April 28–30 in Irving, Texas, our Chairman Doug Haugh took the stage at one of the fuel industry’s most respected gatherings to lay out exactly how operators can move from AI curiosity to AI execution.
SIGMA: America’s Leading Fuel Marketers is the national trade association representing fuel marketers and convenience store chain retailers across the United States and Canada. With approximately 260 corporate members commanding nearly 50% of the petroleum retail market and selling around 75 billion gallons of motor fuel each year, it’s one of the most influential gatherings in the industry. When SIGMA brings people together, the conversations matter.
This past April, our Chairman, Doug Haugh, spoke at the 2026 SIGMA Spring Conference in Irving, Texas. His session covered concrete steps operators can take to put AI to work in daily operations and customer service, with specific tools and starting points rather than predictions about where the technology is headed.
Leadership & culture: getting your team excited, not scared
Haugh opened with the part most AI talks skip: the people. Before any technology decision, a leader has to deal with what the team is actually afraid of. Employees worry AI will take their jobs, that the company’s data isn’t ready for it, or that the whole thing is too complicated to get right. Those worries are legitimate, and his argument was that you answer them directly instead of waving them off.
His framework is a set of six messages he says every leader should deliver to their team before an AI initiative starts:
- AI is a tool, not a replacement for your job. We’re using it to make people more capable, not to automate them out.
- It makes your judgment count for more, not less. The system lays out options; you still make the call.
- Our best people will get the most out of it. The advantage comes from pairing deep domain knowledge with the technology, and that knowledge is already yours.
- This is an investment, not a cost-cutting move. We’re spending to build new capability, not to shrink headcount.
- We expect some experiments to fail. That’s how you find the ones worth keeping, and the trying is what gets rewarded, not just the wins.
- We will govern it with clear policies that protect everyone involved: our customers, our data, and your own careers.
The three classes of AI
One of the more useful frameworks Haugh laid out was a way to sort AI into three classes. The point of sorting it this way is practical: companies that treat “AI” as one undifferentiated thing tend to launch competing initiatives, leave value on the table, and never settle on a strategy.
- Class 1: Embedded AI. The AI already built into systems you own, such as ERP alerts, POS fraud detection, and TMS routing. Most companies are already paying for this and not turning it on. Timeline: weeks.
- Class 2: Point solutions. Focused SaaS tools that handle a single job, such as fuel-pricing engines, loyalty platforms, and demand forecasting. These you evaluate, pilot, and measure on ROI. Timeline: months.
- Class 3: Enterprise AI. An intelligence layer that runs across the whole company: a cross-function data fabric, AI agents working together, a company-wide knowledge base. Timeline: quarters.
The error, Haugh noted, is rarely betting on the wrong class. It’s ignoring the other two. Pick one and neglect the rest and you get shadow AI, vendor lock-in, embedded capability you paid for but never used, and no single vision tying it together.
AI across the full fuel value chain
Haugh walked through where AI actually fits at each link of the fuel value chain, from trading and supply through back-office finance.
- Trading and supply. Price forecasting, automated competitive intelligence, and volume and margin optimizers. Traders still set the strategy; the AI generates execution recommendations across hundreds of terminals around the clock.
- Dispatch and logistics. Predictive tank gauging, dynamic load building, and continuous re-optimization as conditions change. Operators reported a 15–20% drop in miles driven, 40% fewer unplanned runouts, and roughly double the planning capacity per person.
- Marketing and wholesale. AI scores every wholesale account monthly, flags the ones at risk of leaving, and delivers tailored loyalty offers at the point of decision. Operators reported 25–35% better retention among at-risk accounts.
- Retail and c-store. Fuel pricing that reacts to competitor moves within minutes, merchandise decisions driven by POS data, and computer-vision loss prevention that catches drive-offs and internal theft as they happen.
- Back office and finance. AI agents that watch the AP inbox, pull and classify invoices, route them for approval, and post approved vouchers straight into the accounting system. The loop closes without anyone monitoring an inbox by hand.
- HR and workforce. Staffing forecasts down to the store and the hour, which cuts both overtime and understaffing, plus turnover prediction that lets managers step in before they lose someone they need.
Finding your early adopters
Haugh’s advice on getting AI initiatives moving: don’t wait for everyone to be ready. A typical organization splits into about 5% who are eager early adopters, 15% who are open but hesitant, 60% who wait to see what happens, and 20% who are skeptics. The mistake is trying to convince all of them at once. The move is to find the 5% who are already excited, the people using ChatGPT or Copilot on their own and asking why a given task can’t be automated, and put real weight behind them: give them an official title like “AI Champion” or “AI Innovation Lead,” carve out 10–20% of their week for AI work, give them a direct line to IT, and show off their wins in all-hands and leadership briefings. Their job is to pull the 15% along. Once the hesitant group sees results from people they work next to, the case makes itself.
Governance: a competitive advantage, not a constraint
Haugh closed on governance, and his argument was that doing it well is a strategic edge rather than a compliance chore. The numbers make the case: 60% of executives name AI as their top technology risk, 77% of companies are building governance programs, but only 29% have a comprehensive plan in place. Most of the field is moving and very few have arrived, which is exactly where an operator can get ahead. He paired that with the reminder that the technology doesn’t pay off on its own; 95% of generative-AI investments return no measurable ROI, and governance is part of what separates the other 5%.
The cost of skipping it is concrete. Without a plan, employees paste customer contracts, pricing models, and personnel records into public AI tools, and once that data is out you can’t pull it back. Undocumented AI decisions in hiring, credit, or customer treatment open the door to discrimination claims and regulatory action. And unsanctioned tools leave the company running on a patchwork of free consumer AI with no visibility, no control, and no audit trail.
His fix is a three-phase rollout:
- Phase 1, Foundation (months 1–2). Form a cross-functional governance committee across IT, legal, HR, and operations, audit what AI tools are already in use, draft an acceptable-use policy, and put every employee through awareness training.
- Phase 2, Control (months 3–4). Stand up monitoring of AI tool usage and data flows, route all new AI tools through a formal procurement review, build a risk-assessment template for each deployment, and write an incident-response plan.
- Phase 3, Optimization (ongoing). Review policy quarterly as the technology and the rules change, expand the approved-tool catalog as new solutions clear review, track KPIs like shadow-AI decline and incident rate, and align to the NIST AI Risk Management Framework, working toward ISO/IEC 42001 certification as you mature
The window is open
Haugh ended on the stakes: the chance to lead is open now and won’t stay that way, and in this industry the only question is who moves first. SIGMA is where those conversations start, and we were glad to be part of it.