--- title: "Issue 24" canonical: "https://current.tinyfish.ai/issue/24" publication: "TinyFish" publisher: "TinyFish" language: "en-US" guide: "https://current.tinyfish.ai/llms.txt" permalink: "https://current.tinyfish.ai/issue/24" issue: 24 description: "Agents are racing to become invisible infrastructure while the cost curves, trust models, and governance underneath them are still half-built." authors: ["Sable Whitford","Nora Kaplan","Rina Takahashi"] published: 2026-04-13 updated: 2026-11-05 articles: 89 --- # 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 24 Agents are racing to become invisible infrastructure while the cost curves, trust models, and governance underneath them are still half-built. ### [Daily Brief]() Your daily digest of web developments through the lens of enterprise agent deployment - [The Accounts You Locked Were the Ones That Opened: Meet \*sshole]() — By [Sable Whitford]() · Published 2026-11-05 OpenSSH and Alpine Linux **shipped fixes today** for `*sshole`, a decade-old bug that let any password between 257 and 1024 bytes log in on Linux boxes running musl libc without PAM. Locked accounts get a `*` in `/etc/shadow`. musl's `crypt()` also returns `*` when it chokes on a password. OpenSSH compared the two, saw a match, and let you in. The bug was reported Wednesday and patched **this Thursday** with no embargo, because maintainer Damien Miller says AI agents now rediscover bugs within hours. Upgrading alone won't protect you. You still have to restart sshd. - **OpenAI buys the Git plumbing under Codex** — [Source]() — Published 2026-11-05 OpenAI is acquiring Pierre Computer Company, maker of Git-based distributed storage "with API ergonomics designed for machines." Pierre's open-source Trees and Diffs libraries already run file browsing and change review in Codex, and they stay open source. Terms undisclosed. It follows Pando, OpenAI's in-house source-control backend, which reportedly cut deploy latency about 80%. Coverage is calling this a GitHub challenger. OpenAI hasn't said so. - **Meta buys Conductor, users get eviction notice** — [Source]() — Published 2026-11-05 Same day, different buyer: Meta picked up Conductor, the app for running several coding agents at once. Sign-ups and downloads are frozen and the cloud side is winding down. Cloud data deletion starts November 20, so export what you care about now. CEO Charlie Holtz promises something new early next year. Meta hasn't said whether existing integrations survive. - **GitHub picks a terrible day to wobble** — [Source]() — Published 2026-11-05 From 09:49 UTC Wednesday, Git operations, Issues and pull requests degraded, then Actions and the API joined in. Mitigated around 11:30, resolved by 11:50, with a few more Actions hiccups later on. All of it landed the same day OpenAI bought a Git storage company built for agents. Rough optics. Rougher pager. - **Revolut swaps Claude Code for self-hosted GLM** — [Source]() — Published 2026-11-05 About 2,000 Revolut engineers now default to a self-hosted Z.ai GLM model: open weights, Revolut's own US servers, roughly 56 Nvidia B200s, about $0.30 per million tokens, around a tenth of the old bill. Claude Code and Codex stay for the hardest work, and Gemini gets images. Revolut reports 70% better lead time and 60% better coding time. AI ops head David Abitbol says the bottleneck "is not necessarily coding." Cheap and good enough is winning fans, and people are loudly arguing over the stigma around Chinese open weights. - **Rent-a-proxy, now exiting from the White House** — [Source]() — Published 2026-11-05 Layer3 Intel saw residential proxy traffic exit from AS2650, the Executive Office of the President's network, 236 times in a 30-day window ending Tuesday. That was 6 addresses, on 26 of the 30 days, across 28 proxy services. Layer3 wasn't hunting for it. The addresses just turned up in rotation. The pattern looks like someone's everyday device rather than guest Wi-Fi. No breach is shown and the device is unidentified. Sentient, via Cybernews, suspects Rave.io, an app bundling 12 proxy SDKs. The White House hasn't commented. - **CISA flags an actively exploited WordPress flaw** — [Source]() — Published 2026-11-05 CISA added a WordPress Core vulnerability to its Known Exploited Vulnerabilities catalog, so attackers are already using it. Details are thin. Do the boring thing: update WordPress, check that auto-updates actually ran, and poke whoever supposedly "owns" the marketing site. - **Bot swarms are knocking university libraries offline** — [Source]() — Published 2026-11-05 Data-hungry bot swarms are hitting university library websites hard enough to crash them. Sites built for students and researchers are soaking up crawler traffic that will never borrow a single book. Toss it on the scraper-fatigue pile, which keeps getting taller. - **Does Gemini train on paid chats?** — [Source]() — Published 2026-11-05 A complaint making the rounds says Gemini trains on chat history even for paying subscribers. It's an unconfirmed claim that people are arguing about loudly. Either way, it's a decent excuse to go poke at your own settings. - **Nvidia's AI laptop: unplug it to go faster** — [Source]() — Published 2026-11-05 On RTX Spark laptops running GPT-OSS-120B, Windows and WSL hit about 65 tokens per second. Native Linux managed 13.9, roughly 79% slower. Pull the power cord and Linux climbs back to about 61–67. MindStudio blames a Windows-only firmware component with an encrypted power table Linux can't decrypt. That's one study, and Nvidia hasn't confirmed it. Battery mode as a performance feature. Wild. - **AMD expects to outsell Intel, a first** — [Source]() — Published 2026-11-05 Per Tom's Hardware, AMD expects to outsell Intel in Q4 for the first time ever. That's a forecast, so wait for the final count. Still, anyone who remembers AMD as the perennial underdog gets to enjoy that sentence. - **Data centers get voted down, town after town** — [Source]() — Published 2026-11-05 Data centers had a bad night in Tuesday's midterms. Bloomberg headlines say only one town out of dozens voted in favor of them, and AI Weekly counts bans or restrictions passing in at least 14 Ohio municipalities. Wall Street read the night its own way: Bloomberg reports Big Tech rallied as the results reshaped the AI spending outlook. - **Baidu ships ERNIE 5.5 Flash Base** — [Source]() — Published 2026-11-05 Baidu released ERNIE 5.5 Flash Base. For now there's a release page and not much else, so no specs to argue about yet. It lands on a day when Chinese models were already the talk of engineering teams. - **OccupySaaS: laid-off engineers clone everything** — [Source]() — Published 2026-11-05 Under the OccupySaaS banner, laid-off engineers are using AI to build open-source alternatives to, well, everything. Auth0 and Salesforce clones are in the works. Free time plus a code generator turns out to be a hell of a combination. - **AI is calling in 40 years of tech debt** — [Source]() — Published 2026-11-05 An IEEE piece argues AI is forcing organizations to finally pay off four decades of technical debt. The headline alone had people glancing nervously at the oldest box in their stack, the one nobody is allowed to reboot. "We'll fix it later" apparently had a due date. - **A detective game where every suspect lies** — [Source]() — Published 2026-11-05 Uncracked City casts LLMs as your suspects, and they lie to you, on purpose this time. You interrogate them, cross-check their stories and catch the contradictions. A model making things up is the whole point here, and it's suspiciously good practice for reading AI output at work. - **Stringman: a room-sized robot gripper, $800** — [Source]() — Published 2026-11-05 Stringman is an open-source robot gripper that works across an entire room, for about $800. That's roughly used-bike money, which puts room-scale robotics within reach of anyone with a free weekend and a garage. Expect weekend projects. - **MCP Explained, for when the acronym blurs** — [Source]() — Published 2026-11-05 MCP Explained is a standalone site that walks through Model Context Protocol, the standard for hooking AI models up to tools and data. Bookmark it for the next meeting where someone says "just wire it up with MCP" and the whole room nods like they know. - **Download your favourite YouTube videos. Seriously.** — [Source]() — Published 2026-11-05 A warm little essay makes the case for keeping local copies of the YouTube videos you love, because channels vanish and links rot. Ops people will agree on sight: if it matters and there's no backup, you're only borrowing it. - [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. - [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. - [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. - [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 - [The Goal and the Line Item]() — By [Nora Kaplan]() · Published 2026-04-08 · Format: Economic Actors A purchasing card knows what you can't buy. Wrong merchant category, declined at the register. Over the spending limit, declined before a human ever sees it. Thirty years of delegated spending authority, encoded into infrastructure so reliable it's invisible. Now Visa and Mastercard are extending that same infrastructure to AI agents. - **GPT-5.4 Brings Computer Use to One Model** — [Source]() — Published 2026-04-08 OpenAI's first general-purpose model with native computer-use capabilities scores 75% on OSWorld-Verified, clearing the 72.4% human baseline. Professional-matching hits 83% across 44 occupations on GDPval, with a 1M token context window. Capabilities that previously required separate specialized models now ship under a single model ID. Why it matters: the consolidation lowers the integration cost for anyone building agents that need to operate software. - **EY Goes Live With Agentic Audit Globally** — [Source]() — Published 2026-04-08 EY has embedded a multi-agent framework into its Canvas audit platform across 150+ countries, reaching 130,000 assurance professionals and 160,000 engagements. This is a production rollout, not a pilot. Full end-to-end audit support is targeted by 2028, built on Microsoft technology. Why it matters: the largest enterprise-scale agentic deployment to date, in a domain where "correct" is deeply contextual. - **Okta and Microsoft Ship Agent Identity Platforms** — [Source]() — Published 2026-04-08 Okta for AI Agents enters early access with GA on April 30, offering agent discovery, access controls, and a kill switch. Microsoft Agent 365 hits GA May 1 as a control plane for observing and securing agents at scale. The timing reflects the gap: 88% of organizations report agent security incidents, while only 22% have agent-specific identity governance in place. Why it matters: agents are getting treated as first-class identity principals by the two companies that define enterprise identity. - **NVIDIA Open-Sources Enterprise Agent Toolkit at GTC** — [Source]() — Published 2026-04-08 NVIDIA's Agent Toolkit pairs the Nemotron 3 open model family with AI-Q, a hybrid architecture that routes queries between frontier and open models to cut costs by 50%+. Nemotron 3 Nano delivers 4x throughput over its predecessor. Adobe, Amdocs, Atlassian, and over a dozen other partners are building on the stack, with LangChain integration underway. Why it matters: an open-source cost optimization layer for agent inference, backed by a hardware vendor with distribution reach. - **MCP Hits 10,000 Servers and 97M Downloads** — [Source]() — Published 2026-04-08 The Model Context Protocol ecosystem now counts over 10,000 active public servers and 97 million monthly SDK downloads. The 2026 roadmap prioritizes Streamable HTTP and enterprise gaps: audit trails, SSO-integrated auth, and gateway behavior. Forrester predicts 30% of enterprise app vendors will launch MCP servers this year. Why it matters: adoption is running well ahead of the enterprise features the protocol still lacks. - **Kyndryl Launches Blueprints for Agentic Service Management** — [Source]() — Published 2026-04-08 Kyndryl's new offering packages maturity models, structured assessments, and implementation blueprints for transitioning to autonomous workflows. The company already runs roughly 200 million automations per month through 8,000+ certified playbooks. The context is sobering: nearly half of organizations investing heavily in AI report struggling to achieve meaningful returns. Why it matters: a managed-services player is betting that the bottleneck for agentic adoption is operational design, not model capability. - **GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks** — [Source]() — Published 2026-04-08 OpenAI's benchmark pits model outputs against human deliverables across 1,320 tasks in 44 occupations, judged blind by professionals averaging 14 years of experience. It measures one-shot output quality on well-specified tasks, not run-to-run consistency or performance when instructions are vague. - **GPT-5.4 Professional-Matching Results on GDPval** — [Source]() — Published 2026-04-08 GPT-5.4 matches or exceeds professionals in 83% of GDPval comparisons, a 12-point jump over GPT-5.2. Spreadsheet modeling climbed from 68.4% to 87.3%. OpenAI itself acknowledges most jobs involve more than well-specified one-shot tasks can capture. - **Visa/Morning Consult: Consumer and Business Readiness for AI-Mediated Commerce** — [Source]() — Published 2026-04-08 A survey of 2,000 U.S. adults and 512 business decision-makers finds nearly 40% of Americans have purchased something they wouldn't have considered without AI. 53% of businesses say they'd let agents negotiate prices or terms directly with other agents. - **Visa and Partners Complete Secure AI-Agent Transactions** — [Source]() — Published 2026-04-08 Visa completed hundreds of secure agent-initiated transactions in late 2025 and launched the Trusted Agent Protocol with over 10 partners. The open framework helps merchants distinguish legitimate AI agents from bots. Visa projects millions of agent-purchasing consumers by holiday 2026. - [When Both Sides of the Table Are Bots, Whose Deal Is It?]() — Published 2026-04-08 Visa's recent B2AI report found that 53% of U.S. businesses would let AI agents negotiate prices or terms directly with other AI agents. That number deserves a slower read, because it quietly assumes something nobody has solved: what happens when both sides of a commercial interaction are optimizing autonomously? - **Visa and Mastercard Are Building the Rails for Agentic Commerce** — [Source]() — Published 2026-04-08 Primary reporting on how card networks are competing to define trust and liability frameworks for agent-initiated payments. Worth reading for the details on where the liability question actually lands when no human confirms a transaction. - **The Identity Problem at the Heart of Agentic AI Security** — [Source]() — Published 2026-04-08 IBM's RSA 2026 recap surfaces a telling gap: vendors showed plenty of point solutions for securing agents, but nobody demonstrated end-to-end orchestration. The missing middle is the story. - **AI Agents Are Forcing Identity Security Into Real Time** — [Source]() — Published 2026-04-08 Ping Identity's CEO names the structural shift clearly: the control point has to move from authentication at entry to authorization at every action. If you read one piece on why traditional IAM breaks down with agents, this is the one. - **Enterprise Agentic AI Landscape 2026: Trust, Flexibility, and Vendor Lock-in** — [Source]() — Published 2026-04-08 Kai Waehner offers a two-axis framework for evaluating AI vendors once agentic autonomy and enterprise lock-in are factored in. Useful for anyone trying to think about vendor selection as a governance decision, not just a capability one. - **Every Agent Needs an Identity: Okta's Blueprint for the Secure Agentic Enterprise** — [Source]() — Published 2026-04-08 The clearest public articulation yet of centralized agent governance from a major identity vendor. The 88% of organizations deploying agents versus 22% with adequate security controls is a number worth sitting with. - **State of AI Trust in 2026: Shifting to the Agentic Era** — [Source]() — Published 2026-04-08 McKinsey survey data showing organizations can deploy agents at scale but cannot yet govern them meaningfully. Confirms the pattern we keep seeing: the deployment bottleneck has moved from capability to organizational readiness, and nobody's staffed for it. ### [Foundations]() Conceptual clarity earned from building at scale - [The Classification Decision That Makes Agent Cost Models Fragile]() — By [Nora Kaplan]() · Published 2026-04-08 · Format: The Practical Piece Somewhere in a config file, a practitioner is deciding whether "extract pricing data from competitor pages" counts as reasoning or execution. A frontier model or a cheap model gets the job, and the cost architecture of the entire system rides on this single classification. No framework draws a formal boundary. - [Where Agentic AI's 40% Failure Rate Actually Lives]() — By [Rina Takahashi]() · Published 2026-04-08 · Format: The Argument AI captured 80% of global venture funding last quarter. Quarter-trillion dollars, one category. Meanwhile, Gartner projects that four in ten agentic AI projects won't survive to 2027. Both numbers are credible. Both are rational. And the gap between them is where actual deployments go quiet. - **Multi-Agent Patterns That Survive Production (and Why Most Don't)** — [Source]() — Published 2026-04-08 Real failure data from multi-agent deployments: cascading hallucinations, latency traps, and which patterns actually hold up. - **Vendor Lock-in as Economic Liability: The Trust × Flexibility Framework** — [Source]() — Published 2026-04-08 Treats agentic AI vendor selection as a long-term cost structure commitment, not a technical preference you revisit later. - **What Agent Orchestration Frameworks Actually Commit You To** — [Source]() — Published 2026-04-08 Framework choice locks in observability, state management, and debugging costs that compound well past the initial build. - **Agent Cost Optimization as Architecture, Not Afterthought** — [Source]() — Published 2026-04-08 Names a 90% cost reduction from Plan-and-Execute patterns and argues model tiering belongs in design, not ops. - **The $4.63M Price Tag on Unsecured Agents** — [Source]() — Published 2026-04-08 Puts dollar figures on the downside: breach costs, red-team timelines, and blast-radius math for agents with real permissions. - **The Model You Pick Sets the Ceiling** — [Source]() — Published 2026-04-08 Inference accounts for roughly 85% of enterprise AI spend. The price spread is staggering: Claude Opus 4.5 runs $25 per million output tokens; GPT-5 nano costs $0.05 per million input. Same model, different provider? Expect 10x variance. Choosing where to run is a budget decision disguised as a technical one. - **Falling Token Prices Hide a Rising Total Bill** — [Source]() — Published 2026-04-08 Gartner puts the agentic multiplier at 5–30x more tokens per task compared to standard chat. Complex multi-agent workflows regularly blow past a million tokens on a single job. Per-token costs keep dropping. Volume is rising faster. The bill grows anyway. - **Context Windows Are Quietly Burning Your Tokens** — [Source]() — Published 2026-04-08 Every token in the context window gets billed on every LLM call. One agent resending a 25,000-token document across five reasoning steps burns 125,000 input tokens on a single task. MCP tool metadata alone can eat 40–50% of available context. RAG and prompt caching cut this waste by 60–90%. - **Coordination Tax Rivals the Agents' Own Costs** — [Source]() — Published 2026-04-08 When agents hand off to other agents, the coordination layer becomes its own cost surface. A single edge case can trigger retry chains costing 50x the normal path. Framework choice compounds this: in benchmarks, some orchestration frameworks consumed nearly double the tokens of others running identical tasks. - **Below the API Layer, a Whole Other Bill** — [Source]() — Published 2026-04-08 Kubernetes provisioning, vector databases, memory backends, retry queues, load balancers. All separate line items, all easy to forget. Self-hosting above 100K requests per day saves 60–80% versus API pricing. Below that threshold, fixed GPU costs make self-hosting the more expensive option. - **Observability Bills Punish You for Actually Looking** — [Source]() — Published 2026-04-08 Agent observability runs $25–$500/month at startup scale, $2K–$10K+ for enterprise volumes. The billing math gets adversarial fast: a 10-span trace with one evaluation score costs 12 billable units. Some teams have resorted to sampling 0.1% of traffic just to keep observability affordable. - **Let a Frontier Model Think, Not Do** — [Source]() — Published 2026-04-08 Plan-and-Execute splits the expensive reasoning from the cheap labor. A frontier model architects the strategy; distilled or smaller models carry out each step. Production deployments report 80% cost cuts, stretching toward 90% with full distillation. The catch: distilled executors fail on rare edge cases outside their training data. Precisely the cases where failure costs the most. - **Scope Isolation Starves Each Agent of Bloat** — [Source]() — Published 2026-04-08 Monolithic agent architectures stuff every subagent with the full conversation history. Scope isolation walls off each agent's context to only what it needs. LangChain's multi-agent documentation shows a 67% token reduction per request through this boundary enforcement. The cost: each handoff between isolated agents adds 500ms–1.5s of summarization latency and risks dropping context that mattered. - **Match the Model Tier to the Task** — [Source]() — Published 2026-04-08 A frontier reasoning call can cost 190x more than a small model handling equivalent work. Routing each agent step by complexity, sending simple extraction to cheap models and reserving heavy inference for hard decisions, yields 30–70% savings across practitioner reports. The fragility: you now depend on a routing classifier that adds its own latency, and a single provider outage can break a multi-model pipeline in ways a single-model system never would. - **Stop Resending the Same Tokens Every Call** — [Source]() — Published 2026-04-08 Agents hammer the same system prompts, tool schemas, and few-shot examples on every invocation. Provider-level prompt caching (Anthropic discounts cached tokens by 90%) eliminates that redundancy. Practitioners report 45–80% cost reductions on cache-eligible workloads. The caveat worth remembering: when tool schemas update, stale cache hits silently degrade output quality unless you manage invalidation explicitly. - **Cache the Plan, Not Just the Prompt** — [Source]() — Published 2026-04-08 The strongest evidence in this set comes from a peer-reviewed preprint: caching and adapting prior execution plans across five agent workloads cut costs by 50% and latency by 27%, retaining 96.6% accuracy. This goes beyond prompt-prefix caching to reuse the planner's actual reasoning. Requires a Plan-Act architecture to work, and effectiveness drops sharply on genuinely novel tasks where no prior plan applies. ### [Practitioner's Corner]() Lessons from the field—what we see building at scale - [Drop-Downs That Aren't Really Drop-Downs]() — By [Rina Takahashi]() · Published 2026-04-08 · Format: The Self-Orienting Agent Delaware's government portal goes offline at night. A pharmacy invoice site hides its download link behind a field that looks like a drop-down but behaves like a text box. A county records office redesigns its layout without warning. None of these have APIs. Humans navigate the mess without thinking. Scripts shatter against it. Somewhere between those two facts sits a bottleneck that more model capability alone won't close. One engineer spent years watching it consume his ML pipelines before deciding to build directly against it. - [Why Agent Failures Keep Getting Blamed on the Model]() — By [Rina Takahashi]() · Published 2026-04-08 · Format: The Context Gap In surgical teams, individual surgeons hold 83% of the information they need. Anesthesiologists hold 87%. Competent people, by any measure. Yet a study of perioperative handoffs found only 27% of critical knowledge was shared across the full team. The failure lived in the gaps between skilled people, not in the skill itself. - [The 340-Page Document Nobody Budgets For]() — Published 2026-04-08 *Dolores "Dee" Bridgewater spent fourteen years in hotel operations before moving to enterprise IT at a large Midwestern insurance company. She is now, by her own description, "the person who onboards the robots." Her actual title is Senior Process Analyst. Her actual job, documenting the unwritten rules of claims processing so that AI agents stop making expensive mistakes, has no title, no budget line, and no entry on the org chart. We spoke with her over video call from a conference room she'd booked under the name "Agent Training — DO NOT CANCEL," a trick she learned after facilities deleted her recurring hold three times. Dee Bridgewater is not a real person, but the role she describes is one that a growing number of enterprises have accidentally invented.* - [Compile to Code: The Pattern That Turns Agent Reasoning Into Reusable Institutional Memory]() — Published 2026-04-08 A workflow costing $0.50 per LLM-in-the-loop execution becomes $50,000 a month at scale. That math is why a quiet architectural pattern keeps surfacing: let the agent reason through a workflow once, compile that reasoning into deterministic Playwright code, then replay the script on every subsequent run. Think expensive thoughts exactly once. - **What Amazon Learned When It Tried to Evaluate Agents at Scale** — [Source]() — Published 2026-04-08 Primary-source account of production evaluation at Amazon. - **The Friction Isn't in the Model — It's in Everything Around It** — [Source]() — Published 2026-04-08 Practitioner map of where production scaling actually breaks. - **The Compounding Probability Problem Nobody Wants to Do the Math On** — [Source]() — Published 2026-04-08 Simple arithmetic that reframes the entire adoption gap. - **From Building Agents to Managing the Sprawl They Created** — [Source]() — Published 2026-04-08 The problem has shifted from construction to lifecycle governance. - **Half of Enterprise Agents Can't Talk to Each Other** — [Source]() — Published 2026-04-08 Salesforce survey quantifies agent isolation with uncomfortable precision. ### [Echoes]() Past infrastructure decisions echoing in today's production realities - [Why the Internet Still Trusts Strangers]() — By [Rina Takahashi]() · Published 2026-04-08 · Format: The Napkin Protocol In June 2024, a Brazilian ISP announced Cloudflare's IP address as its own, and the internet believed it. DNS resolution dropped across 300 networks in 70 countries. The routes were cryptographically signed. A major carrier accepted the bogus announcement anyway. - [The Fix That Needs What the Problem Needed]() — Published 2026-04-08 BGP's vulnerability was never really technical. It was a coordination problem dressed in protocol clothing. The system assumed honest peers. No verification. Just trust. - [A Routing Researcher on Why Being Right Was Never the Hard Part]() — Published 2026-04-08 *Dr. Renzo Pfister is a composite character, a routing-protocol Frankenstein stitched together from the documented careers of researchers who've spent decades trying to fix the internet's most fragile layer. Any resemblance to real professors at Swiss technical universities who founded spinoff companies and received prestigious security awards is, let's say, structurally inevitable.* - [The HTTPS Exception]() — By [Rina Takahashi]() · Published 2026-04-08 · Format: Chokepoints and Consequences The entire web upgraded its security posture in about five years. For foundational internet infrastructure, that kind of change almost never happens. The conditions that made HTTPS possible turn out to be so specific they raise an uncomfortable question: a free certificate authority, a dominant browser willing to punish holdouts, and a search engine rewarding compliance, all arriving within a narrow window. What about the parts of the internet where no single actor holds that kind of leverage? - [Trust by Design]() — By [Rina Takahashi]() · Published 2026-04-08 · Format: Chokepoints and Consequences In April 2018, a small hosting company in Ohio told the internet's core routers it was Amazon Web Services. The routers believed it. They had no mechanism to do otherwise. The protocol that directs traffic between networks was designed for a world of cooperating institutions that trusted each other. That world disappeared decades ago. The protocol kept running. Patches keep accumulating on top. And unlike the web, no chokepoint exists to force a real upgrade. - **Switzerland Built an Alternative to BGP. Nobody Noticed.** — [Source]() — Published 2026-04-08 SCION carries 220 billion Swiss francs daily, proving the gap between working replacement and actual adoption. - **Is BGP Safe Yet?** — [Source]() — Published 2026-04-08 Only 6.5% of users protected by ROV. Signing routes and enforcing validation are separate coordination problems. - **Switzerland Wants to Replace the Internet's Most Fragile Protocol** — [Source]() — Published 2026-04-08 Dollar figures on BGP hijacking losses make visible the cost that never generates enough political pressure to replace anything. - **Reviewing a Historical Internet Vulnerability: Why Isn't BGP More Secure?** — [Source]() — Published 2026-04-08 Thirty-five years of security proposals failed because the community never agreed on what the threat model actually was. - **SCION Internet Architecture: Deployment and Standardization Status** — [Source]() — Published 2026-04-08 NL-ix connectivity marks SCION's first step beyond Switzerland, testing whether small coordination surfaces can expand without dissolving. ### [Vision]() Where human-AI collaboration is heading - [When the Thing That Disappears Is the Thing That Decides]() — By [Nora Kaplan]() · Published 2026-04-08 · Format: The Trajectory Right now, AI agents have names. They get launched at keynotes with brand identities, marketed as teammates you can @mention in a project tracker. Underneath the fanfare, interoperability protocols are moving into standards-body governance, on track to ship by year's end. The agents are being made ready to disappear. - [The 600 Agents Nobody Hired]() — By [Nora Kaplan]() · Published 2026-04-08 · Format: Two Tensions The most valuable AI agents in enterprise today are background processes running on event triggers across supply chains, compliance pipelines, and procurement systems, operating continuously without human prompts. A security vendor recently mentioned that a Fortune 500 client found hundreds of these agents running inside its environment that nobody knew existed. The audience didn't flinch. That indifference is the governance problem. - [How Do You Sell Something That Succeeds by Vanishing?]() — By [Nora Kaplan]() · Published 2026-04-08 · Format: Two Tensions The most valuable AI agents succeed by disappearing, which creates a governance problem explored in our companion piece. It also creates a commercial one. Atlassian's design docs for Rovo instruct teams to make AI visible so users trust it. The product aspiration is that nobody thinks about Rovo at all. Somewhere between those two instructions lives an unanswered question about how you price, brand, and renew software whose success metric is user indifference. - [The Designer With Two Figma Files and No Good Answer]() — Published 2026-04-08 *Gil Fontenot is not a real person. He is, however, a composite of every enterprise product designer who has ever been told to "make it pop" in a demo and "make it disappear" in production, in the same product review meeting. We spoke over video. He had opinions.* - [The Shrinking Arc: How Fast Technologies Become Infrastructure]() — Published 2026-04-08 Electricity took roughly 40 years to go from novelty to something city dwellers forgot they lived without. The web compressed that to about 20. Mobile did it in eight. - **The Computer for the 21st Century** — [Source]() — Published 2026-04-08 Mark Weiser's 1991 paper argued that the most profound technologies are the ones that disappear. Thirty-five years later, agent infrastructure is finally testing whether he was right. - **The Ambient Revolution: Why Calm Technology Matters More in the Age of AI** — [Source]() — Published 2026-04-08 Picks up Weiser's thread and pulls it into the current AI moment, making a sharp case that good infrastructure behaves like a roommate, not a houseguest demanding constant attention. - **Enabling Agent-First Process Redesign** — [Source]() — Published 2026-04-08 Branded content via Deloitte, but the "humans as governors, agents as operators" framing is useful. The real signal here is that process redesign around agents has become consultancy-grade advice, which tells you something about where the maturity curve sits. - **AI Agents as Colleagues: The Workplace Design Nobody's Planning For** — [Source]() — Published 2026-04-08 Fresh Gensler survey data on how AI power users actually work differently, plus Gartner projections on agent adoption. The Atlassian layoffs add uncomfortable context to the visibility question: agents are becoming colleagues faster than anyone is redesigning the workplace to accommodate them. - **Deciphering the Alphabet Soup of Agentic AI Protocols** — [Source]() — Published 2026-04-08 The most readable map of MCP, A2A, and the growing tangle of agent protocols. Protocol consolidation is how a technology announces it's becoming infrastructure, and this piece captures the moment mid-announcement. - **Enterprise Agentic AI Landscape 2026: Trust, Flexibility, and Vendor Lock-In** — [Source]() — Published 2026-04-08 Practitioner-level analysis of how vendor lock-in now compounds at every layer of the agent stack. Worth reading for the structural argument: the default choices being made today will quietly constrain what's possible for years. ## More - [Publication guide]() - [Issue archive]()