What I’m thinking about: The actual job description for an experienced operator in 2026…and the fellow land investor who showed me exactly what role evolution looks like (we’d never actually had a real conversation before this one, though he’s been reading this newsletter for a while).
His name is Mitch Klein, a water and environmental engineer with 13 years in the field (see his dual RETipster pod appearances on septic/water here and here). He runs his own engineering firm solo, AND runs a land investment business with one full-time VA…while raising 6(!) kids.
(FYI, if you need a civil engineer’s perspective for larger dev projects, site analysis, or AI consulting, with multi-state reach, connect with Mitch here.)
He joined Callan Faulkner’s Accelerate to Automate (A2A) program last fall (Callan’s core 12-week course, designed to upskill business owners in practical AI applications better than any other offering).
~6 months after starting A2A, Mitch has already spoken at a civil engineering conference on applying AI inside firms within that industry (notably, Mitch does not have a tech background, making this jump in capability even more impressive).
Throughout our conversation, I barely talked. I just took notes, occasionally asked clarifying questions, and listened to Mitch walk me through one of the most aggressive AI build-outs I’ve seen from an operator in our space.
And he hadn’t even started using Claude Cowork yet. (More on why that matters in a minute.)
His $250/month receptionist now costs $3
In his land flipping business, Mitch built a custom Retell AI receptionist (Retell is the voice agent platform behind a lot of new AI phone systems) to replace PatLive, the human call answering service many land investors use.
Old cost: ~$250/month.
New cost: ~$3/month.
The system rated itself 7.8 out of 10 on its own first build, with a 54-point improvement plan to get to 9.5. It runs 24/7, collects caller info, and pipes everything into Pebble (his CRM) through Zapier. Background noise is selectable for realism (coffee shop, call center, outdoor).
The broader move was simplification. Mitch went from 10+ tools (RocketPrint, Prycd, DataTree, and others) down to a clean stack: Land Portal for market research and data, Pebble, Launch Control for texts, and his own Retell build.
Mitch’s own phrase for this surgical implementation and consolidation pattern is the ‘AI compounding effect’…each new capability you add reveals the next set of use cases that weren’t visible before.
The 16-hour task he ran in 30 minutes
Mitch told me about a pipe buoyancy calculation spreadsheet (instructions tab, calculator tab, glossary, the works). The traditional build for a working engineer takes ~16 hours, or 2-3 days of pecking away between client work.
He fed Claude the design manual PDF as a knowledge base, specified colors, fields, units, ranges, and figures, and got 85% of the calculation spreadsheet produced in 30 minutes.
He then ran a “man vs. machine” experiment on a separate engineering project, planning to time himself against Claude’s output side by side. He had to abandon the experiment…Claude outpaced him so badly during the research phase of the project that there wasn’t a fair comparison left.
His framing of what happened to his role (the part that stuck with me): his job had shifted from ‘doing’ to ‘managing and QC’ing’, like supervising a junior engineer (QC = quality control).
(Recall, Mitch is a senior civil engineer with 13 years of reps. AI didn’t replace his judgment…it replaced his typing, his research time, and his document drafting. The judgment, the calibration, the “is this actually right?” check is still 100% him. Tech amplifies expertise, never replaces judgment. That’s the durable framing.)
=====
When Mitch needed to build core business documents for insurance (e.g. risk approach, risk registers, A/R (accounts receivable) processes, etc.), he expected weeks of grinding. He built 10 of them in one day.
Same with client proposals. The old workflow took days, sometimes weeks. New workflow: he records the client prospect call on speakerphone with Granola (a meeting transcript tool), drops the transcript into a Claude project loaded with previous proposals, a brand guide, and his website content, and gets back a complete proposal draft, jurisdiction call action plan, and competitive cost analysis in 3-4 hours.
He also pits Claude and ChatGPT against each other for QC. One AI tears apart the other’s draft, iterates, and quality goes up exponentially with no human grinding required.
(Think two junior engineers reviewing each other’s work, with Mitch as the senior engineer making the final call. Same management structure, radically different time and resource cost.)
A Weekly Project Management Office (PMO) Pack that beat his old top-tier firms
Mitch is currently running a complex development project (entitlements, purchase, funding, pro forma, multiple engineering disciplines, builder evaluation, etc.) He built a Weekly PMO Pack (status-tracking deliverable inside engineering/consulting firms) in Claude that generates an action register tracking every action ID, context, contact, next step, and dependency (designed, in his words, “to be ‘turn-your-brain-off’ easy so you don’t make mistakes”).
It includes HTML dashboards for both investor-facing and internal-facing views, a Gantt chart for visual critical path analysis, a change log with an “AI field map” tab so the AI can self-reference the spreadsheet structure when updating it, and auto-generated weekly summaries flagging which assumptions changed and what next week’s priorities are.
His paraphrased direct take: “This change management process is better than what I saw at top-tier engineering firms when I used to work there.”
(To note, a career engineer who’s worked inside the biggest names in the industry just out-built one of their internal systems by himself, using AI he learned to wield in a 12-week course.)
The kicker: he hasn’t touched Cowork yet
Everything I just walked through, Mitch built using standard Claude projects (the chat-based version), with knowledge bases and custom instructions.
Claude Cowork (the desktop agentic version that I now run my entire business through) is just a few months old. It connects directly to local files, Notion, Drive, Gmail, Calendar, GitHub, WordPress, and a stack of other connectors, runs scheduled tasks, and builds full workflows you can call from anywhere.
While I demoed some of our Cowork builds to Mitch (I’ll be presenting some of these on the free 5/5 Masterclass with Callan tomorrow if you want to tune in), he commented that he could see immediate improvements he could make to his current AI projects and workflows (and I have no doubt he’s already done so since we met).
The current A2A program is NOW centered on Cowork. Which means the operators going through this cohort are positioned to ramp up the AI learning curve faster than ever before.
(Remember, Mitch is not unusually technical, and he’s not under-occupied. He has 6 kids, two businesses, and one VA. He decided AI was the leverage point, found the right teacher, and put his head down. As my Coach, Trevor McGregor, routinely says, “education without application is merely entertainment.” Mitch obviously applied himself, and the results speak for themselves.)
In Mitch’s case, saving $3,000 a year on a receptionist is real money. But the role transformation, the major time efficiency gains, the conference talks…those are the compounding wins.
And Mitch is still less than a year into his AI journey. The runway in front of you is the same.
=====
Like Mitch, the operators who learn to build with AI now are documenting their judgment, training systems around their expertise, moving faster than teams 5x their size, and creating operating leverage that compounds every month they keep building.
Love how Section AI put it last weekend:
“Six months ago, the difference between an AI-forward company and a laggard was subtle. You had to squint to see it. Now you don’t. The gap is showing up in throughput, in the scope of what small teams can take on, in how fast decisions get made. And the companies on the wrong side of that gap aren’t falling behind gradually. They’re watching the lead widen every week.”
Exactly that.
My team and I are collapsing projects that would’ve taken a month in the past…into a single day.
Also like Mitch, we wouldn’t be where we are today without Callan and her team, and we leveled up the most when we enrolled in A2A.
The critical distinction is that A2A is not “how to use AI.”
It’s how to become the kind of operator who can build AI employees, AI workflows, and AI-powered business infrastructure around the knowledge already sitting in your head.
Cart closes May 8, and payment plans are available.
