--- title: "Stop Teaching Agents to Click" newsletter: "User Community" date: 2026-09-24 source: https://aaif.live/newsletters/usercommunity/2026-09-24-stop-teaching-agents-to-click --- # Stop Teaching Agents to Click *Plus… 68 ms voice agents, vending-machine benchmarks, agent payments, and engineering playbooks* *User Community — Agentic AI Foundation, 2026-09-24* Huge thanks to everyone that helped us pull off the 32-hour livestream last week.  Good news is the live stream recordings are up now. Check them out in case you want to relive some moments or catch a talk you may have missed. Day one [https://youtu.be/hbH2RbUBZQ0?si=SxD0O1XkpI8VP3DG&t=811] and day two [https://youtu.be/-w6RDNBAI0E?si=k-ZXkhFWVCSni-i-&t=70] ## Gems ## What’s happening across the chapters ## LOCAL ORGANIZERS Singapore held its first event, with Thu Ya K. covering A2A interoperability [https://lnkd.in/p/ed7mBxH6] and Weiyuan Liu sharing his “MCP sandwich” talk [https://lnkd.in/p/eu6PmtP2], while Luxembourg also kicked off its first event [https://lnkd.in/p/gG8DUiCY] with discussions around building and operating reliable agentic systems. In Ulaanbaatar, Munkh-Orgil Erdenekhuyag announced the chapter’s first public event [https://lnkd.in/p/gN7AgnD2], an Agentic Commerce Hackathon with Applied AI Mongolia, with registration open until October 2. Meanwhile, AAIF London organizer Ibrahim Malik reflected on AGNTCon + MCPCon Europe [https://lnkd.in/p/guT9Qymn], where he stepped in to speak, connected with builders and maintainers across the ecosystem, and started bringing ideas back for London. ## Engineering the agentic stack ## AGNTCON + MCPCON NORTH AMERICA Join developers, maintainers, engineering leaders and open-source contributors in San Jose on October 22–23 for two days focused on the systems behind agentic AI. Expect technical talks, implementation lessons and conversations around MCP, agent infrastructure, orchestration, security, observability, evaluation and the open standards shaping how agents are built and operated. It’s also a chance to meet the people behind the projects, compare approaches with teams working on similar problems, and come away with ideas you can apply to your own stack. Use code COMMUNITY25 for 25% off registration. Register for AGNTCon + MCPCon North America → [https://events.linuxfoundation.org/agntcon-mcpcon-north-america/register/] ## Building the next voice agents ## FREE VIRTUAL EVENT What does it take to move voice AI from a slick demo to something fast, reliable, and ready for production? Join us September 16 for 90 minutes on voice-agent architecture, infrastructure, and open standards, including UNMUTE, a new MIT-licensed standard for voice agents, plus a live look at how semantic caching, CDNs, and edge infrastructure can cut latency in voice pipelines. September 16 · 08:30 PDT / 17:30 CEST JOIN LIVE [https://home.mlops.community/home/events/voice-agent-virtual-event-bnzc6hiff8] ## Verification and recovery ## LUNCH AND LEARN WHAT HAPPENS AFTER A VERIFIER SAYS NO? Dolly Sah and Tanmay Sah’s Session 25 looked at what happens when agents can identify unsafe actions but struggle to recover from them. The write-up covers the verifier tax, unsafe success, and EvoUndo’s approach to making agent self-modification reversible. Read the write-up here [https://learn.mlops.community/wp-content/uploads/2026/09/Lunch-and-Learn-Session-25-asset.pdf]. This week’s Coding Agents Lunch & Learn is Friday, September 25 at 9 AM PT / 6 PM CEST. Session 26 drops the usual featured talk for an open discussion on coding agents - what people are building, which tools and workflows are working, where agents still fall short, and the challenges around context, verification, reliability and autonomy. Join the next session here. [https://home.mlops.community/home/events/coding-agents-lunch-and-learn-session-26-open-discussion-on-coding-agents-irtllqmk41?agenda_day=6ab15598c6df653e558f94ae&agenda_track=6ab1559ac6df653e558f94c6&agenda_stage=6ab15598c6df653e558f94b3&agenda_filter_view=stage&agenda_view=list] ## Can agents prove they followed policy? ## READING GROUP The next AAIF Reading Group looks at Measuring Agent Policy Compliance, a paper on evaluating whether AI agents follow their intended policies. The group will work through the paper’s methodology, assumptions and limitations, with plenty of room to question the approach and discuss what it means for agent evaluation. Thursday, October 1 · 9 AM PT / 6 PM CEST Join the reading group here. [https://home.mlops.community/home/events/aaif-reading-group-measuring-agent-policy-compliance-d6gfpflagm] ## The first official MCP certification is live ## NEW FROM AAIF AAIF and Linux Foundation Education have released the Model Context Protocol Associate (MCPA), a vendor-neutral certification based on the 2026-07-28 MCP spec. Candidates answer questions on client-server interactions and the tool invocation lifecycle, with 24% of the exam focused on security and governance. See exam details [https://training.linuxfoundation.org/certification/model-context-protocol-associate-mcpa/] ## Walking Tokyo Talking Agent Protocols Browser agents burn tokens taking screenshots, searching accessibility trees, and clicking through pages they barely understand. WebMCP offers a cleaner path by exposing page-specific tools inside the user’s authenticated session, alongside agent commerce and A2A. * In China, super-app ecosystems reduce integration overhead, but expansion beyond them increases the value of MCP and A2A. * WebMCP could make browser automation faster and more reliable by replacing visual interaction with contextual tools tied to the page. * Agentic commerce still needs identity, authorization, and payment mechanisms, with KYC extending to agents acting for customers. The shift is toward agents operating through explicit tools, permissions, and protocols rather than brittle UI automation. [https://podcasts.apple.com/gb/podcast/walking-tokyo-talking-agent-protocols/id1505372978?i=1000790526231](https://podcasts.apple.com/gb/podcast/walking-tokyo-talking-agent-protocols/id1505372978?i=1000790526231) [https://home.mlops.community/home/videos/walking-tokyo-talking-agent-protocols](https://home.mlops.community/home/videos/walking-tokyo-talking-agent-protocols) [https://open.spotify.com/episode/5nZJ6RJnvyky2ii98UFxxV?si=pQfPmrYrQBmNt7aST2SBGw](https://open.spotify.com/episode/5nZJ6RJnvyky2ii98UFxxV?si=pQfPmrYrQBmNt7aST2SBGw) ## Voice Agent - Virtual Event A semantic cache cut one voice-agent response from 1.1 seconds to 68 milliseconds in a live demo. Across the event, the recurring challenge was making real-time agents fast, controllable, and robust outside clean test environments. * Breaking monolithic prompts into scoped steps more than halved token use while making required checks enforceable. * Production latency depends on the whole stack: model choice, network distance, caching, turn detection, and prefetching. * Real-world reliability needs noisy-audio testing, P50/P95/P99 monitoring, human escalation, and simulations that target failures such as interruptions, transcription errors, and privacy leaks. Strong voice systems come from controlling context, infrastructure, audio, and evaluation together. [Watch the event](https://home.mlops.community/home/videos/voice-agent-virtual-event) ## The Winchester Mystery House Problem in AI development A workflow that cost $1 per 1,000 records at 90% accuracy reached 95% accuracy while costing less, after an optimizer moved 75% of the work out of the model and into code. That result sits inside a broader discussion about when agentic behavior should become a defined workflow. * Models increasingly inherit assumptions from the coding harnesses they were trained around, which can make custom harnesses harder to build. * Repeated agent tasks can often be “crystallized” into smaller models or conventional code once the process is understood. * DSPy separates task definitions from implementation, allowing prompts, models, and even harness code to be optimized without redefining the task. The practical direction is toward identifying which tasks still need open-ended agents and which are mature enough to become cheaper, more reliable workflows. [https://podcasts.apple.com/us/podcast/the-winchester-mystery-house-problem-in-ai-development/id1505372978?i=1000785551986](https://podcasts.apple.com/us/podcast/the-winchester-mystery-house-problem-in-ai-development/id1505372978?i=1000785551986) [https://home.mlops.community/home/videos/the-winchester-mystery-house-problem-in-ai-development](https://home.mlops.community/home/videos/the-winchester-mystery-house-problem-in-ai-development) [https://open.spotify.com/episode/4IjjWOjSq8XQPAXLq6ApRQ?si=TnlufikXRKqxS8ZHSpyYCQ](https://open.spotify.com/episode/4IjjWOjSq8XQPAXLq6ApRQ?si=TnlufikXRKqxS8ZHSpyYCQ) ## Introducing the First Open Source Vending Machine Benchmark A green “completed” result can still hide an agent that loses money. This benchmark turns a simulated vending business into a controlled environment for testing how tool-using agents handle pricing, inventory, purchasing, refunds, and changing demand. * Six machines across three locations expose agents to shared cash, limited working time, supplier failures, and demand shifts. * The simulator uses 22 MCP tools and reproducible rules for weather, seasonality, pricing, marketing, and customer preferences. * Early runs showed large differences in business outcomes between models, while higher API spend did not consistently produce better results. The useful signal is whether an agent can make sound operating decisions repeatedly, not simply finish the task. [Read the blog](https://aaif.io/blog/introducing-the-first-open-source-vending-machine-benchmark) ## 402 payment required: what enterprise MCP servers owe the agents that pay them A paid MCP tool creates more than a billing problem: the agent can choose the work, but it should not control the budget that approves it. x402 v2 and MCP 2026-07-28 make pricing and routing metadata more explicit, which shifts attention to where enforcement sits. * Servers should recompute prices from the current request, rather than trust values supplied by the client. * Replay protection, local payee checks, and independent budget controls become part of the execution path. * HTTP and MCP payment bindings expose different tradeoffs for gateways, clients, and shared infrastructure. The core design question is how to separate agent intent from the authority to spend. [Read the blog](https://aaif.io/blog/402-payment-required-what-enterprise-mcp-servers-owe-the-agents-that-pay-them) ## IN-PERSON EVENTS ## Come and connect * Amsterdam [https://luma.com/tavily-7u0e] - September 25 * Bengaluru [https://luma.com/igokwrsx] - September 26 * Dallas [https://luma.com/28y9e7b8] - September 26 * Sydney [https://luma.com/axep37gt] - September 29 * Liverpool [https://luma.com/syr6fwfu] - September 30 * Seattle [https://luma.com/aaif-sea-02] - September 30 * Melbourne [https://luma.com/ioxvp4zf] - October 1 * Toronto [https://luma.com/ej11d939] - October 5 Find your city here [https://aaif.io/events?tab=community], or start a chapter if there isn't one yet. ## VIRTUAL EVENTS ## Join from anywhere * Lunch and Learn Session 26 [https://home.mlops.community/home/events/coding-agents-lunch-and-learn-session-26-open-discussion-on-coding-agents-irtllqmk41?agenda_day=6ab15598c6df653e558f94ae&agenda_track=6ab1559ac6df653e558f94c6&agenda_stage=6ab15598c6df653e558f94b3&agenda_filter_view=stage&agenda_view=list] - September 25 * AAIF Reading Group [https://home.mlops.community/home/events/aaif-reading-group-measuring-agent-policy-compliance-d6gfpflagm] - October 1 --- Source: https://aaif.live/newsletters/usercommunity/2026-09-24-stop-teaching-agents-to-click