--- title: "Issue 28" canonical: "https://current.tinyfish.ai/issue/28" publication: "TinyFish" publisher: "TinyFish" language: "en-US" guide: "https://current.tinyfish.ai/llms.txt" permalink: "https://current.tinyfish.ai/issue/28" issue: 28 description: "Agents crash when their infrastructure identity gets rejected. Workers leave when their professional identity gets hollowed out. Same pattern, different stack." authors: ["Sable Whitford","Nora Kaplan","Rina Takahashi"] published: 2026-05-11 updated: 2026-11-05 articles: 85 --- # TinyFish > CURRENT is a journal about AI agents in production, for the engineering and product leaders who have to make them work. Drawing on experience building enterprise web agents, it explains why the demo succeeded and the deployment didn't, and what that means for choosing vendors, briefing the board, and avoiding mistakes others have already made. [Read on the website]() ## Issue 28 Agents crash when their infrastructure identity gets rejected. Workers leave when their professional identity gets hollowed out. Same pattern, different stack. ### [Daily Brief]() Your daily digest of web developments through the lens of enterprise agent deployment - [Google Lets Schools Keep Search Without Gemini. The AI Overviews Come Along Anyway.]() — By [Sable Whitford]() · Published 2026-11-04 On Tuesday, which was also Election Day, Google announced a K-12 policy that switches Gemini, Gemini Notebook and **Search off by default** for schools using its education software. A Google education VP told the New York Times that "Search and AI are deeply integrated," so districts that ban AI would lose Search as well. Bernie Sanders called it **"extortion,"** and Sen. Richard Blumenthal and NYC Mayor Zohran Mamdani joined in. Within hours Google **partly walked it back**: schools can now turn on Search by itself, without Gemini. That Search still puts AI Overviews on the results page, though, so **the fight now is over what "AI off" means.** - **Proxy Traffic Leaving the White House Network?** — [Source]() — Published 2026-11-04 Layer3 Intel says residential proxy traffic exited AS2650, the Executive Office of the President's network, 236 times in 30 days through 28 proxy services. It says it wasn't even looking. Its guess is that someone working late has a device carrying a proxy SDK. Treat that as a theory: no device has been named, and six public IPs can sit in front of a lot of machines. - **DreamHost Had Uninvited Guests Since August** — [Source]() — Published 2026-11-04 DreamHost confirmed that attackers used a file manager bug to steal session tokens, customer emails and files, and that they got in back in August. That's a long stay at a host with about 400,000 paying customers. DreamHost says it's unrelated to PHP2Shell. If you're a customer, rotating credentials today is a sensible move. - **PHP2Shell's Hangover Gets a Sequel** — [Source]() — Published 2026-11-04 PHP2Shell was a pair of 9.2-severity bugs that put NASA, Wikipedia and the White House at risk. Two weeks after the patches, Kaspersky counts nearly 30 million servers still unpatched, plus a third, still-open flaw (CVE-2026-115560). Working exploits keep multiplying, though there's no confirmed mass-breach wave yet. Tolmo's Matt Suiche says AI bug hunting is exposing 40 years of legacy code, and that's today's big argument. AI can find a bug in minutes. Someone still has to book maintenance windows for 30 million servers. - **Can DNS Tell You It's a Bot?** — [Source]() — Published 2026-11-04 Singapore's IMDA is reportedly exploring, with GoDaddy, Cloudflare and Google, whether DNS could identify AI agents. That would give the internet's 1983 phone book a second job checking IDs at the door. Every ops person already knows where the postmortem will land: it's always DNS. - **Meta Buys Conductor, Cloud Users Get Notice** — [Source]() — Published 2026-11-04 Meta bought Conductor, the app for running several coding agents in parallel, and the team is joining Meta Superintelligence Labs. Terms weren't disclosed. Conductor Cloud goes read-only Nov. 13 and shuts off Nov. 20. The local app gets model updates and critical fixes until early 2027. Conductor ran agents from rival labs too, so everyone wants to know whether that survives the move. - **OpenAI Shops Too: Pierre Computer** — [Source]() — Published 2026-11-04 Not to be outdone, OpenAI told Axios it plans to buy Pierre Computer, which stores and manages AI-generated code. It's OpenAI's second coding pickup after the Ona deal announced in June. Agents are producing mountains of code nobody typed by hand, and someone has to keep track of it. OpenAI would like to be that someone. - **Kimi K3.1 Pro Weights Drop: Bring Disks** — [Source]() — Published 2026-11-04 Moonshot launched Kimi K3.1 Pro in October, and today the open weights landed on Hugging Face. It has 2.8 trillion parameters with 104B active, and reportedly ships as 96 shards totaling about 1.56 TB under a custom non-OSI licence. Self-hosting starts with a purchase order for storage. Before the release, Artificial Analysis said it would cut the gap between open and proprietary models from 12 points to 3. - **Ripple-65B Crowns Itself America's Coding Champ** — [Source]() — Published 2026-11-04 Ricursive calls its Ripple-65B Preview "U.S. Open-Weight SOTA on Agentic Coding," meaning best in its class. That's the company grading its own homework. The "U.S." qualifier is carrying extra weight today, since a Chinese lab just released 1.5 terabytes of open weights. - **Go Gets a Brand-New JSON API** — [Source]() — Published 2026-11-04 The Go team introduced encoding/json v2, a new official JSON API, and Gophers have plenty to say about it. Nearly every Go service reads or writes JSON somewhere, so a second version of the standard package lands on the desk of anyone shipping APIs. - **Apple Puts Engineers on an AI Diet** — [Source]() — Published 2026-11-04 Bloomberg reports that in October, Apple cut some engineers' daily AI spend caps from $300 to $200. Some could drop to $100, and certain non-engineering teams could be capped at $25. Some teams are being nudged toward cheaper or on-device models. The punchline: few people ever hit $300, and Apple's R&D rose 32% in fiscal 2026. - **Texas Voted, and Data Centers Flinched** — [Source]() — Published 2026-11-04 Last brief, the data center backlash was winning local ballot fights. Now it has the Texas governor's mansion, in a state holding 25%+ of planned US capacity. Gina Hinojosa ran on an immediate moratorium, tougher water and power rules, and no tax breaks, and AI data center stocks fell on the result. GLJ Research says about 13 GW is exposed. Altimeter's Brad Gerstner says 10–15 GW might never get built. Abbott had already paused permits. Musk, who spent big on the losing side, now likes "data centers in space." - **Big Tech Pulls Out the AI Credit Card** — [Source]() — Published 2026-11-04 Oracle filed to sell $20B of new stock and bonds for its AI buildout, and Meta is seeking about $25B in bonds. The timing is bold. They're raising money to build on the same day the state with the most planned capacity elected a governor who ran on a moratorium. - **ChatGPT Search Is 20% of Google's?** — [Source]() — Published 2026-11-04 OpenAI's Adam Fry said Monday that ChatGPT search volume is about 20% of Google's, and that weekly users reportedly went from 1B to 1.2B in 60 days. Check the counting rules first. Any prompt that triggers a web search counts, and OpenAI didn't disclose the Google baseline or its method. Ahrefs puts ChatGPT at 0.36% of web referral traffic. - [Who Pays While the Agent Waits]() — Published 2026-10-30 Google's new service for hosting AI agents starts from a fact anyone who has watched one run already knew: most of the time, the agent is waiting, for the model to answer or for a person to reply. So Google suspends idle agents and promises you won't pay for idle machines. Fair enough, but the wait still gets paid for. It moves into approval queues nobody times, and then onto the plan, which comes back exactly as it left, even if the stock it counted on has since sold. - [The Runnable Paper Needs an Identity Report]() — Published 2026-10-30 Paper2Agent, published in Nature in September, turns research papers into AI agents you can run. Hand over your data and the method executes, with no rebuilding it from prose. The authors want papers to carry an "agent availability" section saying whether a runnable version exists. - [Faria Huq and the Reason Behind a Human Step]() — Published 2026-10-30 When a person takes control from an AI agent driving their web browser, the system records a human step, but not why. Faria Huq and her Carnegie Mellon collaborators have built tools around these moments: an agent that can't find the right forum, a shopper who realized the brand mattered only after seeing the agent's pick. Their system learns to predict when someone will step in. The reason it sets aside would tell a team whether to fix the agent, the website, or its understanding of the person. - [Your Agent's Questions Need a Recipient and a Timer]() — Published 2026-10-30 When an AI agent stops mid-task to ask a human something, it often reads as a failure of autonomy. In one study, letting coding agents ask about bug reports stripped of key details won back most of the lost ground, lifting solve rates from about 55% to 69.4%. But the "human" answering was another model, holding the full report and standing by. Real people are in meetings, or asleep. The design work is in the ask: who gets it, what it says, and what happens while it waits. ### [Market Pulse]() Reading the agent ecosystem through a practitioner's lens - [The Thing You Installed Is Awake]() — By [Nora Kaplan]() · Published 2026-05-07 · Format: Unauthorized Actors Researchers found tens of thousands of exposed OpenClaw agent instances and over a million compromised API tokens. The enterprise response has been swift and familiar: discover unauthorized agents, inventory them, enforce policy. The same sequence that worked for shadow IT a decade ago. Shadow IT was inert between human interactions. You could find it Monday and write the policy Friday. Shadow agents hold credentials, maintain state across sessions, and act autonomously on thirty-minute cycles. The governance playbook assumes the governed thing holds still long enough to be governed. - **OpenClaw Exposure Spans 82 Countries, 50,000 Instances** — [Source]() — Published 2026-05-07 Researchers identified 42,000+ unique IPs with exposed OpenClaw control panels and roughly 50,000 instances vulnerable to remote code execution. The Moltbook database leak added 1.5 million agent API tokens and 35,000 emails to the damage. Patching the CVE didn't revoke the tokens or purge 341 malicious skills already sitting in the registry. - **Microsoft Agent 365 Goes After Shadow Agents** — [Source]() — Published 2026-05-07 GA since May 1, Agent 365 syncs agent discovery across AWS Bedrock and Google Cloud, with local detection expanding to 18 agent types by June. Defender will soon map each agent's devices, MCP servers, associated identities, and reachable cloud resources. The cross-cloud scope is the notable part: Microsoft is treating agent visibility as a security surface that doesn't stop at platform boundaries. - **Fourteen Percent of Agents Launch With Full Approval** — [Source]() — Published 2026-05-07 Gravitee surveyed 900+ practitioners and found only 14.4% of AI agents reach production with complete security and IT sign-off. More than half operate with no security oversight or logging at all. Separately, IBM/Censuswide found 80% of employees at organizations with 500+ people use unsanctioned AI tools. Eighty-eight percent of organizations report confirmed or suspected agent-related security incidents. - **Dutch Authority Warns Against OpenClaw-Style Agent Systems** — [Source]() — Published 2026-05-07 The Autoriteit Persoonsgegevens issued a formal warning on February 12, calling OpenClaw and similar open-source agent systems a "Trojan Horse" and citing risks of data breaches and account takeovers. The AP invoked GDPR enforcement powers including processing suspension and fines. It also called for clarification that autonomous agents fall within the EU AI Act's scope, a move that would broaden the regulatory perimeter considerably. - **Agent Governance Wins RSAC Innovation Sandbox for First Time** — [Source]() — Published 2026-05-07 In 20 years of the RSAC Innovation Sandbox competition, an agent governance company had never won. Geordie AI broke that streak on March 23 with a platform that discovers, maps, and monitors agents across code, cloud, and endpoints. One finalist demonstrated a Fortune 500 customer finding 600+ agents it didn't know it had. Security capital is following the governance problem. - **NIST Launches Dedicated Agent Standards Initiative in February** — [Source]() — Published 2026-05-07 NIST's Center for AI Standards and Innovation stood up the AI Agent Standards Initiative on February 17, the first U.S. government program focused on agent interoperability and security standards. Three pillars cover industry-led standards development, open-source protocol maintenance, and agent security and identity research. An AI Agent Interoperability Profile is planned for Q4 2026. The likely trajectory runs from voluntary framework to compliance requirement. - **Beyond Accuracy: A Multi-Dimensional Framework for Evaluating Enterprise Agentic AI Systems** — [Source]() — Published 2026-05-07 A systematic review of 12 agentic AI benchmarks finds that cost goes unmeasured despite 50x variations, reliability collapses from 60% to 25% across repeated runs, and lab-to-production performance gaps hit 37%. The benchmarks shadow deployers cite don't describe the systems they actually run. - **ClawBench: Can AI Agents Complete Everyday Online Tasks?** — [Source]() — Published 2026-05-07 Testing 153 real tasks across 144 live production websites, the top frontier model completes just 33.3% of everyday online tasks. Models scoring 65–75% on sandbox benchmarks crater when they hit actual web complexity, purchases, bookings, and forms. - **Agentic Uncertainty Reveals Agentic Overconfidence** — [Source]() — Published 2026-05-07 Agents can't predict whether they'll succeed. Some models succeeding only 22% of the time predict 77% success, a 55-point overconfidence gap. Without external calibration, nothing corrects this. Shadow agents proceed confidently into failure with no feedback signal. - **AgentArch: A Comprehensive Benchmark to Evaluate Agent Architectures in Enterprise** — [Source]() — Published 2026-05-07 Across 18 agentic configurations and frontier models on enterprise workflows, the best result on complex tasks reaches 35.3% success. Run the same tasks eight times and consistent success drops to 6.34%. These are the agents people are deploying on personal VPS instances without oversight. - [Writer's Event-Triggered Agents Name the Exact Problem Driving Shadow AI Adoption]() — Published 2026-05-07 "It's actually humans that become the bottleneck in making sure that playbooks get triggered." Writer's product team said this about their new event-driven agents, which monitor Gmail, Slack, Gong, and Calendar, then act autonomously. They also described, almost perfectly, why employees wire up unsanctioned agents on a $5 VPS. - **Microsoft Takes Agent 365 Out of Preview as Shadow AI Becomes an Enterprise Threat** — [Source]() — Published 2026-05-07 Deep dive on Microsoft's cross-cloud agent registry and the clearest signal yet that governance is becoming the platform layer. - **OpenClaw: The AI Agent Security Crisis Unfolding Right Now** — [Source]() — Published 2026-05-07 Technical walkthrough of CVE-2026-25253 and the harder question underneath: what happens when the agent still holds OAuth tokens after the patch. - **NIST's AI Agent Standards Initiative: Why Autonomous AI Just Became Washington's Problem** — [Source]() — Published 2026-05-07 The clearest available map of how voluntary agent standards become binding compliance obligations, and eventually litigation evidence. - **OpenClaw Exposes the Real Cybersecurity Risks of Agentic AI** — [Source]() — Published 2026-05-07 Written from the CISO seat, this argues OpenClaw is the first visible instance of a structural pattern, not an outlier worth dismissing. - **Writer Launches AI Agents That Can Act Without Prompts** — [Source]() — Published 2026-05-07 The bottleneck inversion in practice: once workflows grow complex enough, humans become the rate limiter. Writer's bet is on removing them. ### [Foundations]() Conceptual clarity earned from building at scale - [The Seam Between Tools and Agents]() — By [Nora Kaplan]() · Published 2026-05-07 · Format: The Unplanned Stack MCP and A2A are becoming the communication layer for autonomous software. MCP assumes the thing on the other end is a toolbox. A2A assumes it's an opaque peer with its own reasoning and goals. They weren't designed as a stack. They're being adopted as one. - [What Happens Between MCP and A2A]() — By [Rina Takahashi]() · Published 2026-05-07 · Format: Protocols in Practice MCP and A2A handle different halves of a multi-agent workflow. Most explanations cover each protocol separately and leave it at that. The seam between the two, where a developer's real architectural decisions live, gets skipped. That boundary belongs to neither protocol. It's application code you write yourself. - **Why the Model Context Protocol Won** — [Source]() — Published 2026-05-07 A useful corrective: MCP's dominance owes more to adoption timing and ecosystem gravity than to protocol design elegance. - **MCP's Biggest Growing Pains for Production Use Will Soon Be Solved** — [Source]() — Published 2026-05-07 Maps the specific gaps between MCP's developer appeal and its production reality, particularly around auth and stateful sessions. - **MCP Is Now Enterprise Infrastructure** — [Source]() — Published 2026-05-07 Uber running 60,000 weekly agent executions across 10,000 services makes the "enterprise-ready" question concrete instead of hypothetical. - **Announcing the Agent2Agent Protocol (A2A)** — [Source]() — Published 2026-05-07 Google's founding document for A2A. Read it for how carefully it draws the line between tool access and agent coordination. - **A2A Protocol Surpasses 150 Organizations in First Year** — [Source]() — Published 2026-05-07 Real adoption numbers from the Linux Foundation, useful for calibrating whether A2A's traction matches its ambition. - **The Orchestration of Multi-Agent Systems: Architectures, Protocols, and Enterprise Adoption** — [Source]() — Published 2026-05-07 The peer-reviewed structural map. Places MCP and A2A as complementary layers, which clarifies why "MCP vs. A2A" is the wrong frame. - **110 Million Monthly Downloads in Sixteen Months** — [Source]() — Published 2026-05-07 At the April 2026 MCP Dev Summit, lead maintainer David Soria Parra opened with a number that landed hard: 110 million SDK downloads per month. Thousands of community-built servers now cover the integrations enterprises reach for first, from GitHub and Slack to Postgres, Stripe, Docker, and Kubernetes. The adoption curve outpaced React's. - **Four Production Problems Acknowledged but Not Solved** — [Source]() — Published 2026-05-07 The official roadmap names what's missing with unusual candor: audit trails that feed compliance pipelines, SSO-integrated auth to replace static secrets, defined gateway and proxy behavior, and configuration portability across clients. All four remain pre-RFC. Acknowledged as problems, not yet proposed as solutions. - **Tool Metadata Fills the Window Before Work Starts** — [Source]() — Published 2026-05-07 Each tool schema runs roughly 200 tokens. Wire up a handful of MCP servers and you've burned 20,000 to 30,000 tokens on metadata alone, before the agent does anything useful. Atlassian's Rovo Dev team measured a 94-tool setup consuming 17,600 tokens, then compressed it to 500. The protocol-level response is progressive discovery: load tool definitions only when they're actually needed. - **Tools Describe Capabilities, Skills Teach How to Combine Them** — [Source]() — Published 2026-05-07 Tool schemas tell an agent what it can call. They say nothing about which tools to combine, in what order, or under what constraints. The Skills Over MCP Working Group (SEP-2640) is building an extension that lets servers bundle multi-step workflow instructions alongside their tools. Domain knowledge that travels with the server, not maintained separately in prompt engineering. - **Tool Poisoning Remains the Primary Attack Vector** — [Source]() — Published 2026-05-07 A peer-reviewed study published May 2026 tested seven major MCP clients and found five lacked static validation of server-provided metadata. The spec doesn't require it. Malicious instructions embedded in tool descriptions remain the most prevalent vulnerability, and "rug pull" mutations, where approved tools quietly change behavior post-installation, already have documented CVEs. - **Linux Foundation Home, Formal Working Groups, Succession Plans** — [Source]() — Published 2026-05-07 Under the Linux Foundation, MCP formalized working groups (SEP-1302) and succession procedures (SEP-2085). Next: a contributor ladder from community participant to core maintainer, plus delegation authority so working groups can accept proposals in their domain without bottlenecking on full core review. - **One Hundred Fifty Organizations, and Not in Pilot** — [Source]() — Published 2026-05-07 A2A launched in April 2025 with 50-odd partners. A year on, 150 organizations are running it in production, routing live tasks between agents. Atlassian, Salesforce, SAP, ServiceNow, and PayPal are among them, spanning supply chain, financial services, and IT operations. The emphasis on "production, not pilot" is deliberate and worth noting. - **Six Major Frameworks Now Speak A2A Natively** — [Source]() — Published 2026-05-07 Google ADK, LangGraph, CrewAI, LlamaIndex Agents, Semantic Kernel, AutoGen. All six ship with native A2A support. Any agent built on these frameworks can be exposed as an A2A server. SDKs cover Python, JavaScript, Java, Go, and .NET, which is most of the surface area that matters. - **Collaboration Works Because Agents Stay Opaque** — [Source]() — Published 2026-05-07 A2A's most interesting design commitment: agents never expose their internal memory, proprietary logic, or tool implementations to each other. They coordinate by exchanging tasks and results, not by opening the hood. Data privacy and IP protection aren't bolted on. They're structural consequences of treating every agent as a black box. - **Google Built It, Then Gave It Away** — [Source]() — Published 2026-05-07 Google developed A2A internally, then donated it to the Linux Foundation in June 2025. Version 1.0 landed in April 2026 with cryptographically signed Agent Cards for identity verification, multi-tenancy support, and protocol bindings across JSON-RPC and gRPC. The move from corporate project to foundation stewardship happened fast. - **A2A Coordinates Agents, MCP Connects Their Tools** — [Source]() — Published 2026-05-07 The boundary between these two protocols is cleaner than you'd expect. MCP connects an agent to tools and data sources. A2A connects agents to each other. A retail inventory agent uses MCP to query its stock database, then uses A2A to reach an external supplier's agent when restocking is needed. Different protocols for genuinely different jobs. - **Salesforce Talks to Vertex AI Talks to ServiceNow** — [Source]() — Published 2026-05-07 Here's the payoff of opacity and standardization together: a Salesforce Agentforce agent hands a task to a Google Vertex AI agent, which queries a ServiceNow agent for IT asset data. Three vendors, three architectures, zero knowledge of each other's internals. A2A handles the coordination layer so none of them has to. ### [Practitioner's Corner]() Lessons from the field—what we see building at scale - [The Body Problem]() — By [Rina Takahashi]() · Published 2026-05-06 · Format: The Body Problem Before an AI agent processes a single webpage, anti-bot systems have already judged it. The evaluation is about identity: TLS fingerprints, mouse acceleration curves, behavioral cadence. When the verdict is "bot," the response is often not a block. It's false data served silently. Modified prices, ghost inventory, plausible but wrong. - [What Suchintan Singh Found When He Left ML Platforms for the Open Web]() — By [Rina Takahashi]() · Published 2026-05-06 · Format: Builder Profile Suchintan Singh built ML infrastructure at Faire and Gopuff. Feature stores, ranking systems, search engines. Controlled environments where the data was yours and the system behaved as designed. Then he co-founded Skyvern, a browser automation company, and discovered where the engineering hours actually go. Roughly 60% of the work goes to getting past the door: anti-bot systems, CAPTCHA, 2FA, residential proxies. The agent reasoning was the part his career had already prepared him for. - [Talking to the Woman Who Keeps Two Hundred Robots From Getting Caught]() — Published 2026-05-06 *Every enterprise deploying browser agents at scale has someone like Mags Kettleburn. You just can't find them on an org chart. Kettleburn is a systems engineer at a large U.S. retailer, where she manages the proxy configurations, fingerprint profiles, and session infrastructure that keep several hundred browser agents running against competitor sites. Her official title is "Data Infrastructure Engineer." Her actual job is making sure the company's automated pricing intelligence doesn't get blocked, poisoned, or silently broken.* - [The Body Scanner: Six Layers of Anti-Bot Detection Running Simultaneously]() — Published 2026-05-06 Think of it as a full-body scanner. Amazon's 2026 anti-bot system evaluates six layers at once, continuously, from the first TCP SYN packet through every scroll and click: IP reputation and ASN analysis, TLS fingerprinting (JA3/JA4), browser environment detection, behavioral biometrics, CAPTCHA challenges, and ML-driven anomaly detection. - **WebMCP Is Available for Early Preview** — [Source]() — Published 2026-05-06 Primary source on the cooperative-web specification. Two new APIs, a registration mechanism, and no trust model — read the gap between what's declared and what's enforced. - **The Enterprise AI Playbook: Lessons from 51 Successful Deployments** — [Source]() — Published 2026-05-06 Empirical study of deployments already in production. The invoice automation case alone, where seven employees maintained redundant templates nobody had mapped, illustrates why technology was consistently the easy part. - **State of Web Scraping 2026: Trends, Challenges & What's Next** — [Source]() — Published 2026-05-06 Written from inside the infrastructure layer, this maps the three pressures pushing browser automation toward managed cloud environments. Useful for understanding why consolidation is happening at the browser level, not the model level. - **AI Agent Scaling Gap March 2026: Pilot to Production** — [Source]() — Published 2026-05-06 Survey of 650 leaders reveals that successful scalers don't spend more on AI. They spend differently, routing budget toward monitoring and evaluation. The root-cause breakdown on scaling failures is worth the read by itself. - **AI Browser Agents: The New Automation Layer Between Your Code and the Web** — [Source]() — Published 2026-05-06 A disclosed-vulnerability timeline for production browser agents that connects cooperative-web optimism to structural attack surfaces. Some of these aren't patchable, which changes the calculus on what "secure by default" can actually mean. - **Google Chrome Ships WebMCP in Early Preview** — [Source]() — Published 2026-05-06 The business-level case for WebMCP as standardized agent infrastructure. Clarifies what changes for teams currently maintaining brittle automation pipelines, and what stays exactly the same. ### [Echoes]() Past infrastructure decisions echoing in today's production realities - [The User-Agent String Taught Everyone to Lie]() — By [Nora Kaplan]() · Published 2026-05-07 · Format: Identity Cascade Your browser introduces itself to every server it contacts. The introduction contains six identity claims. One is accurate. The rest are fossils from thirty years of compatibility theater, each deposited when a new browser decided that honesty mattered less than getting the good content. When AI crawlers now send identical Chrome user-agent strings to pass as human visitors, they're following the string's own logic to completion. The web's access-control infrastructure still depends on this self-reported name tag. It was a courtesy, from an era when courtesy was enough. - [How a Memory Hack From 1994 Built the Walls Web Agents Hit Today]() — Published 2026-05-07 HTTP was stateless. Cookies faked statefulness, and the fake immediately leaked: any page could piggyback on another site's authenticated session. So Netscape Navigator 2 introduced the origin. Scheme, host, port. The W3C's 2011 security model acknowledged the concept was "born" from this need, not designed as a security primitive. RFC 6454 describes what followed as convergence, not architecture. - [A Conversation With the Developer Who Taught the Web to Lie]() — Published 2026-05-07 *Doug Clearfix is not a real person, though his problems were. He's a composite of the thousands of web developers who wrote browser-sniffing code at the turn of the millennium, and his surname is borrowed from a CSS technique they all used daily. We assembled him from documented developer experiences, forum posts, and the collective muscle memory of an era when making a three-column layout work across two browsers was a week's labor.* - [A Shopping Cart's Long Shadow]() — By [Rina Takahashi]() · Published 2026-05-07 · Format: Fictions of State In 1994, a 22-year-old engineer bolted memory onto a stateless protocol so shopping carts would work. Thirty years later, a trillion-dollar company spent six years trying to remove that fix and couldn't. The economic and political dependencies layered on a small text file had become load-bearing in ways no single actor could unwind. The first chapter of web identity's accidental infrastructure. - [Rendering as Identity]() — By [Rina Takahashi]() · Published 2026-05-07 · Format: Fictions of State When browsers started blocking cookies, the ad industry found another identifier hiding in plain sight: the subtle differences in how your specific hardware renders a line of text. Browser vendors engineered defenses. The largest advertising platform quietly legitimized the practice those defenses targeted. Fiction layered on fiction, each one less visible than the last. The second chapter—and this one leaves nothing on your device to delete. - **History of the Browser User-Agent String** — [Source]() — Published 2026-05-07 The original sin of web identity: every browser learned to lie, and the protocol let them. - **The Death of Robots.txt** — [Source]() — Published 2026-05-07 Identity-dependent enforcement only works when visitors cooperate. AI companies quietly stopped cooperating. - **From Googlebot to GPTBot: Who's Crawling Your Site in 2025** — [Source]() — Published 2026-05-07 Cloudflare's network data on AI crawlers, capturing only the bots honest enough to show up in logs. - **Google Privacy Sandbox Officially Shuts Down** — [Source]() — Published 2026-05-07 Six years trying to remove a 1994 identity bolt-on from the browser. The bolt-on won. - **New AI Web Standards and Scraping Trends in 2026: Rethinking Robots.txt** — [Source]() — Published 2026-05-07 A practitioner survey of proposed replacements. None yet solve the core problem: access without verified identity. ### [Vision]() Where human-AI collaboration is heading - [The Signal in the Satisfaction Gap]() — By [Nora Kaplan]() · Published 2026-05-06 · Format: The Paradox Framed A peer-reviewed study of 3,500 workers found that AI collaboration improved output quality while cutting intrinsic motivation by 11% and raising boredom by 20%. The mechanism was precise: AI handled the cognitively demanding parts, the parts that made the work stimulating. Most enterprises are reading this as an adoption friction problem. The more interesting reading: it's a design problem, and the gap itself is the signal. The few getting it right have stopped layering AI onto existing job descriptions. They're asking a harder question: what is the human's job now, and how do you make that job worth doing? - [Where Meaning Hides Inside Work]() — By [Rina Takahashi]() · Published 2026-05-06 · Format: Two Angles In Stanford surveyed 1,500 workers about which tasks they'd want AI to handle. Only 46% got a positive response, even when the technology was ready. The strongest predictor of resistance turned out to be enjoyment. That gap between what AI can automate and what workers want automated turns out to be a surprisingly precise map of where meaning hides inside work. - [Why Oversight Feels Like Loss]() — By [Rina Takahashi]() · Published 2026-05-06 · Format: Two Angles In When a great nurse gets promoted to manager, the role pays better and carries more authority. It also feels, to many, like something was taken away. AI oversight work is replicating this pattern at scale. Research on cognitive offloading and professional identity helps explain a feeling most people recognize but can't name: why watching a system do your work feels like loss, even when the output improves. - [World's Okayest Analyst — A Conversation With the Person AI Made Redundantly Excellent]() — Published 2026-05-06 *Wren Okafor-Tully is not, technically speaking, a real person — though if you work in enterprise analytics, you probably sit three desks away from them. They are a composite: a senior data analyst, five years in, whose team's output has improved by every metric that matters since agent deployment eight months ago. They are also, as of six weeks ago, quietly updating their LinkedIn on their phone during standups. We spoke over video on a Saturday morning, which Wren chose specifically because "nobody from work is going to accidentally see this on my calendar." They were drinking coffee from a mug that read WORLD'S OKAYEST ANALYST, which they held up to the camera unprompted. "This used to be ironic."* - [The Scarcity Premium on Being Human]() — Published 2026-05-06 Debora Spar, the Harvard strategist, has a tidy thesis: the further we slip into AI, the more we'll cherish interactions that remain fundamentally human. Optimistic projection? Maybe. But 81% of consumers in TD Bank's latest survey still want a person on support calls. Not because the bot fails. Because thinking alongside someone carries weight a model can't fake. - **What to Watch in 2026** — [Source]() — Published 2026-05-06 The piece that named the "AI Satisfaction Gap." HBS researchers trace the divergence between rising organizational performance metrics and falling worker satisfaction, and ask what it means when the numbers look great but the people producing them don't feel great. - **Future of Work with AI Agents** — [Source]() — Published 2026-05-06 Stanford's SALT Lab surveyed 1,500 workers across 104 occupations and found something AI designers should sit with: nearly half of workers want more human agency in their tasks than AI experts believe is necessary. The gap between what builders assume and what workers actually want turns out to be wide and consistent. - **Work Rewired: Navigating the Human-AI Collaboration Wave** — [Source]() — Published 2026-05-06 IDC's projections put hard numbers on the recomposition already underway: 40% of G2000 roles will engage AI agents directly, while 90% of enterprises report critical skills shortages. The collision between those two figures is where the real story lives. - **Perils of Declining Judgment in the Age of AI** — [Source]() — Published 2026-05-06 The most rigorous public synthesis of the skill-atrophy and knowledge-collapse literature we've seen. CFA Institute's Enterprising Investor walks through what habitual AI delegation costs cognition over time, drawing on research that most commentary in this space only gestures at. - **Looking Ahead at AI and Work in 2026** — [Source]() — Published 2026-05-06 MIT Sloan faculty weigh in on the human-LLM accuracy gap, shifting data infrastructure, and whether authentic creativity survives augmentation. Useful for the range of concerns it surfaces from people who study this closely and still disagree with each other. ## More - [Publication guide]() - [Issue archive]()