoQam_IfJtg-ZzjvAO3HO0jIy808lokeroCfqyrzcKiSRIdFQzyr5SJMC) ### TCP and RDMA are Killing Inference Throughput; Homa can Fix It - Speakers: John Ousterhout, Professor Emeritus, Stanford University - Track: Infra & Inference · Day 4 (2026-07-02) Stanford's John Ousterhout argues AI's shift to small, latency-critical messages breaks TCP and RDMA, and pitches Homa, a receiver-driven transport that cuts tail latency 13x. [Recording](https://web.plaud.ai/nshare/pub_f730dbd9-3287-42ec-8e5b-fa1ffa0a9410::jeMxFJKtp3ezHfi03IEkv99BjgMUoXX37xBQIeJhOWHgVLCGUe0Y807pJzG_r9V418cF6gTGV6MuTd4C) ### State of the Union: Why Local, Why Now - Speakers: Nader Khalil, Director of Developer Technology, NVIDIA · Joseph Nelson, Cofounder, CEO, Roboflow · Alex Cheema, CEO, EXO Labs · Ahmad Osman, Founder & CEO, Osmantic · Matthew Berman, Founder, Forward Future - Track: Infra & Inference · Day 4 (2026-07-02) Five local-AI insiders from NVIDIA, Roboflow, EXO Labs, Osmantic, and Forward Future argue on-device models just crossed from toy to default for privacy, cost, and control. [Recording](https://web.plaud.ai/nshare/pub_54f2f6d7-e1b4-4d8e-ba34-52277c8b717c::2CWvb3y4-M985YKfLf9zwoFk_3boR0zVK9kOJfQKmwz9y6kQ00MvuKNDEvDG2bnvzKr2perB5Y0Xa8IC) ### Multiplayer agentic engineering: enabling your whole team and your best agents to work together - Speakers: Arjun Singh, Co-founder and CEO, Superconductor - Track: Code & SWE · Day 4 (2026-07-02) Gradescope co-founder Arjun Singh's five rules for agentic teams: stay model-agnostic, make agent sessions multiplayer, and sandbox all the work in the cloud. [Recording](https://web.plaud.ai/nshare/pub_bddbd076-061d-4a1b-b6bb-c81385ee47fb::drGnt5Qj2pc-LJZPbyAuTPATOxSSu-2rxRztWSytT7oKo9X1JgvBHlaune5PEDP_xqpw_1GM7pzSkrkC) ## The week's grid Recorded talks per track and day. Day 1 = Jun 29, Day 2 = Jun 30, Day 3 = Jul 1, Day 4 = Jul 2, 2026. | Track | Day 1 | Day 2 | Day 3 | Day 4 | | --- | --- | --- | --- | --- | | Agents | 2 | 2 | 2 | 0 | | Code & SWE | 0 | 3 | 2 | 1 | | Harness & Context | 3 | 0 | 1 | 2 | | Infra & Inference | 0 | 0 | 3 | 2 | | Leadership | 0 | 4 | 0 | 0 | | Product & Design | 1 | 0 | 1 | 1 | | Evals | 2 | 0 | 0 | 0 | | Other | 0 | 2 | 3 | 0 | ## Who sent people Companies with two or more speakers in the recorded set: | Company | Speakers | Talks | | --- | --- | --- | | Amazon Web Services | Elizabeth Fuentes Leone, Sandhya Subramani, Clare Liguori | 3 | | Anthropic | Gagan Bhat, Isabella Kai He, Thariq Shihipar | 2 | | Towards AI | Louis-François Bouchard, Samridhi Vaid, Omar Solano | 1 | | Clickhouse | Lotte Verheyden, Annabell Schäfer | 1 | One voice each: Amplify Partners, Arize AI, Cloudflare, Cursor, Decagon, EpicProduct.engineer, EXO Labs, Fireworks AI, Forward Future, HumanLayer, Kepler, Moonlake AI, NVIDIA, Oracle, Osmantic, PostHog, Ramp, Roboflow, Sierra, Stanford University, Superconductor, T3 Tools & YouTuber, The Browser Company, Together AI. ## Time on tape - Total recorded: 14h 40m across 37 talks - Typical talk: 24m mean, 18m median - Longest: "From Vibes to Production: Evaluating and Shipping AI Agents That Work 101" at 1h 49m 24s - Shortest: "Sentry for Fixing Broken Code" at 4m 40s ## Method This guide is a transformative recap of the AI Engineer World's Fair 2026, built from talks recorded on the floor. Summaries, takeaways, and topic/tool analysis are original synthesis — not transcripts. Full recordings and slides belong to the speakers and organizers; each talk links back to its official session where available. [View the interactive field guide on ANALOG](https://analoghq.ai/aie-2026)