This website uses cookies

Read our Privacy policy and Terms of use for more information.

Sponsored by

Observability you can act on, whether you build it or lead it, plus the few signals worth your time.

This week's Signal Drop is the whole AI-ops governance model in one line: automate the action, never the accountability. Here is how to draw that line before an agent draws it for you.

Automate the action, never the accountability.

This week's Signal Drop

The Gate Stays Human: Who's Accountable When AI-Ops Acts. Agents can act now, not just suggest: roll back a deploy, scale a service, flush a cache. The model that keeps you out of trouble is simple. You do not decide automation by how confident the tool is; you decide it by blast radius, by how much damage the action does if it is wrong. Low blast radius, the gate stays open. High blast radius, a human decides every time, no matter how sure the machine is.

The Gate Stays Human: Who's Accountable When AI-Ops Acts

METRICS & MAYHEM

The Gate Stays Human: Who's Accountable When AI-Ops Acts

00:00
00:00

Prefer to watch? It is on YouTube.

What you'll take away:

  • Why blast radius, not confidence, is the axis that decides what an agent may do alone.

  • The three gates: open for low-impact, closed for high-impact, conditional in the middle (act, but keep a human able to reverse it).

  • The two-incident test: the same tool, the same night, two opposite answers, and the only thing that changed was the damage on the table.

  • Why a machine cannot be accountable, so the gate on anything that matters stays where accountability already lives.

  • Treat every agent like a service: a named owner, a runbook, and a log you can replay. An agent with no owner is just an outage with a schedule.

Latest on the blog

  • Causal AI vs Generative AI: Which One Actually Finds Your Root Cause? One AI is fluent, one reasons about cause. Which you trust for a root cause decides whether you fix the incident or a coincidence. Read it.

What you missed last week

The recent posts, a couple of signals, and what was worth the time.

Tech Tuesday

No Tech Tuesday this week; it runs every fortnight. If you missed last week's How Do You Observe an AI Agent?, catch up there first: ask, decision, action, outcome, and why the trace is the camera, not the gate. The next one is worth the wait: OpenTelemetry Profiles, the fourth signal. Metrics, logs and traces tell you the box is hot. Profiling is the one that finally says which exact function, in production, right now, is burning the CPU. It landed as a standard this year, not a vendor add-on, and it is coming up on the channel.

That is the week. Next time, what happens the first time your own agent hits a gate you forgot to close?

Allan

PS: If you take one thing from this, automate the action and keep the accountability. That is the whole model.

 

Everything in one place

New here? Start with the free chapter.

Read the free chapter →

More from Metrics & Mayhem

The book  ·  Get Metrics & Mayhem →
Podcast  ·  Listen on Spotify & Apple →
Connect  ·  Allan on LinkedIn →

Cap table management that puts your business first

Managing your cap table doesn’t have to be complex. Pulley simplifies equity management for Founders and CFOs with intuitive workflows, accurate, audit-ready reporting, and predictable pricing—so you can plan and scale without surprises. Onboard in days, not weeks, and rely on responsive, expert support every step of the way.  

From issuing grants to 409A valuations or ASC 718 reporting, Pulley gives you the clarity to manage equity, make decisions, and get back to work. Experience a platform built for business owners and finance teams: transparent, reliable, and designed to put your company first.

Reply

Avatar

or to participate

Keep Reading