About Dusty

Dusty Gulleson started his company in 1999, in the middle of the dot-com boom, building custom content management systems with a single engineer he flew in from overseas. Twenty-five years later it’s Tectonic (rebranded from eResources in January) — a vertically integrated technology partner with about 80 people in the States and 130 offshore, around $15–16M in annual revenue, grown in large part through five acquisitions since 2015.

What we talked about

Why most AI projects start in the wrong place — how teams rush to build agents before they can describe the problem, or the outcome they actually want.

The conversation clients don’t want to have — why every AI initiative turns into an uncomfortable audit of people, processes, and data governance.

Who signs and who freezes — the split between large clients who want eight initiatives done yesterday and smaller ones who reach the statement of work and stop.

The collapsing org chart — his case that the CTO role, project management, and even the screen itself are on the way out.

What engineering becomes — why coding may be solved but problem-solving isn’t, and what separates a good engineer now.

Building a second brain — the “librarian” agent, the PARA framework, and where Obsidian starts to break down.

Sharpest moments

On why bad AI output usually isn’t the AI’s fault:

AI is the smartest employee you ever had that knows nothing about your organization... if you feed them bad process, you’re gonna get bad output.

On the hard conversation he keeps having with clients:

You’re gonna have drift so fast in your organization, and you’re gonna be like, this AI sucks. I’m like, no, the AI doesn’t suck. You actually suck because you haven’t done the hard work.

On what actually decides the winners:

Being able to talk to someone and understand their situation, be able to empathize with them, those are the people that are gonna win at the end of the day.

On the limits of the tooling:

AI is just a tool. It is not the relationship builder.

On the fear of missing out:

Google was the eighteenth search engine that ever got developed... It’s not about being first all the time. It’s about doing a good product.

On the second brain debate among fellow CEOs:

The second brain is not a crutch. It’s an enablement tool.

Key takeaways

• Most AI initiatives stall on undocumented processes and missing data governance, not on the technology — the first step of any engagement is a look at people and process.
• Larger clients are more willing to have that uncomfortable conversation; smaller ones more often get to the signature and freeze.
• Any industry built on structured, objective data and defined process is exposed — he names marketing, content, gaming, and education.
• He expects account manager, project manager, and strategist to collapse into one role, with the strategist the most valuable person long-term.
• Coding is largely solved; engineering isn’t — the skill shifts to defining the problem and breaking it into small pieces that can be tested one at a time.
• Legal review is one of the fastest wins he’s seen: contract redlining that used to run about $5,000 in fees now costs closer to $400, with a lawyer’s final sign-off.
• Relationships are the moat — the winning pattern is AI that routes people to a human, not AI that replaces one.