--- title: "Issue 29" canonical: "https://current.tinyfish.ai/issue/29" publication: "TinyFish" publisher: "TinyFish" language: "en-US" guide: "https://current.tinyfish.ai/llms.txt" permalink: "https://current.tinyfish.ai/issue/29" issue: 29 description: "Protocols are writing the specifications enterprises won't. Frameworks are rebuilding architectures they've never read. The defaults are hardening." authors: ["Sable Whitford","Rina Takahashi","Nora Kaplan"] published: 2026-05-18 updated: 2026-11-05 articles: 80 --- # 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 29 Protocols are writing the specifications enterprises won't. Frameworks are rebuilding architectures they've never read. The defaults are hardening. ### [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. - [Lebanon's Two Clocks and the Question Calendars Don't Ask]() — Published 2026-10-30 On March 26, 2023, Lebanon woke up on two clocks. The caretaker government had postponed daylight saving three days earlier. Churches and TV channels moved forward anyway, government offices stayed put, and most phones jumped on their own. Middle East Airlines printed every departure an hour earlier, so its planes still left at the same moment. That week made each calendar show what it had been storing: an exact instant, or a time on the wall that politics can move. Software assistants now make that choice without asking anyone. - [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. - [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. ### [Market Pulse]() Reading the agent ecosystem through a practitioner's lens - [When Infrastructure Writes the Spec]() — By [Rina Takahashi]() · Published 2026-05-13 · Format: Specification by Default A support platform prices its AI agents per automated resolution: no escalation, ticket not reopened within 72 hours. Clean metric — one where a customer who gives up counts the same as one who's satisfied. Most organizations deploying agents haven't specified what "good" means in operational terms. Infrastructure is filling that vacuum. Pricing models, protocol defaults, eval frameworks each encoding their own answer. And the specification decision already happened, whether anyone noticed or not. - **MCP Hits 97 Million Downloads, Governance Lags Behind** — [Source]() — Published 2026-05-13 MCP crossed 97 million monthly SDK downloads by March. 78% of enterprise AI teams report at least one MCP-backed agent in production. Now under Linux Foundation governance with OpenAI, Google, and Microsoft co-sponsoring, the 2026 roadmap puts stateless HTTP transport and a Tasks primitive under review. Neither has shipped. No Enterprise Working Group exists. Server counts range from 5,800 to 17,468 depending on who's counting. - **OpenAI Agents SDK Ships Native Sandbox Environments** — [Source]() — Published 2026-05-13 OpenAI's April 15 update adds sandbox environments where agents can inspect files, run commands, edit code, and handle long-horizon tasks inside controlled, siloed workspaces with scoped access. It targets the blast-radius problem directly: one of the core reasons production agents fail. TypeScript support still pending. - **EU AI Act Timeline Shifts Under Provisional Deal** — [Source]() — Published 2026-05-13 August 2, 2026 remains the legally binding enforcement date for EU AI Act transparency rules. A May 7 provisional agreement between Council and Parliament would defer national regulatory sandboxes to August 2027. The Digital Omnibus package may push standalone high-risk system deadlines to December 2027. The provisional agreement still requires formal adoption. - **Outcome Pricing Spreads as Seat-Based Models Shrink** — [Source]() — Published 2026-05-13 Zendesk charges $1.50 per automated resolution on committed volume, $2.00 pay-as-you-go. In January 2026, a billing change introduced automatic overage charges with no prior notification. Across the industry, seat-based pricing fell from 21% to 15% of SaaS companies in twelve months while hybrid models surged from 27% to 41%. Outcome pricing presupposes tasks that are measurable and bounded. - **Indirect Prompt Injection Now Documented in Production** — [Source]() — Published 2026-05-13 Google and Forcepoint documented production attacks: hidden instructions embedded in web pages, documents, and emails that agents process and execute. The AgentDojo benchmark, used by US and UK AI Safety Institutes, found effective defenses significantly degrade agent utility. Kiteworks 2026 data: 41–44% of organizations lack human-in-the-loop oversight; 55–63% have no kill switches, purpose binding, or network isolation. - **Towards a Science of AI Agent Reliability** — [Source]() — Published 2026-05-13 Decomposes reliability into consistency, robustness, predictability, and safety across 14 models over 18 months. Reliability gains lag capability progress throughout. The paper amounts to specification work: defining what "reliable enough" demands beyond raw accuracy. - **Beyond Accuracy: A Multi-Dimensional Framework for Evaluating Enterprise Agentic AI Systems** — [Source]() — Published 2026-05-13 Analyzed 12 benchmarks and found enterprise-critical dimensions going unmeasured: 50x cost variations, consistency drops from 60% to 25%, a 37% lab-to-production gap. The resulting CLEAR framework is, in effect, the specification document most deploying organizations still lack. - [When Regulation Does the Specification Work]() — Published 2026-05-13 Organizations have been rationally avoiding the hardest upstream work in AI deployment: articulating what their systems actually do, what they optimize for, and why those choices are defensible. That avoidance is political as much as practical. Writing the spec forces disagreements into the open. - **The 2026 MCP Roadmap** — [Source]() — Published 2026-05-13 The maintainers lay out what needs fixing. Enterprise readiness is pointedly left as "extensions, not core spec." - **State of AI Engineering** — [Source]() — Published 2026-05-13 Built from production traces, not self-reports. The surprise: 60% of LLM errors are rate limits, not model failures. - **State of Agent Engineering** — [Source]() — Published 2026-05-13 Survey of 1,340 practitioners. 57% have agents in production, but quality remains the top barrier at 32%. - **State of AI Agents 2026: 5 Enterprise Trends** — [Source]() — Published 2026-05-13 46% say systems integration is the primary challenge. Model capability barely shows up as a constraint. - **Indirect Prompt Injection Is Now a Real-World AI Security Threat** — [Source]() — Published 2026-05-13 Google and Forcepoint disclose production exploits. Documents the bind where effective defenses render agents unusable. ### [Foundations]() Conceptual clarity earned from building at scale - [What a Benchmark Score Actually Contains]() — By [Rina Takahashi]() · Published 2026-05-13 · Format: Reading the Leaderboard Claude Opus 4 scores 64.9% on GAIA in one scaffold and 57.6% in another. Same model, same benchmark, same questions. The seven-point gap comes entirely from the orchestration wrapping the model, and it's larger than many consecutive frontier releases produce. Agent benchmark scores carry information about scaffold and token budget alongside model capability, often in roughly equal measure. The numbers tell you where engineering leverage lives, if you're reading all three variables. - [Every Agent Framework Is an Argument About What's Hard]() — By [Rina Takahashi]() · Published 2026-05-13 · Format: Choosing the Scaffold Framework comparison tables show you what's available. The interesting part starts when a container restarts mid-workflow at 3 a.m. LangGraph, CrewAI, the Claude Agent SDK, and Microsoft Agent Framework 1.0 each embed a different argument about what the hard problem in agent orchestration actually is — control flow, task delegation, permission boundaries, enterprise interoperability. Picking one means committing to a control philosophy that shapes how your system encounters problems you haven't anticipated yet. A better approach: name your dominant constraint first, then find the framework whose core abstraction matches it. - **AI Evals Are Becoming the New Compute Bottleneck** — [Source]() — Published 2026-05-13 The cost data alone is worth the read: a 2% accuracy gap can mean 9x the spend, which reframes what "better" actually means on a leaderboard. Essential context on why benchmarks measure scaffold-model composites, not models. - **Demystifying Evals for AI Agents** — [Source]() — Published 2026-05-13 Gives agent evaluation a shared vocabulary (task, trial, transcript, outcome) and draws a careful line between scoring the final result and scoring each step along the way. The layered methodology here treats evals as ongoing discipline rather than a one-time gate. - **Evaluating AI Agents in Practice: Benchmarks, Frameworks, and Lessons Learned** — [Source]() — Published 2026-05-13 Useful for the gap it names between what metrics detect and where agents actually fail in production. The five-pillar readiness framework makes a concrete case for hybrid evaluation when purely quantitative benchmarks stop telling you anything actionable. - **Agentic AI Frameworks 2026: LangGraph vs CrewAI vs OpenAI SDK** — [Source]() — Published 2026-05-13 Treats framework selection as an architecture decision with real cost consequences. The comparison surfaces metrics vendors tend to skip, like cost-normalized accuracy and pass-at-k, which matter more than feature lists once you're running at scale. - **Agentic Orchestration: LangGraph vs CrewAI vs Mastra** — [Source]() — Published 2026-05-13 A seven-capability matrix that turns vague orchestration trade-offs into concrete decision axes: ramp time, durable execution, MCP support. Helpful for anyone trying to figure out which constraints actually bind in their specific production environment. - **GAIA: Your Scaffold Choice Outweighs Your Model** — [Source]() — Published 2026-05-14 GAIA evaluates general-purpose reasoning across web browsing, tool use, and multimodal tasks. What it won't tell you: how much credit belongs to the model. Swapping scaffolds shifts scores by roughly 30 points, far more than upgrading between frontier models. One run costs $2,829 before caching. When someone cites a GAIA number, the first question is which framework produced it. - **SWE-bench: Verified Flatters, Pro Doesn't Care** — [Source]() — Published 2026-05-14 Claude Opus 4.5 hits 80.9% on SWE-bench Verified and 45.9% on Pro. Three things compound: possible contamination (models reproducing verbatim patches), harder tasks in Pro requiring 10+ line changes, and Pro's controlled harness stripping away scaffold advantages. Verified measures something. Pro measures something closer to your actual codebase. - **τ-bench: The Benchmark That Caught Consistency Lying** — [Source]() — Published 2026-05-14 Sierra's τ-bench runs agents through customer service conversations, then runs them again. And again. GPT-4o passes a single attempt over 60% of the time. Require it to pass the same task across eight runs, and that number collapses below 25%. Most benchmarks report the generous number. τ-bench reports both, and the gap between them is the reliability problem production teams actually live with. - **Online Mind2Web: Nine Times the Price, Same Result** — [Source]() — Published 2026-05-14 Online Mind2Web tests agents on live websites that change underfoot. Browser-Use with Claude Sonnet 4 spent $1,577 to reach 40% accuracy. SeeAct with GPT-5 Medium hit 42% for $171. Two percentage points apart, ninefold cost difference. The benchmark measures web task completion. It accidentally also measures how wildly scaffold-model pairing affects economics. - **HAL: Forty Thousand Dollars of Honest Accounting** — [Source]() — Published 2026-05-14 Princeton's Holistic Agent Leaderboard burned $40,000 on 21,730 rollouts across 9 models and 9 benchmarks, standardizing scaffolds and tracking costs that individual leaderboards quietly omit. Independent sweeps revealed a 33x cost spread on identical tasks from scaffold choice alone. HAL has since paused accuracy updates entirely, pivoting to reliability measurement. Even the meta-benchmark decided the headline numbers weren't the point. - **Full Auditability Over Every Transition: LangGraph** — [Source]() — Published 2026-05-14 If your hardest problem is knowing exactly what happened at each step of a branching, retrying workflow, LangGraph's graph state machine saves typed snapshots at every super-step. Time-travel debugging and human-in-the-loop interrupts are built into the core abstraction, not layered on top. - **Deep Claude Integration: The Agent SDK** — [Source]() — Published 2026-05-14 If your hardest problem is tight coupling with Claude models, the Claude Agent SDK gives you four extension points wired directly into the model loop: MCP for tools, skills for reusable procedures, hooks for lifecycle control, subagents for parallel isolated work. Same architecture that runs Claude Code. - **Agents as Teammates: CrewAI Maps Roles Directly** — [Source]() — Published 2026-05-14 If your hardest problem is getting specialized agents to coordinate without hand-drawing graph topology, CrewAI lets you define agents by role, goal, and backstory. Delegation happens across sequential, hierarchical, or consensual patterns. You can ship a working multi-agent system before sketching a single node. - **Microsoft Consolidates: One Framework Replaces Two** — [Source]() — Published 2026-05-14 If your hardest problem is enterprise buy-in within Azure, Microsoft Agent Framework 1.0 (GA April 2026) merges Semantic Kernel and AutoGen into a single .NET and Python SDK. Both predecessors are now in maintenance mode. Swapping between six model providers takes one line of code. - **Surviving Crashes and Pauses: Durability Across Frameworks** — [Source]() — Published 2026-05-14 If your hardest problem is workflows that wait on humans for days or die mid-run, durability outranks every other feature. LangGraph checkpoints natively at every graph boundary. Microsoft Agent Framework 1.0 supports pause-and-resume across all orchestration patterns. Without durability, a crashed agent restarts from zero. - **Routing Requests to Specialists: OpenAI Agents SDK** — [Source]() — Published 2026-05-14 If your hardest problem is triaging incoming requests to the right specialist agent with safety checks in place, OpenAI's Agents SDK treats handoffs and guardrails as native primitives with automatic tracing. Code-first, no graph to predefine. The tradeoff: less portable, and less suited to stateful long-running work. ### [Practitioner's Corner]() Lessons from the field—what we see building at scale - [The Checkpoint That Approves Everything]() — By [Nora Kaplan]() · Published 2026-05-13 · Format: The Rubber Stamp Most enterprise teams running agent workflows have never checked the approval rate on their human-in-the-loop controls. When someone finally does, the number lands above 95%. Nearly everything gets waved through. The reviewers are behaving rationally given where the gate sits. How organizations build these checkpoints where they do, why the checkpoints persist long after anyone could justify their placement, and why replacing them with something real is structurally harder than it sounds: that's where the number becomes interesting. - [When Should an Agent Stop Thinking?]() — By [Rina Takahashi]() · Published 2026-05-13 · Format: The Autonomy Boundary Government portals, insurance dashboards, vendor procurement systems. No API, no programmatic access. Just a browser and forms built for humans. Suchintan Singh's Skyvern navigates these surfaces by reasoning through a workflow once with an LLM, then compiling the result into deterministic code. No model in the loop on the second run. The architecture is designed to need less intelligence over time, which amounts to a specific claim about where judgment actually matters in agent systems, and where it's just cost. - [Bev Okoro Approves 340 Items on a Tuesday]() — Published 2026-05-13 *Bev Okoro is a Procurement Operations Lead at a mid-market packaging and fulfillment company in the Midwest. She oversees agent-generated purchasing decisions — shipping routes, vendor payments, inventory reorders — approving upwards of 300 per day. Before operations, she spent six years as an EMT. She agreed to this conversation on the condition that we note she is, in her words, "a composite of every ops person who has ever stared at an approval queue at 3 PM and felt their soul leave their body." We have honored this request.* - [BakerHostetler's ROSS Deployment: A Checkpoint Placed Where It Counts]() — Published 2026-05-13 BakerHostetler pointed ROSS Intelligence at 27 terabytes of bankruptcy filings and told it to surface citations, retrieve documents, and identify precedent. Attorneys kept one job: deciding what the research meant. - **The Agentic Enterprise: Where Should Humans Stay in the Loop?** — [Source]() — Published 2026-05-13 The sharpest move here is the distinction between deliberate and reflexive checkpoints, and the diagnostic that follows: if your approval rate sits above 95%, the human in your loop has become decorative. That number should unsettle anyone building oversight into agent workflows, because it implies most checkpoints degrade into confirmation theater the moment they encounter real operational volume. The piece gives you a framework for asking where human judgment actually changes outcomes versus where it just adds latency and a false sense of control. - **Enforcing Human-in-the-Loop Controls for AI Agents** — [Source]() — Published 2026-05-13 Where the Board of Innovation piece asks where to place checkpoints, Prefactor asks what happens at the checkpoint itself. The practical controls perspective reveals a problem that policy-level thinking tends to skip: a well-placed approval gate still fails when the interface starves the reviewer of the context needed to make a real decision. Worth reading alongside the Board of Innovation piece as a reminder that "human in the loop" describes a design surface, not a solution. - **AI Reliability Is a Decade-Old Problem — And We're Still Only Solving Half of It** — [Source]() — Published 2026-05-13 The compound failure math here is the piece's real contribution: a 95% step-level success rate across a 10-step agent workflow yields roughly 60% end-to-end reliability. That arithmetic turns oversight from a governance preference into a structural necessity. It also explains why agent failures cluster in production rather than demos, since demo-length workflows never accumulate enough steps to surface the compounding. - **The 5 AI Agent Failure Modes: Why They Fail in Production** — [Source]() — Published 2026-05-13 A taxonomy built from 591 production incidents, and the headline finding is striking: 88% of failures traced to infrastructure gaps rather than model errors. That ratio matters for oversight design because it suggests most human-in-the-loop systems are watching the wrong layer. If your checkpoint is designed to catch the model doing something wrong, you're auditing the component responsible for roughly one in eight failures. - **EU AI Act Article 14: Human Oversight Requirements** — [Source]() — Published 2026-05-13 With binding enforcement arriving in August 2026, Article 14 transforms "meaningful human oversight" from an engineering best practice into a compliance obligation with real consequences. The legal text is worth reading directly because its requirements, particularly around the ability to "correctly interpret the high-risk AI system's output," presuppose a sensing capability that most deployed agent systems simply do not have. The gap between what Article 14 requires and what current oversight tooling delivers is about to become very expensive. ### [Echoes]() Past infrastructure decisions echoing in today's production realities - [The Specification Nobody Read Twice]() — By [Rina Takahashi]() · Published 2026-05-13 · Format: The Rediscovery When an agent workflow fails at step seven, you need steps one through six to unwind in reverse order. Builders working on today's agent orchestration frameworks are designing exactly this logic, calling it the "saga pattern." In June 2007, a consortium of enterprise vendors published a specification that formalized the same mechanism as a "compensation handler," along with durable execution, state persistence, and human task lifecycles. The problems never went away. But the vocabulary they were solved in got rejected wholesale, and the knowledge went with it. It took nearly eighteen years for anyone to formally notice. - [Four Agent-Era Concepts That Already Had Names]() — Published 2026-05-13 The agent ecosystem keeps coining terms for things that already had perfectly good names. Not analogues or spiritual predecessors. The same mechanisms, wearing different syllables. - [The Same Plumbing, a Trendier Coffee Shop]() — Published 2026-05-13 *Henk Verloop is not, strictly speaking, a real person. He is a composite — assembled from documented BPEL-era practitioner experience, real specification documents, and the kind of career arc that a decade of enterprise process orchestration produces. If you worked in SOA between 2003 and 2012, you probably sat next to someone like him. He would like you to know that the travel-booking example was his first.* - [Every Protocol Inherits What the Last One Deferred]() — By [Nora Kaplan]() · Published 2026-05-13 · Format: Two Returns CORBA deferred its security and versioning problems. SOAP inherited them, declared governance "outside the scope," and passed them forward. REST made the whole burden invisible. Now MCP has shipped to production with authentication still optional and over 1,800 servers exposed on the public internet. A 2006 post-mortem predicted exactly this. The cycle persists because governance makes protocols hard to adopt, so it can never arrive on time. - [RPA's Scar Tissue]() — By [Nora Kaplan]() · Published 2026-05-13 · Format: Two Returns Between 2018 and 2023, the RPA industry cataloged its failures in painful detail: UI fragility, credential sprawl, ownership ambiguity, maintenance costs that dwarfed licensing. Agentic AI is now encountering every one of these failure modes as though for the first time. Gartner projects over 40% of agentic AI projects will be canceled by 2027, citing reasons that read like RPA post-mortems. The scar tissue is there. Nobody packed it for the trip. - **API Design, Part 1: Before There Was REST** — [Source]() — Published 2026-05-13 Troy pinpoints CORBA's real failure: not ambition but the coordination cost of maintaining high-precision interfaces across organizations. MCP quietly inherits that same cost structure. - **From DX to AX: Why the Future of AI Depends on Agent Experience and Open Standards Like MCP** — [Source]() — Published 2026-05-13 The only contemporary source naming the CORBA/SOAP-to-MCP lineage explicitly, and treating it as validation rather than warning. That optimism is itself part of the pattern. - **Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027** — [Source]() — Published 2026-05-13 Puts a number on agent washing: roughly 130 genuine vendors out of thousands. The RPA rebranding cycle, quantified at enterprise scale. - **Was 2025 Really the Year of the AI Agent?** — [Source]() — Published 2026-05-13 Practitioners looking back diagnose the core gap: agentic concepts layered onto infrastructure never designed for autonomy. The RPA echo, named plainly and without nostalgia. - **What's BPEL? Business Process Execution Language Explained** — [Source]() — Published 2026-05-13 BPEL4People specified human-in-the-loop workflow orchestration in 2007. Today's agentic frameworks are rebuilding it from scratch, largely without knowing it existed. - **MCP for Skeptics: Why the Model Context Protocol Is Worth It (Even If It Doesn't Seem Like It)** — [Source]() — Published 2026-05-13 A DevOps veteran recognizes CORBA in MCP, works through the déjà vu, and decides to believe anyway. The rediscovery pattern narrated in first person. ### [Vision]() Where human-AI collaboration is heading - [The Architecture of Foregone Conclusions]() — By [Rina Takahashi]() · Published 2026-05-13 · Format: The Anchor Argument In 1994, a Netscape engineer wrote a four-page proposal for a small piece of browser data. Cookies shipped in the next release. Google spent six years trying to replace them and failed. The quick fix outlasted the replacement project because by the time anyone questioned it, everything else depended on it. That same mechanism is now operating in agent infrastructure and identity, except the dependency graph that took cookies three years to form is building in months. Most governance conversations assume there's time to get the architecture right. The defaults accumulating in your stack this quarter suggest otherwise. - [The Pilot That Worked]() — By [Nora Kaplan]() · Published 2026-05-13 · Format: Two Defaults Hiding in Plain Sight Failed AI pilots generate postmortems. Successful ones generate something quieter: permanent architecture. When a pilot encodes the current workflow and hits its targets, the reporting lines it assumed become the reporting lines. The decision rights it left untouched stay untouched. The window for redesign closes before anyone notices it was open. - [What the Junior Analyst Knew]() — By [Nora Kaplan]() · Published 2026-05-13 · Format: Two Defaults Hiding in Plain Sight When organizations automate entry-level work, the efficiency gains are visible. What's been removed takes longer to miss. Junior analysts processing exceptions and escalating edge cases did measurable, standardized work. They also served as the mechanism through which institutions discovered what their formal processes missed. Nobody is building a replacement for that function, because most organizations didn't realize it existed. - [Writing Down the Weather and Calling It a Forecast]() — Published 2026-05-13 *The following is a hypothetical interview. Nkem Okafor-Whitfield is a fictional character, though the events, organizations, and technical details she references are real. We invented her so we could have a conversation that no actual governance board member would be candid enough to give on the record.* - [The EU AI Act Delay Reveals a Familiar Pattern: Governance Arriving After the Architecture Is Already Set]() — Published 2026-05-13 On May 7, the EU agreed to push the AI Act's high-risk obligations from August 2026 to December 2027. Honest reasoning: harmonized standards weren't ready, conformity bodies weren't appointed. Companies were being asked to comply against benchmarks that didn't exist yet. - **The AI Agent Identity Problem: Why Governance Is the Missing Layer in Enterprise AI** — [Source]() — Published 2026-05-13 The clearest case yet that capability stopped being the constraint. Trust is, and the defaults are already hardening. - **MCP's 2026 Roadmap: From Agent Integration Standard to Production Connectivity Layer** — [Source]() — Published 2026-05-13 Where MCP is heading technically, and why the security boundaries those APIs assumed are quietly ceasing to hold. - **Inside the AI Index: 12 Takeaways from the 2026 Report** — [Source]() — Published 2026-05-13 Gold-medal math, coin-flip clock reading. The jagged frontier enterprises actually encounter looks nothing like the benchmarks. - **AI Will Reshape More Jobs Than It Replaces** — [Source]() — Published 2026-05-13 Augmentation diffuses faster than substitution. The unexamined cost: hollowing out the pipeline that builds senior judgment. - **AI Is Simultaneously Aiding and Replacing Workers, Wage Data Suggest** — [Source]() — Published 2026-05-13 Entry-level displacement is already visible in wages. The apprenticeship model that builds tacit knowledge has no designed successor. ## More - [Publication guide]() - [Issue archive]()