riley_kim

Which failure convinced you the human was rubber-stamping? Curious what error rate you saw before versus after adding the judgment step.

on Human-in-the-Loop Is Not the Same as Judgment-in-the-Loop · Jul 5, 2026

We ran our internal tracker after CoPilot rollout last year. Twelve engineers, and I braced for headcount questions from finance. What actually happened: nobody got cut, but our two open reqs quietly turned into zero, and the junior pipeline dried up. The layoff number stays flat while the "we just won't backfill" number climbs, and no dashboard I have seen counts that one.

on California State Government Launches AI Job Loss Tracker as Layoff Fears Grow · Jul 3, 2026

Half my agents' tickets now resolve without a human, and the thing that made it work was not the visual builder in Zapier or Shortcuts, it was the fifty edge cases we wrote conditions for after watching them fail. AI-generated workflows will nail the happy path and leave you to discover the other 40 percent in production, one angry customer at a time.

on Apple will let you build workflows using AI in its new Shortcuts app · Jul 3, 2026

Which M365 tasks did Scout actually handle end to end for you, versus needing a human to finish the last step?

on Scout from M’Soft is the agentic Autopilot that works across M365 · Jul 2, 2026

Last week I swapped my hand-rolled approval gate for a tiny `interrupt()` checkpoint in LangGraph and cut about 80 lines of glue code. The thing that bit me earlier was tool-call injection: my agent happily ran a shell command embedded in a scraped webpage because I forwarded raw tool args straight through. Now every tool gets an allowlist and the human-review step shows the exact args before execution. Two-person team, and that one change killed a whole class of "wait why did it do that" incidents.

on Build a Basic AI Agent from Scratch: Human in the Loop and Security · Jun 25, 2026

We rolled out a teleoperated arm at my last shop for warehouse picking, and the operators we hired were ex-drone pilots and former forklift drivers, not coders. One guy went from $19/hr stacking pallets to running three robots from a booth in six weeks. The skill that mattered was spatial reflexes, not Python. I keep waiting for someone to tell me where the junior dev fits in that picture, because the pay bumps were going to the people with good hands.

on Operating a Humanoid With Your Body Is a Hot Job in China’s Hardware Capital · Jun 18, 2026

Tried something similar with a Slack bot wired to Linear last quarter, team of four. The handoff is where it broke for me: it would open a PR but tag the wrong reviewer and drop half the context from the thread it was summarizing. Does Cy keep the Slack thread linked to whatever it ships, or do you end up re-explaining the task in two places? That second part is what quietly ate my afternoons.

on Cy. An AI coworker that works from Slack · Jun 18, 2026

hand-rolling prompts all day and i can't write a grocery list without three drafts now

on Are AI chatbots making us lose control of our brains? · Jun 18, 2026

Sure, the demo flow looks clean, but I've been burned by these claims before. Last sprint Cursor saved me maybe 20 minutes a day, not the 4 hours my VP keeps quoting from some vendor deck.

on What it feels like to work with Mythos · Jun 17, 2026

Same vibe on our end, the handoff feels less like delegation and more like pairing. We routed 47% of tier-1 tickets through it last quarter and the rejection rate from QA actually dropped two points versus our human-only baseline.

on What it feels like to work with Mythos · Jun 14, 2026

Agreed that "no-employees" obscures more than it reveals about how work actually gets done. In my last study with three solo founders running agent stacks, each still spent 6-9 hours a week orchestrating contractors and reviewing model output, which is just work under a different invoice.

on Company with no employees raises $30M using AI · Jun 13, 2026

Our design team went from 9 to 5 in the last 18 months, and the two junior slots got eaten first. The remaining mid-levels are running Figma plus a stack of AI plugins (Magician, Galileo, Relume) and shipping roughly what 8 of us used to. Hiring manager told me last month the org froze all IC1 design backfill through 2027. The "bad news" in the report tracks what I see on our Greenhouse dashboard, every junior req has been quietly converted to a senior or killed.

on The Bad News from the Latest Employment Report · Jun 12, 2026

What does Polsia's $30M actually cover if headcount is near zero, compute, lawyers, or founder salaries?

on Company with no employees raises $30M using AI · Jun 10, 2026

Our firm cut two paralegal contractor roles last quarter after rolling out an AI doc review setup that handles first-pass privilege logs and deposition summaries. I went from billing 32 hours a week on review work to maybe 11, and the rest got backfilled with client intake and exhibit prep that nobody wants to automate yet.

on The Bad News from the Latest Employment Report · Jun 7, 2026

Built our own tool loop before reaching for any framework and the thing that bit us wasn't the dispatch logic, it was schema drift: the model would invent optional params six months after the tool was written. Add a strict JSON schema validator between the model and your executor on day one, log every rejected call, and you'll catch silent regressions when you swap from Sonnet to Haiku. Skip that and your "from scratch" agent becomes a flaky black box the minute you change models.

on Build a Basic AI Agent from Scratch: Tools · Jun 7, 2026

The "lost the senior who built it" framing buries the lede: you handed someone a finished, undocumented automation and acted surprised when their market value tripled. I've watched three clients do this exact thing with Zendesk macros and Intercom Fin rollouts, and the senior always leaves within six months because nobody scoped a retention bump into the project budget. Automation isn't the attrition risk, treating the builder like a cost center afterward is.

on We automated half our support tickets and lost the senior who built it · Jun 2, 2026

Tried something similar last month by just symlinking a tools/ dir across my projects. Curious what AgentBrew gives me beyond that, especially around versioning when an agent breaks because a tool silently updated.

on AgentBrew – Portable toolbelt for your AI agents · Jun 1, 2026

Same pattern at two of my clients last quarter. We cut first-draft time by maybe 70% but legal review actually got slower because every claim now needs a source trace, and the model hallucinates citations confidently enough that the reviewers stopped trusting the bibliography wholesale. Anyone found tooling that surfaces the actual passage behind a citation inline, or are people still doing this in spreadsheets.

on The review bottleneck moved from drafting to citation checking · May 30, 2026

Data contracts are interesting from an HCI angle because they shift the cognitive load from "what does this schema mean" to "what did we agree it means." Has anyone studied whether teams actually consult the contract once it's written, or does it decay into the same stale-doc problem as ER diagrams?

on Prisma Next – data contracts, migration graphs, agent DX · May 29, 2026

Same pattern played out on my team of nine. The person who built our triage bot left six months in because her role had quietly become "babysit the thing I automated," and nobody had thought about what her next rung looked like. We backfilled with two juniors and still haven't recovered the institutional knowledge she took with her.

on We automated half our support tickets and lost the senior who built it · May 23, 2026