Click any tag below to further narrow down your results
Links
The author imagines AI making software and labor cheap, then asks who captures the gains. His answer: builders and incumbents get some, landlords and rent-seekers get most, and everyone else gets little unless society adopts something like a land-value tax for other scarce resources.
- Four scenarios are walked through: status quo, a messy transition where new jobs offset displacement, reinstatement failure leading to plutocracy, and distributed ownership, with the first three analyzed as paths toward either instability or a rentier economy.
- Citing Henry George, Enrico Moretti, and Richard Hornbeck, the author argues that when housing supply is tight, landlords capture nearly all productivity gains, so a guaranteed income like UBI at $50k would mostly flow into rent (around $25k).
- He rejects revolution as a reliable route to a better outcome, arguing it more likely produces destruction or a new autocracy, and proposes that a scarce-resource equivalent of a land-value tax is needed to keep gains broadly shared.
Annie Vella argues that AI coding agents have changed software engineering from writing code to supervising it, and that human factors research on supervisory control from the 1970s and 1990s explains what that new job demands. She uses those decades-old frameworks to map today's levels of AI autonomy onto engineers' roles.
- Her Masters research found engineers shifting from creating code to verifying it, through testing and reviewing, with a new activity she calls supervisory engineering work: directing AI, evaluating its output, and correcting it.
- Supervisory control research from Sheridan and Verplank maps closely onto the role: plan and command, monitor, intervene, and calibrate trust.
- Agents hand over whole tasks ("trading") rather than working alongside the engineer ("sharing"), and the level of autonomy is often left to individuals, defaulting to whatever the tool allows.
Mitchell Zemil, a freelance animator and adjunct professor, writes in response to 3blue1brown's AI-in-math discussion, arguing that artists and mathematicians both value the human labor, intentionality, and personal "DNA" behind their work. He contends that generative AI turns making into tastemaking, and that mathematics is more than an answer-producing machine.
- Art is valued for the artist's labor, skill development, and intentional process, so generative AI's ability to produce polished output quickly threatens that value and raises ethical issues around training data.
- Past technologies like cameras and digital tools democratized art without erasing the human practice behind it, but generative AI is different because it shifts the work toward selecting from outputs rather than making them.
- Mathematics is often treated as a purely objective pursuit that machines could finish off, but Zemil argues people still value living mathematicians who explain, interpret, and push the frontier, much as audiences still value dancers at the ballet.
1Password has launched a platform that lets personal AI agents log into websites by auto-filling credentials you approve, so the agent never sees the actual password. You get to see and approve each login, and you can disconnect it at any time.
- Agents currently stall at sign-in pages; the workaround of pasting passwords into prompts is what this is meant to replace.
- You approve each credential request in 1Password, and 1Password fills the login directly into the destination site, keeping the password hidden from the agent and its model.
- Integrations are announced with partners including Meta's Muse, Anthropic's Claude, OpenAI's ChatGPT, and xAI's Grok, with more coming. It's free for Individual and Family accounts on 1Password.com, but not on the .eu or .ca domains.
Jason Wei argues that radiologists will stay in demand despite AI, because cheaper scans will drive elastic demand, rare diseases will keep AI in a triage role for years, and healthcare's slow adoption and regulation will buy time. He uses Geoff Hinton's 2016 prediction that radiology training should stop as the foil.
- AI will handle common, easy-to-see conditions first but lag on rare diseases, since there is less data for them, so expect 3-5 years of triage use before full automation.
- Jevons paradox: if reading a scan becomes nearly free, radiologists might read only 20% of scans while five times as many arrive, and many developing countries with under one radiologist per million people would add demand.
- Cheap interpretation could push device makers toward cheaper, more frequent scanners, and medical regulation, legacy infrastructure, and data access limits will slow adoption.
The author argues that fears of mass AI unemployment rest on the "lump of labor fallacy"—the mistaken belief there's a fixed number of jobs. History shows that as technology automates old work, humans invent new desires and new jobs to fill, a pattern unlikely to break now despite AI's capabilities.
- Unemployment rates haven't risen over 80 years despite continuous automation since the Industrial Revolution, yet 71% of Americans expect AI to cause job losses in the next 20 years—a historical pattern reversal that needs stronger evidence than speculation.
- Humans have an infinite capacity to generate new wants once current ones are satisfied (the "hedonic treadmill"), which creates perpetual demand for new jobs; jobs like "professional screamer" and "sorority rush consultant" aren't signs of economic decadence but proof this mechanism still works.
- Current data contradicts AI displacement predictions: Indian IT companies' headcounts haven't fallen, radiologist salaries hit all-time highs, and over 329,000 bank tellers still work in the US despite ATMs and mobile banking supposedly making them obsolete.
This is a repository that provides a single collection of 202 agents, 184 skills, and 105 commands designed to work across seven different coding environments—Claude Code, OpenAI Codex CLI, Cursor, OpenCode, Antigravity CLI, GitHub Copilot, and Pi. Each tool uses a "harness" that loads and runs these components according to its own capabilities, with adapters translating the source Markdown format into whatever format each tool needs.
- The system includes 94 plugins (92 local, 2 external) covering Python, JavaScript, code review, testing, infrastructure, and security work, with installation methods varying by target tool.
- Components are maintained once in Markdown source format, then automatically adapted to each harness's requirements—model aliases, tool permissions, and command formats differ by platform.
- Installation ranges from simple marketplace commands (Claude Code, Cursor) to cloning the repo and running make commands (OpenCode, Antigravity, GitHub Copilot, Pi), with individual skills installable via gh skill or npx.
This is a sign-up page for Zhipu AI's GLM-Legal API, inviting law firms and legal teams to apply for access and discuss how the tool can integrate into their practice. The pitch covers using it for daily legal work, business operations, and client services.
- GLM-Legal is positioned as a tool for routine legal tasks and evolving legal challenges
- The offering targets three use cases: legal practice, business operations, and client services
- Access requires form submission and direct conversation with the Zhipu team about collaboration terms
Google Cloud is launching Gemini Agent, a unified AI assistant that handles knowledge work, content creation, coding, and task automation across enterprise systems. It works persistently in the cloud, maintains context across devices and apps, and can orchestrate multiple sub-agents to complete complex workflows.
- Gemini Agent operates as a single interface for work—you assign objectives rather than give instructions, and it completes tasks across Gmail, Drive, Docs, Sheets, and other Workspace apps while maintaining memory and context across devices
- The agent uses multi-model orchestration (Gemini models plus Claude and others) with smart routing and cost controls, automatically selecting the best model for each task rather than always using the largest one
- It connects to enterprise systems (Salesforce, ServiceNow, BigQuery, Git, Jira, etc.), learns from reusable skills and tools stored in company registries, and can spawn temporary sub-agents or persistent "coworker agents" with their own identities to coordinate multi-step workflows
- Early customers like Bradesco cut document review time from 1 hour to 5 minutes and achieved 10% financial efficiencies; BNP Paribas deployed it to 65,000 employees for streamlined workflows
Korn Ferry published "I Need a Job!", a career guide by CEO Gary Burnison that hit Amazon's top spot in three job-search categories. The book addresses a paradox in today's job market: roles people want are being automated away while companies struggle to fill less desirable positions.
- Job seekers now face a fundamentally different hiring process requiring online assessments, video responses, AI navigation, and ATS systems—not just resumes.
- Korn Ferry places roughly one professional every 3 minutes and draws on data from 115 million professionals to inform the book's advice.
- The job market creates a "vicious cycle" where more applicants chase fewer viable openings because automation eliminates desired roles while leaving less sought-after positions unfilled.
Google froze its open source bug bounty program starting October 1 because AI-generated submissions overwhelmed the system with invalid reports and hallucinations. The company won't reopen it until Q1 2027.
- Google engineers and open source maintainers were flooded with automated submissions that were mostly invalid or contained AI hallucinations, making the program unmanageable.
- The pause confirms cybersecurity experts' earlier warnings that AI-generated content poses a real threat to bug bounty programs by degrading their signal-to-noise ratio.
- Google is redirecting participants to its other bug bounty programs while it figures out how to handle the AI submission problem.
@DAcemogluMIT on X
Acemoglu argues that the AI boom faces a bind: for the industry to earn back its investment, it would need roughly $3.7 trillion in annual revenue by 2032, which would push capital's share of national income and inequality sharply higher. He doubts that revenue will materialize, so he sees a crash as the more likely alternative, and he worries that either outcome strains democracy.
- Van Nieuwerburgh estimates AI investment averaging 3.6% of GDP from 2025 to 2032, requiring about $3.7 trillion in yearly revenue by 2032 at a 10% return, up from roughly $200 billion today.
- Capital's share of US national income is already at a record high of about 47%, and a revenue surge would drive it higher, with capital income far more unequally distributed than labor income.
- Acemoglu expects AI diffusion to be slow, open-weight models to cap pricing, and AI to fall short of automating whole occupations, so revenues will likely undershoot and a costly crash becomes the more probable outcome.
AI handles code generation fine, so what companies actually need from engineers now is judgment—asking the right questions, understanding context, and owning work end-to-end. The "10x engineer" mystique dies when the hard technical work becomes commodity.
- AI commodified technical output, so the premium skill is now judgment: knowing what questions to ask, understanding user impact, catching edge cases, and recognizing when a machine answer is wrong for your specific situation.
- People waste time with bloated documents and vague asks instead of being direct—attention is scarce and you need to earn it by being clear about what matters and what you actually need.
- The soft skills (accuracy, follow-through, evidence over theory, reaching done) were never actually soft—they're what separates someone who ships from someone who doesn't, and they're the only thing left that's actually hard to automate.
Companies are struggling to break down static job descriptions into granular tasks that AI can handle, creating demand for tools that map which parts of roles can be automated. RoleGrid is a proposed platform that deconstructs jobs into micro-task bundles and connects them to current AI capabilities.
- Legacy HR platforms treat jobs as indivisible units, but AI executes discrete tasks—managers can't pinpoint which 20% of a role is automatable without precise task mapping.
- Workers are already unbundling their jobs informally (17x more likely to ask chatbots than bosses), forcing enterprises to build orchestration layers before shadow AI creates operational debt.
- RoleGrid's competitive advantage would come from building a proprietary taxonomy of micro-tasks tied to real-time agent capabilities, with deep integrations into HRIS platforms creating high switching costs.
This is a tool that extends ChatGPT to run on your actual desktop, read and edit files, execute tests, and split work across multiple AI workers that maintain context between tasks. It's a Chrome extension paired with a local app that gives ChatGPT real capabilities beyond conversation.
- ChatGPT can now access your filesystem, run terminal commands, keep processes open, and see results in real-time instead of just talking about what it would do
- You can spawn multiple workers to handle independent jobs in parallel, and they retain context so the next task picks up where the previous one left off
- You can interrupt and correct long-running tasks mid-execution, and save/resume entire sessions with full worker history using Compact & Resume
Cline, a VS Code extension used by 11 million developers, now has a standalone desktop app that lets you run multiple AI agents in parallel, choose from 300+ models, and automate recurring tasks. You can also import conversations from Claude Code or other agents and continue them with cheaper open-weight models.
- Run parallel agents simultaneously with scheduled cron job automations for recurring work like nightly repo checks or weekly documentation updates.
- Switch between 300+ models across 50+ providers, or use local models—you can even mix different models for planning versus execution.
- Import tasks and conversations from Claude Code, Codex, or other agents to continue work without starting over, useful when hitting subscription limits.
- The app extends beyond coding to research, document review, reporting, and other non-code work through plugins, MCP servers, and skills.
@cyrilXBT on X
A step-by-step guide from @cyrilXBT, shared as a quoted X article, that walks a beginner from zero to a working AI agent that completes tasks without constant supervision. The post itself is a short promotional wrapper; the substance lives in the linked article.
- The guide is framed as ordered, practical steps rather than theory, aimed at getting a functioning agent built.
- The stated goal is an agent that does real work without the user reviewing every response.
- The post is mostly a pointer to the full article, so the actual technical claims can't be assessed from this text alone.
Census Bureau surveys tracking business AI use from 2023 to 2026 show that even as AI adoption tripled, firms report almost no net employment changes. About 95% of businesses say AI hasn't affected their headcount either way.
- AI adoption jumped from 3.7% of firms in September 2023 to 10% by late 2025 (18% when counting any AI use), but employment impact remained flat: roughly 2-3% reported job increases, 2% reported decreases, and 95%+ reported no change across both survey periods.
- Among firms actually using AI, 44% say it supplements existing work, 10% say it replaced employee tasks, and 11% say it created new tasks—but most firms (64%) made no business changes to implement AI and only 1% hired new AI-skilled workers.
- Task substitution is growing within the small subset of firms where it's happening: the share reporting AI took over "a large number" of tasks jumped from 2.4% to 7.1%, but this group still represents only about 2% of all firms.
OpenClaw released a massive 2.0 update built by 933 contributors over nearly two months, completely overhauling installation, the browser app, and core infrastructure. The update lets people start with existing AI subscriptions and models, then grow their automation workflows from simple tasks to complex multiplayer collaborations.
- The release contains 16,000 pull requests (50% of all PRs ever merged) and touched every part of the platform because simplifying installation forced a complete foundation rebuild
- Installation now uses what's already on your computer—existing ChatGPT/Claude subscriptions, API keys, local models—cutting setup time so you can start having conversations immediately
- OpenClaw introduced shared cloud sessions that turn automation into a multiplayer experience, letting teams collaborate on tasks with full context intact
@eng_khairallah1 on X
The post describes a solo founder who runs seven Claude agents from a MacBook to scan Google Maps for small businesses with weak or missing websites, build mockup landing pages, and send personalized cold outreach. The author claims this brings in about $18,800 a month on roughly $480 in API costs, and frames it as a near-passive income system.
- The pipeline is fully automated: it finds prospects across three cities, builds a mockup landing page, renders a short video of it, and sends the cold message before the operator wakes up.
- The claimed numbers are 47 clients a month at $400 each, or $18,800 monthly revenue against $480 in API costs.
- The operator is only pulled in when a deal exceeds $3,000 or reply rates fall below 12%, and approvals and booked calls are handled from a phone.
Claude Code lets your independent sessions message each other to share findings, coordinate work, and report status without you manually copying information between terminals. Messages are delivered between tool calls and filtered through permission controls that prevent sessions from approving actions or changing configuration on each other's behalf. You can prompt Claude to send a message, or it'll detect when another session needs information and send it automatically.
- Claude Code sessions (v2.1.224+, macOS/Linux) can now message each other automatically or on request, using ListAgents/SendMessage tools, without interrupting running processes.
- Cross-machine messaging is one-way by default—remote or web sessions can only reply to existing threads, not initiate them, and replies from sessions without Remote Control arrive without a return address.
- Incoming messages are treated as system info, not user commands, so they can't approve permissions, change config, or run commands—permission prompts still trigger as normal.
- Inbound behavior is configurable per session via crossSessionInbound, with automatic naming (e.g. "myapp-3f") and /list-agents to see reachable sessions.
This page outlines e360Digital Pro’s offerings in custom software development, from ERP and web applications to AI-driven automation and cross-platform mobile apps. It also covers eCommerce store setups, digital marketing services, agile project management practices, FAQs, and contact details.
- e360Digital has over 10 years of custom software experience, with offices in St. Petersburg, FL and Houston, TX
- Their services span ERP, web/desktop apps, AI automation (chatbots, workflow tools), and cross-platform mobile apps (Flutter, React Native), plus Shopify/WooCommerce eCommerce builds and digital marketing (SEO, PPC, social, CRO)
- They use Asana for agile project management with real-time tracking and regular client check-ins, and transfer full IP ownership upon project completion
- They advertise a 24-hour response time on contact form inquiries
The author warns that “agentic design systems” often blur the line between using AI for tasks and handing off core judgment to autonomous loops with no human oversight. He argues design systems are governance tools requiring human-owned gates and accountability, and that removing those humans risks unchecked drift.
- "Agentic design systems" conflates using agents to do work with letting them make final judgment calls—two very different things.
- Agentic loops differ from vibe coding because they add hard gates (token checks, linting, accessibility tests, design-parity reviews) after generation, each owned by a named human.
- Without human-owned gates, systems produce drift that just gets mistaken for official, validated output.
- The right split is agents handling the mechanical 80 percent while humans own the final 20 percent of judgment.
Google Apps Script is now a core Google Workspace service with enterprise-grade data protection and technical support. Administrators who previously disabled Apps Script can re-enable it to deliver secure custom automations. No action is required for organizations that already had it enabled or for end users.
- Google Apps Script became a core Workspace service on June 23, 2026, now covered by the same enterprise-grade security and compliance policies as Gmail and Drive.
- Admins who previously disabled Apps Script over security/compliance concerns can now re-enable it without needing extra compliance reviews.
- No action is needed for orgs that already had it enabled or for end users—functionality and settings remain unchanged.
- It's now visible in every Workspace customer's admin console regardless of Rapid or Scheduled Release track.
This GitHub repo provides a coding-agent skill that runs automated security audits in six phases—recon, hunting, validation, reporting, structured output, and independent verification—to identify exploitable vulnerabilities. It uses parallel agents to generate and disprove findings, outputs structured JSON conforming to a schema, and independently verifies each claim against the source code. Each run reads prior findings to skip known issues and improve coverage.
- Cloudflare open-sourced the actual skill behind their own vulnerability discovery harness, structured as six phases: recon, hunting, validation, reporting, structured output, and independent verification.
- Findings are adversarially checked—the agent that finds a bug never validates it, and separate fresh agents re-verify each claim against the source code before reporting.
- A single audit pass only catches about half of total vulnerabilities, so the tool reads prior findings.json runs to skip known issues and improve coverage over repeated runs.
- Output is enforced as machine-readable JSON validated against a schema via a zero-dependency Node.js script, alongside human-readable REPORT.md and FINDINGS-DETAIL.md files.
This post breaks down Amazon’s lawsuit against Perplexity over its Comet AI agentic browser, which browses and transacts on users’ behalf while disguising itself as Chrome. It explains how these browsers work, the security risks they introduce—like prompt injection attacks—and why sites like Amazon demand transparent agent identification.
- Amazon's lawsuit claims Comet disguises itself as Chrome, hiding the AI agent's true identity from Amazon's systems.
- Agentic browsers share the user's actual cookies/passwords/session data, so a compromised AI agent can act with full access as if it were the user.
- Amazon's complaint cites a public report of prompt-injection attacks hijacking embedded AI assistants to steal private data.
- Amazon argues Comet's covert automation bypasses and degrades its personalized shopping features built over years.
AWS CEO Matt Garman says Amazon will bring on 11,000 interns and new grads this year even as it rolls out AI agents for recruiting, coding, security, and customer service. He argues AI will reshape entry-level roles rather than eliminate them, pointing to past technology shifts and a growing overall labor force. His upbeat stance on hiring sits alongside Amazon’s broader plans to cut corporate jobs and automate half a million roles with robots.
- Amazon is hiring 11,000 interns/new grads this year even while deploying AI agents for coding, security, and recruiting.
- This hiring push coexists with cutting 30,000 corporate jobs since October and plans to automate/robot-replace up to 500,000 roles.
- Garman argues AI reshapes entry-level work rather than eliminating it, comparing it to how spreadsheets displaced calculators but grew the labor force.
- He warns that companies which stop training junior talent risk long-term stagnation.
Google Cloud and Nokia introduced AI agents in Nokia Assurance Center to automate telecom network operations, with a router agent and event triage agent now live. Four more agents—covering KPI selection, anomaly reasoning, action recommendations and dashboard reporting—will roll out via a SaaS launch on Google Cloud Marketplace in September 2026. Nokia keeps humans in the loop for critical approvals and plans continuous updates through 2027.
- Google Cloud and Nokia's new router and triage AI agents in Nokia Assurance Center claim to cut troubleshooting time by 50-80%
- Four more agents (KPI selector, anomaly reasoner, action reasoner, dashboard agent) launch as SaaS on Google Cloud Marketplace in September 2026, with rollout continuing through 2027
- Nokia's "glass box autonomy" keeps humans approving most fixes, only allowing full automation for low-risk, preapproved actions
- Alphabet's 2025 telecom revenue (~$3B) trails Microsoft ($3.85B) and Amazon ($4.2B) per MTN Consulting
This issue of TLDR Marketing covers nine new LinkedIn features and how to use them, argues that sentiment scores alone miss real social insights, and offers practical tips on subject lines, AI use, and experiment design. It also highlights debates around under-16 social media bans, AI’s role as an augment rather than replacement, confidence scoring flaws, and emerging hybrid AI verification models.
- Yale found zero job losses from AI automation over 33 months, but a 12.2% task completion boost and 25.1% speed gain when used as a helper
- Australia's under-16 social media ban still let 70% of kids retain partial access, showing enforcement limits
- Sentiment scores alone (positive/negative) miss the real insights buried in comment threads and recurring themes
- Rigid p<0.05 significance thresholds can leave up to 25% of potential experimental gains on the table
Orca lets you run multiple code-generation agents (Codex, ClaudeCode, Pi, etc.) side-by-side in isolated git worktrees and compare or merge their outputs. It combines split terminals, UI scraping, remote execution, commit workflows, and real-time mobile notifications to manage and steer agents without context switching.
- Orca runs multiple AI coding agents (Codex, ClaudeCode, OpenCode, Pi, etc.) in parallel using isolated git worktrees, so you can fan out one prompt to several agents and merge the best output.
- It bundles a full dev environment—WebGL terminal, embedded VS Code with autosave, and Chromium-based UI scraping to capture HTML/CSS/screenshots directly into prompts.
- Mobile apps (iOS/Android) let you get notified when agents finish or need input, then steer them, review PRs, and comment on diffs remotely.
- It's MIT-licensed and open source, installable via Homebrew, AUR, or prebuilt binaries, with every action scriptable via CLI commands.
The author uses a local LLM called pi that reads a global CAPABILITIES.md index in markdown, pointing to task-specific docs like SSHing into a studio Mac or driving a browser. Pi only loads detailed instructions when a task matches a capability, and new setups are documented and added to the index so agents grow more useful over time.
- A single symlinked AGENTS.md pointing to a CAPABILITIES.md index lets every project share one growing set of agent capabilities instead of duplicating setup per folder.
- Pi only loads detailed docs when a task matches one of the index's listed capabilities, keeping context small while still scaling to many tools.
- Adding a capability is just three steps (write the doc, add an index line, commit), so the library compounds over time as new setups get documented.
- The agent-docs repo is itself maintained almost entirely by AI, treating documentation upkeep as its own agent task.
Armin Ronacher breaks down two layers of agent loops: the internal cycle where a model calls tools and edits code, and the external harness that re-queues and re-runs tasks until a goal is met. He says loops shine for experiments, code ports, and security scans, but they generate brittle, over-defensive code and threaten human understanding, even as defenders must adopt loops to match automated attackers.
- Autonomous agent loops tend to produce defensive, bloated code—stacking fallbacks and exception-handling instead of fixing root design flaws.
- Loops work best on bounded, verifiable tasks like code ports (Zig-to-Rust, MiniJinja-to-Go), performance tuning, and security audits, where results can be mechanically checked or judged by a second LLM.
- For production or mission-critical systems, Ronacher still prefers deterministic, fully "peelable" code over the unpredictable, living-system feel of loop-generated software.
- Despite the risks, defenders may be forced to adopt loop-based approaches simply to keep pace with automated attackers using the same technique.
@RohOnChain on X
Citadel CEO Ken Griffin says AI agents now do work that used to take master's and PhD-level finance staff weeks or months, finishing it in hours or days. The post is a short promotional teaser, with the real detail behind the quote left to the linked content.
- Griffin claims work usually handled by master's and PhD finance staff over weeks or months is now completed by AI agents in hours or days.
- The post frames this as a look at what AI is already doing inside Citadel, but the body itself offers no specific tasks, tools, or numbers.
This article shows how to export Terraform plans to JSON and evaluate them with conftest policies written in Rego, enabling deterministic auto-apply of safe changes. It walks through examples for allowed actions, resource types, field diffs, blast radius limits, and environment gating to keep control while boosting velocity.
- Terraform plans can be exported to JSON (`terraform show -json`) and tested against Rego policies via conftest to auto-apply only compliant changes.
- AI review tools like Overmind are non-deterministic and fail audit requirements for human sign-off, whereas codified policy checks give repeatable, versioned decisions.
- Policies can be as granular as needed: restrict to no-op/create/read actions, gate specific resource types (RDS, IAM), limit changes to certain fields (tags), cap the number of modified resources, or vary rules by environment tag (staging vs. production).
Founders from the Department of Government Efficiency built SpecialOS, an AI-driven platform that automates tasks in Main Street service industries. Their first target is eldercare via Figure Health, where they’ve acquired a Texas provider, plan to open-source billing claims, and use efficiencies to boost nurse pay.
- Ex-DOGE founders launched SpecialOS to acquire and run Main Street service businesses with AI automation rather than just sell software to them
- First acquisition is Figure Health, a Texas home-health provider with 1,400 patients, where AI-driven billing/scheduling savings will fund higher nurse pay
- Figure Health will open-source its Medicare/Medicaid billing claims for public transparency
- Backed by a16z and a roster including Brian Armstrong (Coinbase), Shyam Sankar (Palantir), and several former DOGE officials, with plans to expand into other regulated, labor-intensive industries and eventually go public
@marcusyul on X
Marcus Yul, writing in Spanish, says Anthropic has published a workshop showing how to build a whole company using only AI agents, with no employees and no meetings. He has subtitled it into Spanish and is urging readers to save it.
- Anthropic's workshop teaches building an entire company run only by AI agents.
- The agents divide up tasks and execute business processes among themselves, with no human staff.
- The post is a Spanish-subtitled version of the workshop, shared as something to save for later.
This article explains how to install and configure the open-source Hermes AI agent on your own server and use it to run a self-improving weather-trading bot on Polymarket. It covers Hermes’s memory, skill-creation, and always-on execution features, then walks through cloning a weather bot repo, setting up wallets, approvals, and trading prompts.
- Claims a $100 → $5,000 result from running a self-learning weather-trading bot on Polymarket via Hermes Agent.
- Hermes auto-generates and refines markdown "skill" files roughly every 15 tool calls, unlike OpenClaw which needs manual skill installs and third-party memory add-ons.
- Setup is pitched as under 5 minutes on a rented VPS: SSH in, run a one-line install script, pick ChatGPT or Claude, connect a messaging gateway (Telegram/Slack), and launch.
- The bot workflow scans forecasts across 20 cities and uses expected value plus Kelly Criterion sizing to generate trade alerts.
Obscura is a Rust-based headless browser engine that runs real JavaScript via V8 and implements the Chrome DevTools Protocol for Puppeteer and Playwright compatibility. It ships as a single 70 MB binary with just 30 MB RAM usage, supports stealth mode with fingerprint randomization and tracker blocking, and delivers faster startup and page loads for large-scale scraping and AI agent automation.
- Obscura uses ~30 MB RAM vs Chrome's 200+ MB, and loads pages in ~85 ms vs Chrome's ~500 ms
- Ships as a single 70 MB binary with no Node.js or Chrome dependency required
- Speaks the Chrome DevTools Protocol, so it works as a drop-in for Puppeteer/Playwright workflows
- Built-in stealth mode randomizes fingerprints and blocks 3,500+ tracking domains for scraping/automation at scale
Matthew Gallagher built MEDVi, a telehealth service for GLP-1 weight-loss drugs, using only AI tools and one sibling in under two months. He outsourced medical and logistics functions, automated marketing end-to-end with AI, and drove $400 million in revenue his first year while targeting $1.8 billion next.
- One person (plus one sibling) ran MEDVi to $400M revenue and 250,000 users in a single year by outsourcing prescriptions/logistics and using AI tools for everything else
- MEDVi's 16.2% net margin is roughly double competitors like Hims & Hers
- Gallagher's prior startup, Watch Gang, hit $11M revenue but collapsed under 60 employees, teaching him that headcount kills profitability—directly shaping MEDVi's lean, AI-driven model
- He's targeting $1.8 billion in revenue for 2026 despite criticism over "grey-area" marketing tactics
Claude Code Auto Mode automates permission checks by assessing the risk of each action instead of prompting you every time. It blocks or escalates unsafe operations—like mass deletions or external network calls—while allowing routine tasks to run headlessly. This differs from the dangerous “skip permissions” flag, which removes all guardrails.
- Claude Code Auto Mode uses risk assessment (reversibility, scope alignment, risk surface, cascading effects) to decide whether to allow or block actions instead of prompting a human every time.
- Unlike Auto Mode, the "--dangerously-skip-permissions" flag removes all safety checks and should only be used in disposable test environments.
- Auto Mode is configured via .claude/settings.json or CLI flags (--allowedTools/--disallowedTools), letting you explicitly allow patterns like "Read(*)" or "Write(src/**)" while denying destructive commands like "Bash(rm -rf*)" or "WebFetch(*)".
- It's best used in sandboxed or controlled environments (dev containers, test VMs) so Claude can handle routine tasks in headless workflows while still blocking genuinely risky operations.
The author argues that modular “Skills”—reusable markdown workflows loaded on demand—outperform standalone AI agents by cutting token bloat and maintenance overhead. A live GEO audit system built with Skills shows how you can turn domain expertise into scalable, service-ready products without managing dozens of agents.
- Claude's "Skills" load modular markdown playbooks on demand instead of baking everything into prompts, citing 53 tokens for passive reference vs. embedding a full prompt every time
- A live GEO audit system built entirely on Skills scrapes visibility across ChatGPT/Gemini, flags gaps like missing Wikipedia entries, and auto-generates client-ready reports without spinning up separate agents
- The whole GEO pipeline is public and forkable, letting anyone productize it without building custom infrastructure
- Documenting expertise once in a markdown file and iterating on it lets teams ship service-ready AI products in days rather than maintaining fleets of bespoke agents
This article explains how to create and configure Claude Code routines—autonomous cloud sessions triggered on a schedule, via API, or by GitHub events. It covers setting up prompts, repositories, environments, and connectors through the web UI, CLI, or desktop app, and illustrates use cases like issue grooming, alert triage, code review, deploy verification, docs drift, and library porting.
- Claude Code routines run autonomously on schedules, API calls, or GitHub events (like pull_request.opened), not just interactively.
- Each run clones repos, executes shell commands or connector calls (Slack, Linear, Google Drive), and pushes changes to claude/-prefixed branches for review as PRs.
- Real use cases include nightly issue grooming with Slack summaries, API-triggered error-threshold triage that opens draft fix PRs, and automated code porting between SDK repos on merge.
- Routines can be built via web UI, CLI (/schedule commands), or the Desktop app's "New remote task" option.
Chrome now lets you save reusable Gemini prompts as “Skills” for one-click access instead of retyping or copy-pasting each time. Saved Skills sync across devices when you’re signed in, and you trigger them with “/” or the plus button in Gemini. They still require confirmation for actions like messaging or calendar entries.
- Chrome's Gemini now lets you save reusable prompts as "Skills," triggered via "/" or the "+" button instead of retyping or pasting them each time
- Skills sync across desktop devices when signed into Chrome
- Skills can operate across multiple open tabs at once, useful for tasks like comparing prices or aggregating research
- Actions like sending messages or adding calendar events still require user confirmation, keeping the same security checks as regular Gemini prompts
Anthropic has introduced repeatable routines in Claude Code that run on its web infrastructure, so tasks execute even if your Mac is offline. The feature, now in research preview, lets Pro, Max, Team, and Enterprise users schedule automations with repo and connector access, subject to daily run limits. The update also includes a redesigned Mac app with parallel sessions, an integrated terminal, file editing, and preview tools.
- Claude Code now runs scheduled "routines" on Anthropic's servers, so automations execute even when your Mac is offline
- Daily routine limits scale by plan: 5 for Pro, 15 for Max, 25 for Team/Enterprise
- Routines come pre-connected to repos and connectors, eliminating manual cron job or server setup
- The redesigned Mac app adds parallel sessions with a sidebar, plus a built-in terminal, file editor, and HTML/PDF preview pane
The article argues that AI can now generate and manage design systems and dashboards better than humans, making manual frameworks and large UI teams obsolete. It predicts a shift from uniform, high-cognitive-load interfaces to conversational, intent-driven experiences that deliver only the insights users need.
- AI can now generate and customize full design systems on demand, undercutting the need for large design teams and paid seats in tools like Figma
- Design systems have become their own worst enemy—organizations spend more time managing them than they save, producing generic, uninspired interfaces
- Dashboards demand too much cognitive effort and assume users already know what to look for, making them a poor fit for actual decision-making
- AI chatbots that query databases directly and generate charts on the fly are replacing fixed dashboards, delivering precise answers without manual analysis
This article compares two main options for setting up an AI agent: the user-friendly Claude and the customizable Hermes. It explains the differences between AI agents and chatbots, outlines the setup processes for each option, and emphasizes the importance of .md files for effective AI interaction.
- Claude's Cowork mode requires a paid subscription starting at $20/month to let it act autonomously on your computer.
- Hermes has no known security vulnerabilities as of April 2026, whereas OpenClaw has had multiple.
- Hermes offers better memory and learning from past interactions than OpenClaw, at the cost of a more involved setup.
Career-Ops is an AI-driven tool that simplifies job searches by evaluating offers, generating tailored CVs, and tracking applications in one place. It uses a structured scoring system to help users focus on high-fit opportunities without spamming companies. The system is customizable and designed for efficiency.
- Creator personally evaluated 740+ job offers and generated 100+ tailored CVs using this system before landing a Head of Applied AI role
- Uses an A-F scoring system across ten weighted dimensions to rank job fit and cut wasted applications
- Automates job scanning across Greenhouse, Ashby, and Lever for 45+ companies with batch processing of multiple offers at once
- Generates ATS-optimized CVs as PDFs tailored to specific job descriptions, run through Claude Code with a dashboard for tracking applications
Paperclip is a platform that manages AI agents to streamline business operations. It allows users to set goals, hire agents, and monitor their performance from a centralized dashboard. Unlike traditional task managers, Paperclip integrates organizational structures and budget controls for efficient agent coordination.
- Paperclip orchestrates multiple AI agents as a unified company rather than just managing individual tasks, adding org charts, budget controls, and goal alignment on top.
- It runs locally via Node.js/pnpm with an embedded, self-managed PostgreSQL database, and can run multiple isolated companies from one deployment.
- Agents run on schedules or event triggers with built-in cost tracking to prevent budget overruns and maintain persistent state.
- The platform supports plugins, allowing extension beyond its core orchestration and governance features.
While AI tools can automate tedious tasks like sorting emails and taking notes, they may inadvertently limit creative thinking and problem-solving. The risk lies in losing valuable insights that often arise during repetitive activities, highlighting a potential downside to increased productivity.
- Boring, repetitive tasks (sorting emails, note-taking) often give the mind space to wander, and that wandering can spark unexpected creative insights or solutions.
- Automating those tasks with AI removes the "downtime" that fuels incidental problem-solving, so productivity gains may come at the cost of creativity.
- The tradeoff is subtle and easy to miss, since the lost value (an idea that never occurred) isn't as visible as the time saved.
A recent bug in Claude Code's CLI, caused by a changelog format change, highlights the challenges of rapid AI-assisted development. As release velocities increase, existing systems struggle to keep up, leading to potential drift between components and insufficient oversight on changes. Automation tools are needed to manage this new pace effectively and prevent issues like "changelog drift."
- A changelog format change broke Claude Code's CLI, shipping a bug that basic checks should have caught
- Faster AI-assisted release cycles are outpacing the review/testing systems meant to catch such issues
- This creates "changelog drift," where components fall out of sync because no one's watching closely enough
- The fix isn't slowing down but building automation that matches the new development speed
Explore around 30 pro-tips for maximizing the efficiency of Gemini CLI, an open-source AI assistant designed for command-line use. The guide covers setup instructions, essential features, and advanced techniques for coding, debugging, and automating tasks through natural language prompts.
- Gemini CLI is a free, open-source command-line AI assistant, and the repo compiles ~30 practical tips for using it effectively in coding workflows
- Covers concrete techniques like setting up context/config files, using custom slash commands, and chaining prompts for debugging and automation
- Focuses on efficiency tricks (e.g., scripting repetitive tasks, managing context windows) rather than just basic feature explanations
The article draws parallels between the early internet era and the current landscape of artificial intelligence, highlighting the dichotomy of optimism and pessimism surrounding AI's impact on employment and productivity. It explores how different industries will experience varying outcomes based on the balance between unmet demand and automation capabilities. Historical perspectives on past technological shifts provide context for understanding AI's potential future.
- Radiology jobs and pay grew despite predictions AI would replace radiologists, because Jevons Paradox kicked in—cheaper/faster scans increased overall demand for imaging.
- Whether AI creates or destroys jobs in an industry depends on whether demand is already saturated: textiles boomed then crashed once automation met demand, while motor vehicles kept growing because demand stayed unmet.
- Job displacement risk is concentrated in repetitive, easily automated tasks rather than complex expert work.
- Software engineering faces a unique open question: automating app development could hit a demand saturation point, unlike other tech-driven fields.