Why 70% of Enterprise AI Seats Sit Idle | Russ Fradin

Russ Fradin (0:00)

I don't think people that are in Silicon Valley and spend their time talking to Silicon Valley startups understands how different Silicon Valley is from the rest of even America. Forget the rest of the world, the rest of even America in terms of actual usage of these tools. The average company has rolled out Microsoft Copilot, and six or seven in 10 of their employees are using it basically zero.

SPEAKER_2 (0:25)

Hello everyone, and welcome back to MTS. Today, we are joined live by Russ Fradin. He is the co-founder and CEO of Larridin, and Larridin helps companies understand how their employees and AI agents are actually using AI, what it costs, and whether it's actually improving business outcomes, which is an interesting conversation that I want to talk to you about. So welcome Russ to the show.

Russ Fradin (0:46)

Thank you. Yeah, I feel like people should care about that. They're spending all this money. They should probably figure out why.

SPEAKER_2 (0:50)

Exactly. You see all these funny graphs of what we're actually spending on cost and then productivity, and there's a huge gap between the two. And so how do you actually measure this conversation about return on investment? How do you actually measure that?

Russ Fradin (1:04)

So I think people have to start purely with the basics, which is if you ask most companies, they're not really sure about usage. Who's using the tools? What are being used? Where are they being used? And how effectively are they being used? So that's number one. We just start with like, what's actually happening? But before we get to anything else, then we try and help companies understand how work actually gets done across the organization. I saw, actually it was a good tweet, I saw it earlier today or I forgot what they're called now, but a tweet earlier today where they were talking about one of the issues with productivity in the enterprises. No one really understood how work got done before AI came along, right? And these large companies, they're kind of weird, kind of, there's a great book that just came out. They're really kind of messy bundles of tasks. So, you know, how would you replace it with an agent? No one's really sure. So the first thing to work on is just what's actually getting done in the org, right? Who's using it? What are they using it for? Where are they using it? Where are they using it? Well, are they getting better at it? Second thing we work on is just what are workflows? And then we do our best with companies to help tie it into various backend systems. So on the coding side, on the sales side, what you want to understand is, you know, are my engineers using the frontier models actually producing more PRs that are more complicated, that make it into production, than my engineers that are using cheaper models? Who are my most productive sellers? And what are the workflows of my most productive sellers?

So there's no number, just like there's no answer to is Jerry productive, right? It's more of a framework for CFOs and CIOs about what's happening, who's using it, where are they using it, what is work, what else should I automate with AI, all things like that.

SPEAKER_2 (2:37)

But how do you actually figure that out? So a little bit of what you're saying is before we even had this conversation around AI and agents as a whole, we didn't actually even have a metric for how you measure productivity in general, right?

Russ Fradin (2:48)

Well, there's a million metrics. There's not a single metric. So look, for a very long time, there's a very large business.

McKinsey has one called the Corporate Health Index. Accenture has one where they'll just send surveys to employees every six months, and fundamentally understand, are you productive? Obviously, that is not the best way to measure any of this, but you need something, right?

SPEAKER_2 (3:08)

Elon has one where it's like you tweet three things you did today.

Russ Fradin (3:11)

That's been around since the 60s. And so what we're trying to really understand is what is the work you actually got done? This is a bunch of tools we have that people deploy.

What is the work that actually got done? And then what's the downstream impact? You're only getting that through integrations with various systems. So on the sales side, it's a bunch of Salesforce and HubSpot and a bunch of the backend tools there, Glean and Gong and things like that. On the engineering side, it's fundamentally GitHub and Cursor and Cloud and Codex. So we really want to understand kind of what's actually happening when we spend money on these tools? Are people increasing the amount of kind of slop they're creating? Or are they actually shipping more code? Are they actually closing more deals? Are they actually responding to e-mails more quickly and need less help from their colleagues and coworkers? So to do that, you can't do that unless you understand how work is getting done in the org. So it's a bunch of tools that are effectively kind of, without paying attention to employee level data, effectively kind of monitoring what's actually happening, how work gets done across the organization.