Aug 27, 2026 · 25 min · 10 segments
This GO AI episode features a conversation with John Capobianco, Head of Artificial Intelligence and DevRel at Itential, specializes in agentic ops, when AI agents start making decisions across an…
John CapobiancoGuestJennaHostConnorHostWell, and so, John, you you've identified three major areas, I sense, of effective agentic ops governance, role based access, control and reuse.
How do you define each of those and what's the significance of each?

So I think it's some of the non sexy, flashy stuff about AI, but that's what gets you from a proof of concept or an MVP into production.

They need to have the role based access control into what tools and what systems they have.

When did it reason? When did it call a tool? What was its reasoning? It's not just a black box.

And I think that in terms of a platform, right i think everyone understands a platform where we can centralize these things so that we avoid agent sprawl you know everyone has their own python script john's python script working version two underscore three right jenna's copy of john's working underscore v3 and now we have two copies of this agent connor gets a copy now we have three different agents all with drifting from apart drifting from intent who's got security who's got control We don't want to do this.

There's some real key lessons to be learned that need to be brought forward before we let agents loose on our network, Jenna.

The very first thing I do is give them read access and say, go make sure all my documentation is up to date.
Right? Like it's not a very

Can we give it the right context? Can we give it guardrails and permissions and have it earn trust? And over time, maybe it's over a year, that agent goes from reading things to writing things and changing things, right?
Well, and so, John, you you've identified three major areas, I sense, of effective agentic ops governance, role based access, control and reuse.
How do you define each of those and what's the significance of each?

So I think it's some of the non sexy, flashy stuff about AI, but that's what gets you from a proof of concept or an MVP into production.

They need to have the role based access control into what tools and what systems they have.

When did it reason? When did it call a tool? What was its reasoning? It's not just a black box.

And I think that in terms of a platform, right i think everyone understands a platform where we can centralize these things so that we avoid agent sprawl you know everyone has their own python script john's python script working version two underscore three right jenna's copy of john's working underscore v3 and now we have two copies of this agent connor gets a copy now we have three different agents all with drifting from apart drifting from intent who's got security who's got control We don't want to do this.

There's some real key lessons to be learned that need to be brought forward before we let agents loose on our network, Jenna.

The very first thing I do is give them read access and say, go make sure all my documentation is up to date.
Right? Like it's not a very

Can we give it the right context? Can we give it guardrails and permissions and have it earn trust? And over time, maybe it's over a year, that agent goes from reading things to writing things and changing things, right?
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.