See The New Fuel Live Recording
BOSTON, September 23, 2026 — The next finance hire companies want isn’t someone faster in Excel. It’s a “finance engineer,” someone who builds the AI tools that handle the routine work. That was the message from Alyona Mysko, founder and CEO of Fuelfinance, to Boston University students on September 23.
Her guest lecture, The Finance Engineer: The Future of Finance in the AI Era, covered three things: what AI already does in finance, where it still breaks, and which skills will matter over the next five years. Her description of the role was simple. FP&A managers, controllers, heads of finance, CFOs, and accountants are now building software. “They didn’t plan to become engineers,” she said. “They vibe-coded so hard they accidentally became ones.”
She opened by asking for a show of hands. Who plans to work in finance? Keep your hand up if someone has told you AI is coming for that job. Keep it up if anyone has told you what to do about it. Almost nobody could answer yes to the last question, and the rest of the lecture was her answer to it.
Alyona started with a quick history of finance tools. Paper ledgers came after the 1929 crash, IBM mainframes in the 1960s, Excel in 1985, ERP systems in the 1990s, and cloud software in the 2000s. ChatGPT arrived in 2022, and AI agents that work in real systems came soon after. “The spreadsheet ran finance for 40 years,” she said. “The next shift took two.”
To show what that means for one person’s job, she compared a head of finance at a $15 million company in 2020 with the same role in 2026. In 2020, the job was three-statement models, the month-end close, and board decks built in PowerPoint. In 2026, that person evaluates and deploys AI tools, vibe-codes dashboards and automations, builds agents for close, reconciliation, and forecasting, and decides what AI is allowed to touch in the books.
She said the finance org chart keeps its shape. The CFO still sits above FP&A, the controller, and finance operations. The new part is a finance engineer layer that cuts across all of them and builds the close automation, forecast agents, reporting dashboards, and data pipelines the other teams depend on.
Halfway through, Alyona built a tool live. She gave an AI model messy exports from a contracts system, a billing platform and QuickBooks, along with a single prompt: “Build an executive dashboard.” Within minutes, a working revenue reconciliation dashboard was on screen. It matched signed contracts against billing and recognized revenue, and it flagged the months where the three didn’t agree.
According to Alyona, the best CFOs she talks to have already automated about 90% of their work, including the monthly close, reporting, cash reconciliation, accounts receivable, board reports and forecasts. She was just as direct about where AI fails. “It’s not the model,” she said. “It’s the two layers under it.”
The first layer is data. “You can’t automate chaos,” she said. Messy data just means AI does the wrong things faster. Anyone who has closed a set of books would recognize her examples: payroll recorded under four different names in the general ledger, $3 million of intercompany transactions that were never eliminated, and a February bonus run that never synced.
The second layer is context. The rules for how a business works usually live in the finance team’s heads, where no agent can read them. Fuelfinance writes that knowledge down as 17 context files. Each file covers one topic and has an owner and a date. She recommends starting with six of them: company, business rules, data sources, metrics, chart of accounts and accounting policy. To check whether your context is good enough, she suggested one test: could a new hire answer a client’s question using only the files?
She also argued that every finance team now needs a GitHub repository. That gives every skill, rule, and context template one home, lets the team test changes before they touch the books, and keeps a full history of who changed what and why.
Alyona said she hears that question on every CFO call. Her answer is that companies aren’t paying for someone to compete with the agents. “They’re paying for someone who directs them.” The agents do the matching. People decide what gets matched, where the workflow stops, and who checks the result.
She pointed to the goals of AI-native finance teams, citing OpenAI CFO Sarah Friar on what her own team is working toward. One goal is a zero-day close, where the books close on the day the month ends. The other is a forecast that updates itself as new actuals, contracts, and hires come in.
To make the shift concrete for students, she sorted every finance job into three piles.
“The old path was three years in Pile 1,” she said. “That rung is gone. Pile 2 is the new ladder.”
The demand is already showing up on job boards. Alyona said search interest in “finance engineer” rose about 700% in 2026 compared with 2025. Postings went from a couple in June to hundreds by late July. The same job appears under at least seven titles, including Agentic Finance Engineer, FP&A Data Engineer and Finance Automation Manager.
She shared senior-level base salary ranges from live postings. Every role reports into finance, not IT:
After reviewing 100 finance engineer postings, she found five skills that get candidates hired:
“Finance is a portfolio job now,” she said. “Bring a URL, not a resume.”
She gave students a four-part plan to start this week. First, keep learning finance properly, because you can’t check an agent’s work if you don’t understand the work yourself. Second, try the tools on your own: take raw company data and build a monthly close, a context file, and an agent for one task. Third, get comfortable with Claude Code, Codex, Python, and the terminal. Fourth, share what you build.
Alyona Mysko closed with a story about her own team. Fuelfinance’s youngest financial analyst became a finance engineer faster than anyone else on staff. She built agents for all of her tasks, including one that sends customer updates in Slack. “Experience isn’t the head start,” she said. “Building is.”
In a follow-up note to attendees, she told students not to wait for a finance job before they start. Her advice was to pick one real task, build a tool that does it, and keep the work somewhere they can show it.
Fuelfinance builds the agentic platform for finance. Fuel AI agents run a finance team’s recurring work, including close and controlling, AR and AP, analysis, forecasting and reporting, as scheduled workflows. The customer’s finance team approves every decision that changes the books. Our finance engineers, many of them former CFOs, build the first workflows with each customer on the customer’s own data. Fuelfinance works with founders who have outgrown spreadsheets and with CFOs whose teams are still spending days on work agents can run. Book a demo today.
Just imagine how that would transform your team’s productivity and focus? Talk to our financial experts to know more.
.webp)
