Speaking engagements

Most firms don't fail at AI because the tool was wrong. They fail because they skip the unglamorous questions first: is there real appetite to change, what does the data actually look like before a dollar gets spent, what's genuinely safe to touch, and which single problem is worth solving before anything else gets attempted.

This one lays out the sequence that actually works: readiness before tools, data before use cases, one committed pilot before a second one, and a real way to measure whether it worked before scaling. It spends real time on the piece most efforts quietly die on: whether people actually use what gets built. Built as the entry point to everything else on this list.

A wire fraud attempt rarely looks like an attack. It looks like a routine vendor update: same project name, same tone, a note that the payment account changed. Real money moves on tight timelines, across dozens of vendors, in a culture built for speed rather than double-checking.

This session moves through three connected realities: how a real attack unfolds and the verification habit that stops it, how ransomware reaches into schedules and billing long before it reaches a server, and the coverage gaps sitting inside most cyber insurance policies, particularly the low sublimits on funds-transfer fraud nobody reads until they need them. The room leaves with a specific procedure and three questions for their next insurance renewal.

Bid comparisons, quantity takeoffs, schedule slippage, change-order tracking: the daily ground of running a job, and exactly where AI claims get loudest and hardest to check.

No hype, no blanket dismissal. Just a clear read on where the estimate and the schedule actually get faster and more accurate, and where it's still a vendor's slide.

Most teams still lose part of every week re-keying the same numbers: once from the field, once into the project system, once into whatever report leadership actually reads.

This session looks at what changes when field data, billing, and the back office are actually connected instead of hand-stitched together, and what that's worth in hours during a close-out, not just in theory.

Nationally, a large share of AI pilots never show a real return, and it's rarely the AI's fault. It's the data underneath: scattered, duplicated, half-labeled, sitting in systems that don't talk to each other.

The argument here is direct: a single reliable source of truth is the actual precondition for AI working, not a nice-to-have that comes later, and it's the real reason so many pilots quietly stall out.

History and cost data are real leverage in a bonding or lending conversation, but only if you actually control where that data lives and who can act on it. Most organizations don't. It's scattered across a field tool, an accounting system, email, and a spreadsheet someone built three years ago.

This one reframes data ownership as underwriting and strategy instead of an abstract IT concern, in language leadership can bring straight to their surety and bank.

Most teams are already running AI against real operational and financial data before anyone formally decided what's allowed. That's the actual exposure: not that the tool is dangerous, but that adoption is outrunning any governance around it.

Three concrete questions replace the usual vague advice, alongside a real example: a firm that mapped its own tool sprawl, found the same data sitting in five different places, fixed exactly one process, and got roughly 600 hours back in a year.

Acquirers are increasingly checking whether a firm's systems and records are clean enough to actually run AI on, not just whether the books reconcile. Messy data is a cost a buyer inherits at close.

Built for owners who are years, not months, from a sale or an ownership transition, and want to know what's actually being scrutinized.

Industry benchmarking puts technology spend in many trades and project-based businesses at roughly one to two percent of revenue. The average company across the broader economy spends closer to five or six percent.

Most leaders have never seen that gap sitting next to their own number. This session turns the question from "what should we spend?" into "why are we this far below everyone else, and is that catching up with us?"

Skilled labor is still the most cited operational worry across most industries, and most organizations cannot hire their way out of it.

Real numbers anchor this one: a firm that automated a single process and reclaimed roughly 600 hours a year for its team. It closes with the harder half of the conversation: where automation genuinely takes work off people's plates, and where it still takes a person.