Ai-Work

A group of developers argues that AI-assisted work should be judged by its outcomes, its clarity, and the review burden it puts on other people.

A weekly AI roundup covers Nvidia's CEO transition, model comparisons, Adobe's creative tools and new AI product interfaces.

IBM's Mixture of Experts panel debates AI creativity and Nvidia's leadership change; Confluent's Shawn Falconer explains a streaming-data architecture.

Coding agents make ambitious personal software easier to attempt, raising questions about understanding and maintaining what they produce.

Matt Wolfe tours Peter Yang’s AI-assisted creator workflow and builds smaller examples: an inbox brief, a human-owned goals document, a bass-practice check-in and skill edits reviewed across sessions.

In a sponsored tutorial, Peter Yang uses Opus for routine execution and Fable for difficult planning and review across mail, career advice, content production, app building and care coordination.

Scott Farrar shows Claire Vo a household vault of appliances, warranties and projects that Claude Code can read, with Obsidian for inspection and a small wall display for next actions.

A Reuters Institute survey finds that people increasingly use AI to get answers, but only 12% are comfortable with news made entirely by AI.

For a repeated operating procedure, ReTrace uses a few logged agent runs to compile a decision list. It extracts the conditions and tool inputs from the agent’s explanations and actual calls, merges them into cited rules, then generates …

An internal ticket-labeling case study puts a smaller decision model on live traffic, with a 98% estimated cost cut and important gaps in the quality evidence.

Vorflux's Prasanna Sankar describes cloud development setup, tests, review, merge queues and the limits of always-on work.

Francesco Galletta and Felippe Felisola Caso use a wedding-budget workshop to show how a financial-agent harness can separate questions, calculations, evidence, decisions and independent checks.

Untangle's Ryan Carson describes manager and builder agents, a timed check-in, QA and a merge queue. His hiring process uses agent-ranked candidate recordings and deserves separate scrutiny.

William Imoh built a voice-first Norwegian practice app in Swift and Kotlin despite having no Swift experience. The interview also exposes the cost and access-control trade-offs of his agent workflow.

Pratyusha Duvvuri shows why a token bill needs provider reconciliation and two ownership maps: the reporting tree engineers use and the cost centers finance uses.

A father-son team built an AI-generated financial model. An auditor found two errors that canceled out, making its forecast look better than its assumptions.

Hamel Husain and Angelina Yang work through trace review, judge calibration, and the human decisions behind useful product evaluations.

Matthew Jewkes separates errors, unsettled definitions, and exceptional cases to automate routine work and recover expert attention.

A Git-backed task queue and a human decision inbox aim to make parallel coding-agent work less overwhelming.

McKinsey practitioners describe smaller teams, shared context and quality checks for AI-assisted development, while acknowledging that measuring impact remains hard.

Peter Yang and OpenAI's Ashley Alexander discuss organizing records, preparing questions and the emotional work of caring for parents with cancer.

Range's Matt Ragone explains how planners turn repeated work into reviewed AI skills, and why ownership continues after release.

OpenAI’s Stefano Fabbri and Oliver Steffmann connect financial data, skills, plugins, and custom interfaces, then discuss deployment, evaluations, and writeback.

Nori Agentic investigates changed financial records, cites evidence, and asks a human to approve. Its reported five-minute figure measures human review effort.

Former Cursor designer Ryo Lu uses a journal-app demo to show how editable plans, small changes, and collaboration help people develop their own judgment.

Sid Pednekar uses two hypothetical asset-management firms to explain problem selection, shared capabilities, and learning after AI deployment.

Midwest AI Summit speakers connect an AI charter, human accountability, maintenance costs, and deployed agents that check work orders and answer data questions.

Nous Research’s Dillon Rolnick argues that agent platforms create switching costs, and makes the case for open foundations with replaceable models.

AI can produce a result that passes its checks. Who can defend it, question its assumptions and repair it when something goes wrong?

Matt Wolfe builds reusable scripting and editing skills, then fixes a rough cut through feedback, waveform checks and saved workflow changes.

A shared patient brief, organized records, better questions and expert updates: Peter Yang's five-step caregiver workflow, with clinical decisions left to the care team.

Ben Stolovitz runs five or six coding agents, but finds human attention, engineering judgment, and the pleasure of making things harder to replace.

Two migration campaigns opened 2,333 requests and landed 729. Stronger checks helped, but owner review and deployment remained real work.

A complaint about an apparently automated support reply gets a warmer apology. Whether anyone fixed the bug remains unanswered.

AI-assisted discovery challenges a profession built around individual achievement. Three mathematicians argue for the joy of understanding; readers ask who sustains the people doing it.

Simile's Joon Sung Park explains interview-grounded behavioral simulation, its validation problem, and the gap between customer panels and predicting society.

Cristóbal Valenzuela demonstrates Continuum, a research prototype for interactive visual explanations and shared visual work with coding agents.

A dashboard for human decisions, agent run histories, workflow changes and regression tests, backed by Archon and Postgres.

As AI makes producing proofs easier, a mathematician argues for rewarding understanding, good questions and the work that connects results.

Figma's Ezra Mechaber shows a shared product-management skills repo, agent-assisted testing and an experiment workflow.

Adam Nahari argues that AI projects need a measurable business problem, workflow redesign and committed users to produce financial impact.

Zack Proser shows how he turned writing experience into a tool for colleagues, and applies the same idea to bedtime stories for his children.

Nick Nisi turns code changes into simulated smells and proposes a risk-based review pipeline.

Maddy Zhang argues that verification, judgment, agent workflows and communication matter as coding agents change engineering work.

Jaclyn Konzelmann discusses short strategy documents, user co-creation, experiments, and designing for an agent-filled world.

Compound's Yijia Liang explains shared agent inboxes, Office-compatible engines, and checking loops for financial deal teams.

Barrow explains how Perplexity Decider scores classification choices in one model pass, and why fast scores still need careful validation.

Cole Medin examines a study of AI-influenced expert decisions and suggests ways to keep judgment in the planning process.

Bumble's Supriya Subudhi shares five lessons from testing AI photo feedback: adoption, bad outputs, costs, model choice and regional rollout.

Automating routine incident response can remove the practice engineers need for unfamiliar failures. HN adds training ideas and a challenge to the premise.