Guru

Rick Nucci founded Boomi 26 years ago. Yesterday he was on the Boomi World stage in Chicago, this time as CEO of Guru , announcing a partnership with the company he originally built. He wrote that “full circle doesn’t even cover it.” Standing in that room, I’d have to agree. There was an incredible energy around the announcement. And that energy spilled into the hallways, where I noticed a different story unfolding. Partners who have been with Boomi for decades. Customers who said it changed the trajectory of their business. Employees who came up to tell him that the culture he created all those years ago still holds, that it’s one of the reasons they’re still there. No drama. No ego. Just good people who work hard and play hard, solving real problems. Rick is one of the most humble people I’ve ever worked with. He wasn’t seeking any of it, and seemed a little shocked by it. But watching those conversations happen, one after another, it was something. I’ve worked at Guru under Rick for almost 9 years. Seeing his lasting impact while at Boomi World, I realized I’ve been living a version of that same culture the whole time, the same way of treating people, the same way of building. Rick builds things that last. Products that hold up, evolve, and keep earning loyalty long after launch moments fade. Boomi is evidence of that. So is Guru. And now the two are official partners, making sure that companies with decades of infrastructure behind them can move confidently on AI, with accurate, trusted knowledge powering every agent and every decision. Twenty-six years is a long time to keep anything together. Culture, product, the trust of the people who bet on you. Getting to witness that up close, as someone who’s had a front row seat to him doing it again, was a true privledge. 💚

A recent MIT study found that people who wrote essays using only ChatGPT showed weaker brain activity than those who didn't. But that’s not the most important insight in the study. There was actually a third group with the strongest brain activity, and they didn't actually fall into either category. They thought first, then brought in AI to push their thinking further. They showed enhanced cognitive engagement - more than the brain-only group and significantly more than the ChatGPT-only group. But of course, every headline was some version of "ChatGPT makes your brain lazy." A more accurate headline: outsourcing your thinking makes your brain atrophy. And that was true long before AI existed. I love my GPS. But if you put me more than a mile from home without it, I'd have to start a new life there. I've outsourced my sense of direction to a device, and I know it. The GPS wasn’t the problem, but how I chose to use it was. The same is true for AI. 80% of writers in the ChatGPT-only group couldn't accurately quote a single line from the essay they had just "written." That’s a legitimate problem. But the opposite use case is just as real. The most underrated use of AI is pressure-testing. You bring a fully formed view and AI plays devil's advocate. It surfaces the counter-argument, finds the blind spot you didn't see, and pushes back. Building a fully formed view is where the critical thought muscle gets developed. The collaboration with AI often strengthens it further. And for learning, the leverage is even bigger. A motivated person using AI to accelerate learning can build domain knowledge in weeks that used to take years. The feedback loop that used to require a mentor or expensive mistakes now happens instantly. The research is a reminder that the thinking you do before reaching for AI is more important than ever. Here’s the full study: https://lnkd.in/eW9XfrGk

26 years ago, I founded Boomi to connect enterprise systems. After we were acquired by Dell, I co-founded Guru to make sure companies could actually trust the knowledge powering their teams. Today, Guru and Boomi are officially partnering. Full circle doesn't even cover it. Here's the problem we're solving together. A lot of companies are bought in on AI transformation and ready to go. But they hit a wall because their critical systems were built long before technologies like MCP existed. Those systems aren't going anywhere, and right now, agents can't see them or orchestrate them. Boomi Connect changes this. With 1,000+ connectors, customers can build MCP solutions on top of whatever systems they already have. So when a Guru Knowledge Agent needs to orchestrate actions with a CRM, an ERP, or a legacy database, it can read the data, understand what it means, and execute the right sequence of calls to act on it. Companies with decades of infrastructure behind them can now move quickly and confidently on AI without ripping and replacing everything. This partnership is 26 years in the making, and Guru Knowledge Agents are officially available now in the Boomi Marketplace.

I wonder if this family knows their PB&J tutorial is a masterclass in AI prompting. We showed this video at a recent internal AI enablement session and told the team “you’re the kids, the agent is the dad.” Essentially the dad asks his kids to write down instructions for making a peanut butter and jelly sandwich. Then he follows them exactly as written. "Put the peanut butter on the bread." -> he puts the jar on the bread. "Push the pieces together." -> He smashes them. Every step, he does exactly what they wrote, but the result is chaos. The kids didn't write bad instructions necessarily, they just wrote instructions full of assumptions. This is exactly how agents work. An agent will do exactly what you say, in all of the good and bad ways. It won't fill in the gaps. It won't guess what you meant. It doesn't know that "put the peanut butter on the bread" means open the jar, grab a knife, and spread it. Even then, how much do you get? How do you spread it? The more you think of all the potential gaps, the more you’ll find. In this case, the worst possible outcome is a bad PB&J (which is brutal, don’t get me wrong). But with AI, the stakes are a lot higher. Unclear instructions compromise the quality of every response your agent delivers.

Every few weeks, there's a headline like this that makes you go: wait, what? Except this one is from 2019, talking about GPT2. It was a big step forward, but in hindsight, I think we can all agree these claims were a dash overstated. It's pretty obvious what was going on here. GPT2 was a real and meaningful advancement in the field. The "it's going to end humanity" framing was an excellent marketing tactic. Both were true at the same time. And the same thing is happening now. Claude Mythos just got announced, and we're already hearing it's being withheld from the public for safety reasons. It's worth keeping that in mind next time a new model drops and the discourse spins up. There is transformative stuff happening in this space. There are also incentive structures that run deep. When your metric is attention, you're going to make the most daunting, scroll-stopping claim you can. That's why the discernment filter has to be turned up really high right now. We can get excited about the technology and still ask what someone is trying to get out of the claim they're making. We don't need to end the world every time a new model comes out. If you're mid-AI panic and want a bit of balance, looking in the past can frame some of what's happening. This article is a good place to start: https://lnkd.in/eFy9eGky

At Guru , there's a 0% chance AI replaces you. We've said that to the team since we began internal AI training 2 years ago. But that promise comes with an accountability contract. Upskilling with AI is what we call a mandatory gift. Because it's not just a requirement here. It's where all knowledge work is going, no matter which company you work for. Our VP of engineering made this point the other day: If you're an engineer, you'll find jobs where people are still hand-coding. They exist, but they're shrinking. The jobs going up are the ones where people know how to delegate work to agents effectively. The same will be true for every team. So the accountability contract looks like this: we provide the resources, the training, all of it. The team commits to leveraging the technology to change how they work. In practice, that means everyone is a manager. If engineering ships an agentic path that introduces a bug, the team owns it. "AI made this" isn't an answer. The accountability lies with the people who told the agent what to do. The job has changed. But career growth hasn't.

Faire took their accuracy score from 76% to 96% with one automation on one Guru Knowledge Agent. Here's how: Faire wanted to go deeper with AI across their CX org. Chat, Slack auto-answers, more agents. But none of it could move forward until the team trusted the knowledge those agents were pulling from. And trust was stuck. Carly Byrge , their Ops Enablement lead, inherited a knowledge base with a massive backlog of unverified content. Their Trust Score (a measurement within Guru of the recency, frequency, and current state of knowledge accuracy across a knowledge base) was in the 40s and 50s. She got it into the mid-70s by manually reviewing and fixing what she could, but that was the ceiling, and she was burning 6 hours a week to hold the line there. That's the loop most teams get stuck in: AI adoption stalls because the knowledge isn't accurate, and getting it accurate is manual, endless work. Guru's Knowledge Agent quality features were the missing link. Carly started with the one agent that served their broadest CX audience, turned on the automation, and got these results: → Trust Score: 76% → 96% → Verification time: 6 hours/week → under 1 → Carly went from full-time content maintenance to actively scoping new Knowledge Agents and AI use cases across the org Her words: "We're getting people who don't even have stakes in it to see the value." Senior leaders who'd never really used the platform started asking about it." Full case study is live on our site.

The internal knowledge that makes your company valuable is the stuff you'll never make public. That's your moat. And LLMs have no access to it. At its core, your company is your team's collective know-how. How you've learned to build products. Your differentiated product strategy. The insight you have about a customer or a market that nobody else has. How you deliver great customer experiences at scale. These are the differentiators that you would never share. But LLMs operate on what's public. They're intentionally predictable. That's what makes them powerful. They can pull from the aggregate and combine topics in ways that humans can't. But the more you defer to them on the things you're actually great at, the more you start to look like everyone else. Harvard Business Review found that Gen AI productivity gains aren't actually translating to competitive advantage, because every company has access to the same models. If your company is genuinely great at customer service, and you outsource it to AI, you become the lowest common denominator. M&A buyers are paying attention to this, too. 85% say commoditization and loss of differentiation are the biggest long-term risks AI poses to companies. When they're worried about AI making your company vanilla, it's worth paying attention to. The moat was never the tech stack alone. It's always been what your people know and have figured out that nobody else has. That spark of ingenuity, the thing that's distinctly yours, that still has to come from inside the organization.

Yesterday, I ran a webinar on the new Knowledge Agent capabilities we just launched at Guru . The discourse in the chat gave me a real-time read on where people are at with AI adoption. One person wrote, "In a single thread, Claude processes our IT tickets, scans our IT Guru Collection, authors and drafts knowledge gap-filling cards, and 2 clicks later in Guru the gap is plugged." Then a minute later, someone was asking how to get their company to adopt any of this when they have a long-standing IT department that doesn't work this way and leadership doesn't want to deviate from existing process. My answer was the same thing I'd say to anyone trying to get an AI initiative off the ground inside a team that isn't already bought in: forget Guru for a second. What pain are you trying to solve? Because if the IT team isn't drowning in shoulder taps and a backlog of tickets they're processing manually, they don't have a pain. There's no opening to introduce anything new. But if they are, that's the door open. That's where you can say, here's a knowledge agent that can take some of this off your plate. Thank you to everyone who participated and to our innovation partners. Looking forward to the next one.