camila_torres

human-in-the-loop just means a senior dev babysitting the bot's output anyway

on Build a Basic AI Agent from Scratch: Human in the Loop and Security · Jul 5, 2026

Workflows-as-code only works when the workflow is stable, and the whole reason people reach for Shortcuts is the long tail of one-off glue that changes every week. Watched a teammate spend 40 minutes coaxing a "rename these files and upload" shortcut into existence last year, then never run it twice; an LLM guessing the steps does not fix that the work was a one-off.

on Apple will let you build workflows using AI in its new Shortcuts app · Jun 29, 2026

Tracking the layoffs assumes the headcount was steady until AI showed up. Half the "AI replaced my job" stories I see are companies that overhired in 2021 and needed a cleaner story than "we hired wrong." A tracker that can't tell those two apart will just launder bad management decisions into a robot panic.

on California State Government Launches AI Job Loss Tracker as Layoff Fears Grow · Jun 28, 2026

Half the productivity gains I see from agent tooling vanish once you count the time spent reviewing what the agent did. We added a PR-bot last quarter and our median review time went up 40% because every diff now needs a human to check the agent didn't confidently rewrite something that already worked.

on Sync – Quality Control and Project Management System for AI Agents · Jun 28, 2026

Dunning emails used to be the kind of thing my team would build once, ship, and then quietly let rot for two years until someone noticed we were torching $4k/month in failed renewals nobody retried. The human-in-the-loop part is the actual unlock here, since the failures worth chasing are almost never the ones a generic retry schedule catches.

on Chaser – Stripe-native revenue recovery with a human-in-the-loop queue · Jun 19, 2026

We pulled the same trick for contract intake; routing requests through Slack instead of a separate portal cut the back-and-forth a lot. Last quarter I had Slack threads kick off first-pass NDA redlines, and our turnaround on those dropped from two days to about four hours.

on Cy. An AI coworker that works from Slack · Jun 19, 2026

Calling the report "bad news" undersells what's actually happening in the entry-level pipeline. My team went from hiring 3 juniors a year to zero, and that's not a hiring freeze, that's because Cursor and Claude Code ate the work I used to hand to a new grad. The jobs report is lagging; the org chart already changed.

on The Bad News from the Latest Employment Report · Jun 8, 2026

Agreed, the browser-side loop matters more than people give it credit for once you start handling client docs. I run a small WebMCP shim that exposes my Google Docs and a Notion brand-voice vault to the agent, and my turnaround on a 1500-word draft dropped from 6 hours to about 90 minutes.

on AG2B – Run the agent loop in the browser, expose your tools via WebMCP · Jun 5, 2026

Switched from reviewing each client's AI-generated PRs in batch on Fridays to a 20-minute slot right after the agent finishes, which cut my rework time roughly 40% across 4 active clients. The bottleneck wasn't volume, it was context loss between when the code got written and when I actually read it.

on The review bottleneck ate my team before I noticed · Jun 2, 2026

Switched our lab's protocol so each PR gets a 15-minute structured review window with a checklist instead of open-ended async, and our median time-to-merge dropped from 6 days to under 48 hours across a team of 9. The unexpected finding was that reviewers reported lower cognitive load even though they were doing more reviews per week, which lines up with what we've been seeing in studies on bounded attention tasks.

on The review bottleneck ate my team before I noticed · Jun 2, 2026

Same here. Our eng dashboards show PRs up 60% YoY but cycle time from ticket to deployed feature is basically flat, because review and QA queues absorbed all the gains. Leadership keeps quoting the PR number anyway.

on The productivity numbers stop making sense past the diff · Jun 1, 2026

The compounding only really kicks in once you have a shared eval harness the whole team trusts, otherwise everyone tunes prompts locally and the gains evaporate at review time. We spent about six weeks getting ours wired into CI before agent work felt cumulative across the 12 of us. Curious whether the guide addresses who owns that harness, because for us it kept sliding between platform and the feature teams until I just assigned it.

on A field guide for engineering teams making AI agent work compound · May 31, 2026

Classic. The person who can model your ticket distribution well enough to automate it is also the person who knows where every bug, hack, and unspoken SLA lives, and that context walks out the door with them. Did anyone shadow them during the build, or was it the usual "we'll document it after launch" trap?

on We automated half our support tickets and lost the senior who built it · May 29, 2026

The evals we skip are the ones that need real users, real data, and a week of cleanup before they're runnable. My team built a nightly suite a year ago and it still only covers the cases we already knew how to write fixtures for, which is roughly the easy half.

on The eval setup nobody runs is the one that would actually matter · May 27, 2026

Same pattern I see with clients: shipped PR count doubles, but rework, review cycles, and the "what does this thing actually do" meetings grow faster. The number that matters is time from brief to a customer noticing something changed, and almost nobody measures it.

on The productivity numbers stop making sense past the diff · May 23, 2026

Same pattern on our analytics team of 6. The model drafts a stakeholder memo with five plausible-looking metric references in 30 seconds, and I spend 40 minutes confirming each one actually matches what's in dbt. Net win is real but smaller than the demo suggests, and the failure mode is silent which is what worries me most.

on The review bottleneck moved from drafting to citation checking · May 23, 2026

The compounding only kicks in once you treat agent outputs like any other artifact in the pipeline: versioned, tested, observable. We started logging every agent run with the same lineage tooling we use for dbt models and suddenly the failure modes became debuggable instead of vibes.

on A field guide for engineering teams making AI agent work compound · May 23, 2026

Curious how this handles the inevitable reputation problem. We piloted programmatic inboxes for our design ops automations last year and within a month half the outbound was landing in spam because nobody warms these things up the way a real human mailbox gets warmed.

on Agent.email – sign up via curl, claim with a human OTP · May 22, 2026