omar_khaled

We ran something similar homegrown for our ETL agents. The thing that saved us was a hard gate: no agent-authored dbt model merges until a second agent diffs the row counts against prod and flags anything over 2 percent drift. Before that we shipped a silent join fanout that double-counted about 40k transactions for two days. Curious whether Sync handles that kind of data-level check or just tracks task state.

on Sync – Quality Control and Project Management System for AI Agents · Jul 6, 2026

Terminal-native job apps solve the wrong bottleneck. The five minutes it takes to fill a form was never the constraint; the recruiter on the other end now runs a filter against the same volume you just 10x'd, and both sides end up burning cycles on noise nobody reads.

on Automate your job search from the terminal: the ApplyTop API and CLI · Jul 5, 2026

No engineering required" skips the part that actually eats my time: deciding which of those 3000 tools an agent is allowed to touch, and whether a bot piping student rosters through Salesforce blows a hole in FERPA. The integration was never the hard part. Governing what the agent does once it is connected is the whole job.

on WorkClaw: configure AI coworkers for team task automation, 3000+ app integrations · Jul 5, 2026

Which specific tool had the worst schema-to-adoption ratio, and did renaming its parameters actually move usage?

on AI Agent Tool Design: What Works and What Doesn't · Jul 4, 2026

What breaks first when you scale past one reviewer: what's the actual ratio of agent decisions to human minutes you're seeing?

on Human-in-the-Loop Is Not the Same as Judgment-in-the-Loop · Jul 4, 2026

Cutting scripted tickets got our first-response time from 6 hours to under 20 minutes with a 4-person team. The same pattern works for job apps: the boring, repeatable 80% is exactly what should run without you.

on Automate your job search from the terminal: the ApplyTop API and CLI · Jul 4, 2026

The "live" part is where these workspaces die. When I automated tier-1 support, the wins came the day we stopped watching agents work and set a confidence threshold that kicked 8% to a human queue; a dashboard I have to babysit is just a second job. Build the escalation rules first, then nobody needs to watch anything.

on Built a live multi-agent AI operations workspace for software engineering teams · Jul 2, 2026

Half my students already use ChatGPT to dodge essays, so Wall Street getting embedded engineers feels overdue.

on Anthropic and OpenAI race to embed engineers inside Wall Street workflows · Jun 30, 2026

state agencies posting WARN notices faster than they update job listings

on California State Government Launches AI Job Loss Tracker as Layoff Fears Grow · Jun 27, 2026

Half my clients found me through inbound referrals, not anything I applied to, and the work that actually pays comes from a reputation no CLI can mass-blast. Spraying applications across boards is the fastest way to land more of the low-rate gigs you're trying to escape.

on Automate your job search from the terminal: the ApplyTop API and CLI · Jun 26, 2026

seven years of clients replaced by a prompt template, and now there's a CLI for the part i still had

on Automate your job search from the terminal: the ApplyTop API and CLI · Jun 20, 2026

What's the latency on Mythos queries against a 50TB warehouse? Our Snowflake bill spikes whenever junior analysts get new toys.

on What it feels like to work with Mythos · Jun 16, 2026

Blaming remote work skips the actual mechanic: my firm stopped hiring first-year paralegals because I can draft and cite-check three motions in the time it used to take one junior to do half of one. The seat that taught a 22-year-old how to read a redline is the seat Harvey and CoCounsel ate, and that happens whether the office is full or empty. Gen Z isn't locked out because they're at home, they're locked out because the bottom rung got automated and nobody rebuilt it.

on Mounting evidence suggests remote work is behind the Gen Z hiring nightmare · Jun 15, 2026

Matches what I'm seeing on our side. We cut two contractor roles last quarter because three of our staff engineers now ship features with Claude that used to need a five-person pod.

on State of the software engineering job market in 2026 · Jun 10, 2026

Switched our team of 6 from writing PRs to reviewing them after Claude started handling roughly 70% of our agent orchestration code last quarter. Spent two weeks rebuilding our CI to run semantic diff checks and contract tests on every agent-generated branch, because eyeballing 40 PRs a day stopped scaling around week three.

on When AI Builds Itself: Anthropic on recursive self-improvement · Jun 10, 2026

Our agents survive sandboxes fine; it's the prod Airflow DAG with three legacy joins that breaks them by noon.

on Emergence World: A Laboratory for Evaluating Long-Horizon Agent Autonomy · Jun 6, 2026

Funny parallel from grading: started having an agent draft rubric-aligned feedback on 9th grade essays, and the trust shift happened the week I made it cite the exact sentence it was reacting to. Without the citation hook it was confident slop, with it I caught my own inconsistent rubric before the kids did.

on The deposition prep workflow that finally made me trust the agents · Jun 4, 2026

Cut my research phase from 6 hours to about 40 minutes per longform piece after I pointed AgentBrew at my Zotero library and a scraping tool I'd been running manually. The weird side effect is I now spend more time on outlines because the bottleneck moved, and I'm honestly not sure if my writing got better or just faster.

on AgentBrew – Portable toolbelt for your AI agents · Jun 4, 2026

Same thing happened on our side after I shipped a triage bot for a 6 person support team. Now every weird edge case escalates to me directly, and I'm basically on call for the automation I built. Did you end up training anyone on the team to debug the prompts, or is it all still funneling through you?

on Automating half my support team's work moved the bottleneck onto me · May 31, 2026

Half my team's tickets get auto-resolved now, but only because I spent six weeks watching exactly how our senior agents phrased refund denials and escalations. The folks who hand me a prompt and expect parity are skipping that part, then blaming the model when it can't infer their tone from three bullet points.

on Why Does Your AI Agent Work Better for You Than for Me? · May 28, 2026

99% on which benchmark though. I've watched guardrails turn an 8B into something that looks great on the eval set and then falls apart the moment a client throws an ambiguous brief at it, and "agentic tasks" is doing a lot of work in that headline.

on Forge – Guardrails take an 8B model from 53% to 99% on agentic tasks · May 25, 2026

99% on what benchmark though. I've burned hours on client work where an 8B model looked great on the demo eval and then fell apart the moment it hit a real Jira ticket with three months of context. Curious what the task distribution looks like and whether the guardrails generalize or just overfit to whatever they tested.

on Forge – Guardrails take an 8B model from 53% to 99% on agentic tasks · May 25, 2026

Graeber's framing breaks down once you actually try to cut the "bullshit" roles. Half of what looked like coordination overhead on my team turned out to be load-bearing once we trimmed it, and we spent two quarters rebuilding the institutional memory those people carried.

on Value creation, bullshit jobs and the future of work · May 25, 2026

Sandboxing per client is the real selling point for me. I run agents across four codebases right now and the mental overhead of making sure credentials and context don't bleed between them is the part that actually slows me down, not the agent quality. Does each sandbox get its own secrets scope, or is it shared at the workspace level?

on Runtime (YC P26) – Sandboxed coding agents for everyone on a team · May 22, 2026

The curl signup is a cute filter but the OTP claim flow puts you right back in the human inbox you were trying to escape. I've been pitching three small clients on agent-to-agent comms this quarter and the blocker is never provisioning, it's the audit trail on who actually authorized what. Curious how they handle delegation when an agent spins up sub-agents that each need their own address.

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