Conference Guide
AI Engineer World's Fair
模型不再是变量，挽具、验证与注意力才是
97talks17worth watching32.9hours total
What this conference was about
01挽具压过模型
02审查债正在结算
03注意力才是瓶颈
04确定性代码回收职责
05小模型经济学反转
06Agent 基建即分布式系统
07Web 长出 agent 接口
Conference
Scenario
Who it’s for
Topic
Showing 97 of 97 · orange dot = unread
- In the AI era, building is no longer the bottleneck — distribution is. A practical session on turning yourself, the founder, into your strongest growth channel.Victoria Melnikova · AI Engineer World's FairFor · Technical founders / Early-stage dev tool teamsAI Product1 min on screen13 min totalWorks as a podcast
- A ready-to-use four-part checklist for reviewing skills—trigger, structure, steering, pruning—to help you get out of Skill Hell.AI Engineer World's FairFor · Skill authors and reviewers / Agent tuning engineersAgentContext1 min on screen21 min totalWorks as a podcast
- Five minutes on an edge case no RAG tutorial covers: what to do when all of your documents are relevant and the whole dataset gets replaced constantly.Luis Romero-Sevilla · AI Engineer World's FairFor · Engineers choosing a RAG architecture / Teams shipping GraphRAGContext1 min on screen6 min totalPart listen, part watch
- It's not that the model is bad — the agent has no eyes: a practical checklist for building it verification toolsJohan Lajili · AI Engineer World's FairFor · AI coders working on a brownfield codebase / Engineering leadsAgentAI CodingEvals2 min on screen10 min totalPart listen, part watch
- Turn production logs into replayable test beds, so every fix is proven to help and proven to break nothing that already workedSoheil Feizi · AI Engineer World's FairFor · Agent engineers firefighting after launch / Algorithm and platform leads working on agent optimizationAgentEvalsContext2 min on screen23 min totalWorks as a podcast
- Watch the "one all-purpose agent plus MCP" path get taken to its breaking point, then get a compositional alternativeJustin Schroeder · AI Engineer World's FairFor · Enterprise agent architects / AI cost decision-makersAgentContextAI Product2 min on screen31 min totalWorks as a podcast
- Frontline view from Meta: agents don't fail on intelligence, they fail on infrastructureNishant Gupta · AI Engineer World's FairFor · Agent platform architects / Infrastructure engineersAgentContextAI Product2 min on screen7 min totalPart listen, part watch
- Measured, not guessed: 90% of your AI coding bill is context that was not relevant. One local search layer cuts 94% of it.Rajkumar Sakthivel · AI Engineer World's FairFor · Engineers paying their own bill / Team tech leadsAI CodingContext2 min on screen11 min totalPart listen, part watch
- A view from inside Meta: agent evaluation isn't benchmark chasing, it's production engineeringNishant Gupta · AI Engineer World's FairFor · Agent platform lead / Engineering manager shipping agentsAgentEvals2 min on screen8 min totalPart listen, part watch
- Co-founder swyx previews AIE 2026's five vertical bets and how to get your ticket money backswyx · AI Engineer World's FairFor · Engineers planning to attend / CTOs / VPs of AIAgentAI Product2 min on screen18 min totalWorks as a podcast
- Uses Cursor's measured numbers to puncture "RAG is dead" and spell out what retrieval has actually becomeKuba Rogut · AI Engineer World's FairFor · Engineers building code base retrieval / Technical decision-makers weighing whether to indexAgentContextAI Coding2 min on screen11 min totalPart listen, part watch
- AI speeds up finding bugs and writing them at once; his answer isn't patching, it's replacing the foundation.Jack Cable · AI Engineer World's FairFor · Security lead / Infrastructure leadAI CodingAgent3 min on screen20 min totalPart listen, part watch
- After 500,000 sensor names confused the LLM, a tree-structured rebuild won back 100% accuracy and a 300x cost cutAI Engineer World's FairFor · LLM application architect / Agent cost ownerAgentContextEvalsAI Product3 min on screen16 min totalPart listen, part watch
- 12 minutes on the three layers of infrastructure the internet of agents is missing, and how MIT fills them with an open index and a sandbox you can actually runRamesh Raskar · AI Engineer World's FairFor · Agent platform and protocol developers / Multi-agent systems researchersAgent3 min on screen12 min totalPart listen, part watch
- Break "done" out of a single green checkmark into a verifiable object, or your agents will produce slop at scaleDotta · AI Engineer World's FairFor · AI engineering lead / Platform architectAgentEvals3 min on screen7 min totalPart listen, part watch
- A 100-person factory grew a 36-agent brain with zero GPUs, for a fraction of the agency's quoteRushabh Doshi · AI Engineer World's FairFor · Owner of a small or mid-sized manufacturer / Successor in a family businessAgentContextAI Product3 min on screen10 min totalPart listen, part watch
- Manage agents like new hires: three copyable trust blueprints -- commander's intent, the librarian, the jury.Alex Bauer · AI Engineer World's FairFor · GTM / growth leads / Enterprise agent buildersAgentContextAI Product3 min on screen17 min totalPart listen, part watch
- Notion's Geoffrey Litt reframes "do we still need to read code in the AI era" as a question of participation, and gives you fixes you can use today.Geoffrey Litt · AI Engineer World's FairFor · Engineers who lean heavily on agents / Tech leadsAgentAI Coding3 min on screen20 min totalPart listen, part watch
- He publicly retracts his own "the models have hit a wall" call, and redraws the line for what's worth building.Theo Browne · AI Engineer World's FairFor · Indie developers and founders / Senior engineersAI CodingAgentAI Product3 min on screen16 min totalPart listen, part watch
- 9 minutes on why "can you verify" is replacing "can you code": a hands-on retrospective of building Vector Harness.Talha Sheikh · AI Engineer World's FairFor · AI coding engineers / AI tooling leadsAgentAI CodingEvals3 min on screen10 min totalPart listen, part watch
- An agent fleet across three machines: five things that broke, and an honest list of what's still unsolvedKyle Jaejun Lee · AI Engineer World's FairFor · Engineers running multiple agents / People building agent frameworksAgentAI CodingContext3 min on screen9 min totalPart listen, part watch
- Turns backdoor detection into a build-time unit test costing one forward pass, and admits recall is only a quarterSachin Kumar · AI Engineer World's FairFor · Model safety lead / ML platform engineerEvals3 min on screen14 min totalPart listen, part watch
- Six minutes to learn how to turn a jargon-heavy AI product into a story that lands in 20 secondsVeronica Hylak · AI Engineer World's FairFor · AI startup founders / B2B product managersAI ProductAgent3 min on screen6 min totalNeeds your eyes
- Once AI writes the code, the real moat is deciding what to build — and the toolkit is a set of old trades worth picking up on Monday.Balázs Horváth · AI Engineer World's FairFor · Engineering / development leads / Product managersAI ProductAI CodingContext3 min on screen16 min totalPart listen, part watch
- An ablation that undercuts its own pitch: self-healing ETL reliability comes from structure, not RLAnna Marie Benzon · AI Engineer World's FairFor · Data platform engineers / Tech leadsAgentEvals3 min on screen15 min totalPart listen, part watch
- Validated on 3M real records: compliance risk lives between the documentsVarsha Shah · AI Engineer World's FairFor · Risk and fraud-detection tech leads / Architects of cross-border financial complianceAI ProductEvals3 min on screen19 min totalPart listen, part watch
- In five minutes, see the five leaks that quietly burn money in production agents — and the code to stop each oneErik Hanchett · AI Engineer World's FairFor · Agent backend developers / Anyone bleeding on their token billAgentContext3 min on screen6 min totalNeeds your eyes
- How Block used 50 Champions to grow automated PRs 21x — the complete scaling roadmapAngie Jones · AI Engineer World's FairFor · Engineering effectiveness leads / Platform/DevEx engineersAgentAI CodingContext3 min on screen18 min totalPart listen, part watch
- The full spec-driven development flow, tool-agnostic: requirements, design and tasks documents before any code is written, to keep a wandering AI assistant on the railsErik Hanchett · AI Engineer World's FairFor · Engineers already using coding assistants / Tech leadsAI CodingContextEvals3 min on screen18 min totalPart listen, part watch
- Four minutes of live runs: giving the agent better "eyes" beats giving it a better modelKushan Raj · AI Engineer World's FairFor · Browser agent developers / Product leads for agent toolingAgentContext3 min on screen4 min totalNeeds your eyes
- Reclassify the log from system exhaust to the agent itself, and reliability, scale and lock-in all come loose at onceIshaan Sehgal · AI Engineer World's FairFor · Runtime developers / Platform decision-makersAgentContext3 min on screen15 min totalPart listen, part watch
- "I did a search" is often a lie — a live GPT-5 run fails all five sites, then lands all fiveRafael Levi · AI Engineer World's FairFor · Developers wiring web access into agents / Product leads dealing with dead citationsAgentContextAI Product3 min on screen16 min totalPart listen, part watch
- Don't put the repo into context — let the model write code to inspect it and recursively call sub-models; the open source RLM implementation, demoed live end to endShashi · AI Engineer World's FairFor · Coding agent engineers / Developers who want to reproduce RLMAgentAI CodingContext4 min on screen17 min totalPart listen, part watch
- Don't change your terminal workflow — put four parallel coding agents on your phone exactly as they areConnor Adams · AI Engineer World's FairFor · Terminal-based multi-agent developers / Anyone who wants to check on progress from their phoneAgentAI Coding4 min on screen9 min totalNeeds your eyes
- Why chat plus citations is a dead end for vertical AI, and how to shift toward delegation-style products that actually do the work for your users.Atul Ramachandran · AI Engineer World's FairFor · Vertical AI founders / AI product managersAgentAI Product4 min on screen15 min totalPart listen, part watch
- Recognize the worse version of CI/CD you're rebuilding by hand, and stop it with five gatesSumaiya Shrabony · AI Engineer World's FairFor · Solo agent builders / Content automation authorsAgentEvals4 min on screen11 min totalPart listen, part watch
- The talk is itself an AI game: runtime decision-making running live, and an NPC match that plays out differently every timeDavid Hoe · AI Engineer World's FairFor · Indie AI game developers / Content platform product leadsAI ProductAgentAI Coding4 min on screen18 min totalPart listen, part watch
- Replacing 15 tools with a single REPL: the full post-mortem of taking a spreadsheet agent from 50% to 92%Nuno Campos · AI Engineer World's FairFor · People designing agent tool interfaces / People who own an agent evaluation systemAgentAI CodingEvalsContext4 min on screen19 min totalPart listen, part watch
- Watch AI run a chess-commentary YouTube channel entirely on its own: the engine can play, the LLM can talk, and combining them is the holy grail.Stephan Steinfurt · AI Engineer World's FairFor · LLM agent engineers / AI content foundersAgentAI Product4 min on screen17 min totalPart listen, part watch
- Once your agent ships, how do you find problems, send PRs, and gate them? Wandero's production practice: a half-hour loop.Raphael Kalandadze · AI Engineer World's FairFor · Engineering leads on agent products / Developers with an agent already in productionAgentEvalsAI Product4 min on screen20 min totalPart listen, part watch
- Can't reproduce that production agent failure? Stop fighting for determinism — record and replay turns it into a free regression test.Tisha Chawla · AI Engineer World's FairFor · Production agent engineers / AI testing leadsAgentEvals4 min on screen14 min totalPart listen, part watch
- Around the 200ms tyranny of latency: three engineering rules for real-time voice in, visuals out agentsAllen Pike · AI Engineer World's FairFor · Voice agent builders / Real-time AI product leadsAI ProductAgentContext4 min on screen13 min totalPart listen, part watch
- Three levers for getting research into production: a handoff doc, a repo skeleton, and stacked-diff decompositionVaidas Razgaitis · AI Engineer World's FairFor · Tech leads of mixed teams / ML researchersAI ProductAI Coding4 min on screen15 min totalPart listen, part watch
- Let anyone ask business questions in natural language, then freeze the answer into a widget with zero ongoing LLM costGarrett Galow · AI Engineer World's FairFor · Data and analytics teams / AI agent developersAgentContextEvalsAI Product4 min on screen19 min totalPart listen, part watch
- How PostHog refines trillions of noisy signals into green PRs an engineer can merge on waking upJoshua Snyder · AI Engineer World's FairFor · Platform and infrastructure engineers / AI pipeline leadsAgentAI CodingEvalsAI Product4 min on screen16 min totalPart listen, part watch
- Two years of GitHub data, one claim: the AI application layer is changing its name to TypeScript.Roberto Stagi · AI Engineer World's FairFor · Whoever owns the agent stack decision / Full-stack application engineersAI CodingAgent5 min on screen14 min totalPart listen, part watch
- Data from 22,000 engineers ends the "should we still read AI code" fight: whether you read is routed by the risk of the change.Alex Volkov · AI Engineer World's FairFor · Engineers leaning hard on agents / Tech leads setting review standardsAgentAI CodingEvals5 min on screen22 min totalPart listen, part watch
- The OpenAI Codex team breaks it down: from watching ten terminal windows to talking to one long-running manager agent.Romain Huet · AI Engineer World's FairFor · Engineers working many things in parallel / Engineering team leadsAgentAI CodingAI ProductContext5 min on screen25 min totalPart listen, part watch
- Takes "one version for everyone" apart as a cost calculation, then hands you the architecture and verification problems of a version per user.Iris ten Teije · AI Engineer World's FairFor · Horizontal SaaS product architects / CI and release system ownersAgentAI CodingAI ProductEvals5 min on screen20 min totalPart listen, part watch
- A designer who doesn't ship code, on how a 30-day experiment with 500 people turned her into someone who ships productsSanja Grbic · AI Engineer World's FairFor · Designers who don't write code / Team leadsAI CodingDesign-to-CodeContextAI Product5 min on screen18 min totalPart listen, part watch
- A billion tokens in: even the strongest agent finishes only one project-scale task in fourRishi Desai · AI Engineer World's FairFor · Agent eval designers / Engineers picking a coding agentAgentAI CodingEvals5 min on screen13 min totalPart listen, part watch
- Watch a company actually make "the specification is the product, the implementation is generated" work, instead of stopping at the slogan.Dominik Tornow · AI Engineer World's FairFor · Distributed systems engineers / Platform/infrastructure foundersAgentAI Coding5 min on screen18 min totalPart listen, part watch
- Don't blame the model for bad graphics. Change the medium: HTML is the agent's native language for visuals.Amol Kapoor · AI Engineer World's FairFor · AI presentation product leads / Engineers wiring tools to agentsAgentAI CodingDesign-to-CodeAI Product5 min on screen7 min totalNeeds your eyes
- Don't let eval signal die in the dashboard — feed it back as retrieval weight, with benchmarks and a live demoSonam Pankaj · AI Engineer World's FairFor · Agent memory / RAG engineers / Developers stuck with agents that keep failing the same wayAgentEvalsContext5 min on screen16 min totalPart listen, part watch
- Accuracy falls to 13.6% at 741 tools: measured numbers on how tools should be selectedSohail Shaikh · AI Engineer World's FairFor · Agent engineers / AI platform architectsAgentContextEvals5 min on screen28 min totalPart listen, part watch
- A booking failure that forced out a four-layer prompt architecture: turning what the brand must never say into a hard rule.Isadora Martin-Dye · AI Engineer World's FairFor · LLM product leads / Conversational AI engineersContextAI ProductAgent5 min on screen21 min totalPart listen, part watch
- One brief Slack message — "fix this thing" — and OpenClaw already knows what you mean, with agents writing the code in parallel.Jeffrey Lee-Chan · AI Engineer World's FairFor · Experienced AI coding tool users / Platform / DevEx engineersAgentAI CodingContext6 min on screen15 min totalPart listen, part watch
- As models keep getting more powerful, the harness you fix ahead of runtime becomes the bottleneck — let it emerge instead.Rajiv Chandegra · AI Engineer World's FairFor · Multi-agent systems engineers / AI architectsAgentAI Coding6 min on screen37 min totalPart listen, part watch
- Same model, only the search tool swapped, and it nearly hits the ceiling - the bottleneck is retrieval, not reasoningAmir · AI Engineer World's FairFor · RAG / search agent engineers / People writing agent harnessesAgentEvalsContext6 min on screen14 min totalNeeds your eyes
- Carbon accounting has no single right answer, so he moved validation off the answer and onto the process.Andrew Dumit · AI Engineer World's FairFor · Agent platform leads / Product leads in domains with no answer keyAgentAI CodingEvalsContext6 min on screen17 min totalPart listen, part watch
- Models improve by the week, and the prompt box between you and them is still running the protocol of a punch card.Ted Johnson · AI Engineer World's FairFor · AI interaction designers / Agent developersAI ProductAgent6 min on screen20 min totalPart listen, part watch
- Three months hand-building a pocket AI terminal, from blown-up boards to a working RPGLech Kalinowski · AI Engineer World's FairFor · AI hardware makers / Embedded firmware engineersAI ProductAgentContext6 min on screen25 min totalPart listen, part watch
- Turns quantization, caching, and distillation for diffusion inference into a roadmap you can start on todayZiv Ilan · AI Engineer World's FairFor · Inference deployment engineers / Generative product leadsAI Product6 min on screen19 min totalPart listen, part watch
- 500 美元、21 小时 RL 训练，4B 模型工具调用打赢 235BKobie Crawford · AI Engineer World's FairFor · Agent developers / Enterprise tech leadsAgentEvals6 min on screen21 min totalPart listen, part watch
- 524 real PRs show what is dragging review down is not AI, it is a complexity bill nobody has put a number onSachin Gupta · AI Engineer World's FairFor · Engineering team lead / Senior code reviewerAI CodingEvals7 min on screen25 min totalPart listen, part watch
- A Microsoft engineer's four-part guardrail kit: with guardrails you're a 20x engineer, without them you're 20x the slop.Chris Noring · AI Engineer World's FairFor · Developers doing AI-assisted coding / Tech leadsAgentAI CodingContext7 min on screen23 min totalPart listen, part watch
- On Fable's launch day, an Anthropic member on why what contains the models is usThariq Shihipar · AI Engineer World's FairFor · Engineers who write prompts / Heavy Claude Code usersAI CodingAgentContextAI Product7 min on screen19 min totalPart listen, part watch
- A rare enterprise production record: a self-built agent loop, an A2A contract, sandboxing and humans in the loop, end to endGabe de Mesa · AI Engineer World's FairFor · Agent platform architect / AI application backend engineerAgentEvalsContextAI Product7 min on screen19 min totalPart listen, part watch
- Coding agents aren't short on intelligence, they're short on reliability. Recursive decomposition turns the lucky golden session into a repeatable result.Raymond Weitekamp · AI Engineer World's FairFor · Heavy coding agent users / AI agent buildersAgentAI CodingEvalsContext7 min on screen24 min totalPart listen, part watch
- Design agentic systems with the engineering discipline you already have — method you can copy, no trend talk.Angie Jones · AI Engineer World's FairFor · Agentic systems engineer / Tech leadAgentAI CodingContext7 min on screen20 min totalPart listen, part watch
- How an $85K failed banking chatbot shipped in 8 weeks — with the model picked in week 7Sandipan Bhaumik · AI Engineer World's FairFor · AI platform engineering lead / Data platform leadAgentEvalsAI Product7 min on screen37 min totalPart listen, part watch
- When agents scale infinitely, the bottleneck becomes your attention — a working developer's flow for not getting friedZack Proser · AI Engineer World's FairFor · Developers running agents in parallel / Engineering leadsAgentAI CodingContext7 min on screen25 min totalPart listen, part watch
- No Docker rebuild: add a decorator to a Python function and throw the code onto a GPU cloud with hot reload, right from your local IDEAudry Hsu · AI Engineer World's FairFor · Developers self-hosting models / Developers orchestrating multiple modelsAI CodingAI Product7 min on screen20 min totalPart listen, part watch
- Hand the whole agent iteration loop over to agents; the eval gate is the only termination condition.Benedikt Sanftl · AI Engineer World's FairFor · Production agent engineers / AI tech leadsAgentEvalsAI Coding8 min on screen35 min totalPart listen, part watch
- An open source library that turns PDFs into structure an LLM can actually understand — and lets you drop the vector databaseCedric Clyburn · AI Engineer World's FairFor · RAG engineers / Teams running everything in-houseContextAgent8 min on screen21 min totalPart listen, part watch
- From one odd nesting in ChatGPT's DOM to the full set of trade-offs behind isolating third-party UIFrédéric Barthelet · AI Engineer World's FairFor · MCP app developers / App submitters stuck in reviewAgentAI Product8 min on screen20 min totalNeeds your eyes
- Top-ten on the open leaderboard from a single GPU: DeepMind on when self-hosting Gemma 4 pays off, with a live demo.Gus Martins · AI Engineer World's FairFor · Leads evaluating self-hosting / Enterprise architects and legalAgentAI CodingEvalsAI Product8 min on screen21 min totalPart listen, part watch
- 18 minutes from an empty file to an ACP agent that reads and writes files and opens a terminalBennet Fenner · AI Engineer World's FairFor · Agent authors who want to plug into an editor / Technical decision-makers evaluating ACPAgentAI Coding9 min on screen18 min totalNeeds your eyes
- An hour on every MCP Apps primitive and how to get into the three big stores, ending on the claim that MCP is the new websitePietro Zullo · AI Engineer World's FairFor · MCP server developers / AI product foundersAgentAI Product9 min on screen29 min totalPart listen, part watch
- One metaphor, the amnesiac genius, explains the agent's multi-repo and memory problem — then a live demo fixes it without changing a line of code in any repoVictor Savkin · AI Engineer World's FairFor · Platform engineers / Tech leadsAgentAI CodingContext9 min on screen20 min totalNeeds your eyes
- One claims review case, walked end to end through the four steps from requirements to productionApoorva Joshi · AI Engineer World's FairFor · AI application developers / Tech leadsAI ProductAgentEvalsAI Coding10 min on screen29 min totalPart listen, part watch
- He takes the horror of an LLM with tool calls and settles it with Lean and formal proofAI Engineer World's FairFor · AI engineers wiring up tools / Owners of agent reliability11 min on screen21 min totalNeeds your eyes
- One copy change on screen raised reviewers' rejection rate by 21 pointsAngel Ortmann Lee · AI Engineer World's FairFor · Human-in-the-loop product designers / Owners of high-stakes review workflowsAI ProductEvalsAgentAI Coding11 min on screen26 min totalNeeds your eyes
- Expose your local agent safely, then change code from your phone and watch the UI updateNick Taylor · AI Engineer World's FairFor · Self-hosted agent developers / MCP tool developersAgentAI Coding12 min on screen17 min totalNeeds your eyes
- Swapping cloud Sonnet for on-device Llama 3.2 without losing performance: a right-sizing eval you can actually copyRachel Lee Nabors · AI Engineer World's FairFor · LLM application engineers / AI product leadsEvalsAI Product12 min on screen31 min totalNeeds your eyes
- The Chrome team explains WebMCP with real demos: register tools on your site so agents can stop screenshotting and clickingTara Agyemang · AI Engineer World's FairFor · Front-end and full-stack developers / Website tech leadsAgentAI Product12 min on screen22 min totalNeeds your eyes
- A hands-on audio AI class carried almost entirely by live demos, from transcription to singingThor Schaeff · AI Engineer World's FairFor · Choosing a real-time voice stack / Voice feature developersAI ProductAgentAI Coding12 min on screen20 min totalNeeds your eyes
- Same model, only the harness changed — the score jumped more than 20 points. The evidence and the how-to are both in this talk.Aditya Bhargava · AI Engineer World's FairFor · Developers building their own agents / Anyone deciding which model to useAgentAI CodingEvalsContext13 min on screen32 min totalNeeds your eyes
- Real eval scores on screen: the full loop for letting a coding agent tune an AI agentAlfonso Graziano · AI Engineer World's FairFor · AI application engineers / Tech leadsAgentEvalsAI Coding13 min on screen30 min totalNeeds your eyes
- Prime Intellect's applied research lead opens up the whole open-source RL stack: the real cost sheet for a $50K full RL run on a frontier modelWill Brown · AI Engineer World's FairFor · RL training engineers / Technical decision-makersAgentEvalsAI Coding15 min on screen47 min totalPart listen, part watch
- Ten thousand notes in practice: drop the vector database, let plain markdown make your knowledge reusable by agentsPaul Iusztin · AI Engineer World's FairFor · Engineers building their own memory layer / Heavy second brain usersAgentContextAI Coding15 min on screen40 min totalPart listen, part watch
- Psychometrics moves into LLM evals: flag the bad items, cut 80% of the questions, and detect distillation kinshipAlejandro Vidal · AI Engineer World's FairFor · Benchmark maintainers / Whoever decides which model to useEvals17 min on screen24 min totalNeeds your eyes
- The Local AI inflection point in an hour: once a GPT-4o-class model runs on your phone, privacy and cost start rewriting the industry.Alex · AI Engineer World's FairFor · Anyone evaluating local deployment / Whoever owns the AI budgetAgentAI CodingAI ProductContext19 min on screen44 min totalNeeds your eyes
- An OpenAI engineer's argument: every agent sandbox ends at micro VMs — skip two years of grief.Abhishek Bhardwaj · AI Engineer World's FairFor · Sandbox infrastructure engineers / People choosing a platform for agentsAgentAI Coding20 min on screen45 min totalNeeds your eyes
- A reproducible experiment proves it: your character-AI evals can't catch a Hamilton who has read his own Broadway musicalJacob E. Thomas · AI Engineer World's FairFor · Eval leads for character AI / Prompt engineering leadsEvalsContextAI ProductAgent25 min on screen58 min totalNeeds your eyes
- RAG running live on an all-free local stack, proving clean chunking decides more than a big modelAbed Matini · AI Engineer World's FairFor · Engineers shipping RAG / Teams limited by budget and complianceContextAgentAI Product26 min on screen46 min totalNeeds your eyes
- Five kinds of agent hallucination, five code demos run live side by side, all open source and reproducibleElizabeth Fuentes · AI Engineer World's FairFor · Engineers productionizing agents / Agent platform architectsAgentContextEvals42 min on screen55 min totalNeeds your eyes