Logicata

25/09/2026

Happy Friday! Check out Karl Robinson's Newsroundup.

Link in the comments below👇 to read and sign up for the newsletter

22/09/2026

In Season 5, Episode 32 of LogiCast, Kate Gawron made a useful point about AI governance in DevOps.

Giving AI access to infrastructure is useful, but it also raises an obvious question.

How much control should it actually have?

The concern is not about AI helping with investigations. It is what happens when an agent starts making changes on its own.

A server at 100% CPU might look like a problem to an agent.

To a human, it might be doing exactly what it is supposed to do.
That is why governance matters.

Read-only investigation is one thing.

Deleting resources, restarting servers or making production changes without human context is something else entirely.

For now, keeping a human in the loop still feels like the sensible line.

Full episode link in the comments.

16/09/2026

In Season 5, Episode 31 of LogiCast, Charity Francis made a useful point about retention policies.

A lot of the difficulty is not storing the data. It is knowing how long to keep it, especially around things like audits.

Variable retention helps take some of that guesswork away.

The audit happens, the hold starts, and once the required period has passed, the object can be archived, expired or deleted based on the policy.

It is a small change, but it can make life easier for the people who would otherwise have to work all of that out manually.

Full episode link in the comments.

10/09/2026

In Season 5 Episode 30 of LogiCast, Brian Tarbox made a sharp point about MCP security risks.

A simple MCP server can look useful today, then become risky later if trust is assumed without review.

It is not that MCP is bad. It is that connecting agents to tools, data and systems changes the risk profile very quickly.

As Brian put it, MCP can feel like taking the guardrails off a circular saw. Easier to use, right up until it is not.

Maybe what AI agents need next is a better “pushing stick”.

Question: how are you thinking about trust and safety when adding MCP servers to your AI workflows?

Full episode link in the comments.

04/09/2026

Happy Friday! Check out Karl Robinson's AWS Weekly News Roundup

https://www.linkedin.com/pulse/aws-weekly-news-roundup-issue-302-karl-robinson-hpmkf/?published=t

�Link in the comments below sign up for the newsletter

�* *

25/08/2026

In Season 5 Episode 28 of LogiCast, Jéssica Cristina Coelho shared a practical example of how AI agents are starting to support data teams.

For database troubleshooting, agents can help analyse logs, internal tables and queries to identify what may be causing CPU alerts or slow performance.

That might mean spotting a query issue, suggesting an index, or pointing the team towards the likely source of the problem.

The real value is not replacing the engineer.
It is helping them spend less time finding the issue, and more time fixing it.

Question: where could AI agents remove the most investigation time in your team?

21/08/2026