What the work looks like when it is done properly. The first two are SSTRAIT engagements; the rest are from my executive roles as CIO and senior IT leader. Company names are withheld where I do not have permission to use them.
AI training for 25 consultants and an AI agent for lead management
Situation. A coaching and consulting firm had 25 consultants using AI tools in 25 different ways: some well, some not at all, and nobody was sure what was safe to put into a prompt about a client. At the same time, inbound leads were being handled by hand, inconsistently, and slowly.
My role. SSTRAIT engagement as fractional Chief AI Officer: training, governance, and the build of one working AI agent.
What I did. Ran AI literacy, ethics, and practice training for all 25 consultants, built around their real work rather than generic prompting tips: what the tools do well, where they are convincingly wrong, what client data may and may not go into them, and how to check an answer before acting on it. Then designed and built an AI agent for lead management that qualifies incoming leads, drafts the first response, and routes each lead to the right consultant with the context they need, with a person reviewing before anything is sent.
Results. A consulting team that uses AI consistently and safely, and a lead process that no longer depends on who happens to check the inbox.
What it means for you. The training and the agent together are a small version of what the fractional Chief AI Officer role does at a larger company: make the use of AI consistent, governed, and useful, and prove it with one working automation.
Fall Line Outdoors: from ten people on free email to a 250-member nonprofit
Situation. Fall Line Outdoors, a chapter of the North Carolina Wildlife Federation that helps adults enter conservation through ethical hunting and fishing (including the NC Learn to Hunt program), was about ten people running events from personal Gmail accounts and spreadsheets. No shared data, no data protection, no repeatable process, and every event depended on one volunteer’s memory.
My role. Joined as a student in 2022 and have worked my way through apprentice, mentor in training, to mentor, and also as a leader of the chapter since. Unpaid, which makes it the purest test of whether a process works: nobody has to follow it.
What I did. Won a Microsoft nonprofit grant and moved the chapter onto Microsoft 365 (Teams, SharePoint, Power Automate) with Zapier and a WordPress publishing pipeline. Built the event lifecycle end to end: registration, volunteer sign-up, background-check reminders, mileage tracking, and editorial automation. Cleaned and deduplicated the people master data. Put proper data protection around member and participant information, and gave the chapter a consistent brand and communications process.
Results. Roughly ten members to about 250. Volunteers follow the process because it makes their evening shorter, not because anyone can make them. The chapter was recognized at a Governor’s Award ceremony.
What it means for you. Adoption is the hard part of every technology change, and it is hardest when nobody is paid to comply. The lessons are in What leading 30 volunteers taught me about leading AI transformation.
Eleven ERPs, one data platform, and the first usable AI forecast
Situation. A specialty materials manufacturer with 16 sites on three continents, built through acquisitions, had 11 separate ERP systems. Every month, someone hand-built a $2 million demand forecast in Excel from exports out of all of them. The spreadsheet was the company’s most important application, and nobody called it that. The board was asking about AI.
My role. First CIO the company had ever had, reporting to the CEO.
What I did. Walked the plants and asked one question everywhere: what takes you the longest every week? The answer was almost always a spreadsheet. Instead of starting with AI, I started with the plumbing: consolidated the fragmented data into a single warehouse, automated the dashboards that were being rebuilt by hand every Monday, and cleaned up master data that had been copied between systems for years. In parallel, ran a cybersecurity program because you cannot share data across sites on a network you do not trust (see below).
Results.
- One data platform in place of 11 disconnected ERP data sets; the monthly hand-built forecast retired.
- Yield variation across production lines, demand signals, and quality correlations became visible for the first time.
- Predictive yield optimization on the production floor, and a sales forecast that no human had been able to estimate before, later produced by AI at 80% accuracy.
What it means for you. The highest-return AI projects in manufacturing rarely come from the boardroom; they come from the middle of the organization, and they usually need a data step first. This is exactly what the AI Opportunity Assessment is built to find. Original article: The $2M spreadsheet.
Raising a NIST cybersecurity score by 150+ points in nine months
Situation. The same 16-site manufacturer. Security had grown up site by site with each acquisition. Insurers, customers, and the data-consolidation program all needed a level of control the company did not have, and there was no shared baseline to measure against.
My role. CIO, accountable for the program.
What I did. Assessed every site against the NIST Cybersecurity Framework to get one honest number. Prioritized the controls that unlocked the business goals (site-to-site data sharing and the data platform) rather than working down a checklist. Reported the score to the executive team monthly so security became a business measure, not an IT one.
Results. NIST score up by more than 150 points in nine months. Just as important: the network could now be trusted to carry the consolidated data that made the AI work above possible and, on the side, also secured DoD contracts.
What it means for you. Security is not a separate project from AI; it is the foundation for it. Security ownership is part of every fractional CIO engagement.
Rescuing an IT outsourcing relationship without changing the vendor
Situation. An organization whose IT outsourcing relationship was, in their words, on fire. Tickets piling up, staff frustrated, leadership ready to terminate the contract and start over with a new vendor, which would have cost months and solved nothing.
My role. Incoming senior IT leader, asked to fix the vendor problem.
What I did. Asked the executive team one question: can you show me the governance structure for this relationship? Silence. No regular business reviews, no escalation framework, no KPIs beyond “keep the lights on,” and nobody on the client side owned the relationship. They had hired a vendor and walked away. We did not replace the vendor. We built a governance framework (monthly business reviews, quarterly strategic check-ins, clear escalation paths), appointed an internal IT liaison who owned the relationship, and aligned service delivery to the organization’s actual business priorities.
Results. Six months later, the same vendor, same team, same contract, was being praised in board meetings. Cost of a vendor change avoided entirely.
What it means for you. Outsourcing usually fails on the client side, in governance, before it fails on the vendor side. Managing the MSP is one of the first things I take over as a fractional CIO. Original article: Why outsourced IT fails.
Twenty-plus acquisitions, integrations, and carve-outs
Situation. Across senior IT roles at Coty, Element Solutions, Kymera International, and Bonduelle, I have been on the technology side of more than twenty deals: acquisitions into a platform, carve-outs from a parent, and being the acquired company.
What I learned. Value leaks in three places: contracts that do not transfer, key people who leave at close, and integration timelines set by people who have never run one. We check each of those in the M&A IT Due Diligence engagement.


