Julio Martínez, Co-Founder and CEO of Abacum – Interview Series

Julio Martínez, Co-founder and CEO at Abacum, is a fintech entrepreneur with nearly two decades of experience spanning investment banking, corporate development, venture building, and technology leadership across major global financial centers. He began his career in finance roles covering capital markets, M&A, and private equity before pivoting into fintech, where he helped launch and scale multiple digital financial products and platforms. Prior to Abacum, he co-founded and scaled Banco Sabadell’s corporate venturing arm, leading product launches, strategic investments, and acquisitions across Europe, the Americas, and Asia. Today, as CEO of Abacum, he applies deep operational finance expertise to building tools that modernize how finance teams plan, forecast, and drive business performance.

Abacum is an AI-native financial planning and analysis (FP&A) platform designed to help mid-market finance teams simplify and modernize planning, forecasting, reporting, and budgeting by connecting operational and financial data with collaborative workflows and automated insights. Built to replace manual spreadsheet-driven processes, the platform centralizes real-time data, supports advanced scenario modelling and multi-dimensional financial modelling, automates repetitive tasks such as reporting and forecast updates, and integrates with hundreds of systems to give teams a single source of truth. Abacum’s features drive accuracy, efficiency, and strategic decision making, enabling finance organizations to reduce manual work, accelerate planning cycles, and focus on growth-oriented insights

You spent nearly two decades working in finance and fintech before founding Abacum. What specific frustrations with how finance teams planned, forecasted, and reported performance made you and Jorge realize the tools you were using were not fit for purpose and that you needed to build the product you wished you had yourselves?

I realized that finance was losing influence not because the analysis was wrong, but because it arrived too late. Across banking, fintech, and high-growth startups, I kept running into the same moment. In an exec meeting, someone would ask a reasonable question like: “How many months of runway do we really have if we slow hiring?” or “What happens if revenue slips next quarter?” And could not answer the room in real time.

It wasn’t because I didn’t understand the business or because the math was hard. The problem was structural. Cash lived in one system, headcount in another, revenue somewhere else, and expenses in spreadsheets. To answer confidently, you had to pull everything together, rebuild the model, reconcile discrepancies, and hope nothing broke.

By the time I could return with an answer, the decision window had closed. That was the real problem. Finance earns its seat at the table through rigor but keeps its seat through timing. If you cannot show up with confident decision support in minutes or hours, you lose influence, even if your analysis is perfect a week later.

What made it worse was the false choice finance teams were given. They were either using spreadsheets that were flexible and fast, but fragile and ungoverned. Or legacy platforms that were powerful but assumed a static business and required heavy administration just to function.

Meanwhile, modern companies operate in sprints, even at the exec level. Plans shift constantly. Decisions stack up. Finance cannot afford to be the team that is always “coming back with the answer.”

That is why we founded Abacum. We wanted a planning system built for speed and trust at the same time, so finance can apply rigor early enough to shape direction while choices are still negotiable.

When you started building Abacum in 2020, how did you validate that this problem extended far beyond your own experience and was shared by fast-growing mid-market companies?

The first validation came in a conversation I expected to disprove my thesis. I called my co-founder Jorge because he was the smartest finance person I knew, and I assumed he would tell me there was a better way I had somehow missed.

Instead, we compared notes for hours and realized we had lived the same pattern in different environments. Finance teams are drowning in reconciliation, constantly rebuilding models, and always one step behind the business. That was the moment it clicked for us. This was not a personal failure or a process issue. It was a structural problem shared across companies.

We then spoke to CFOs and finance leaders across industries, geographies, and growth stages. The language changed, but the story did not. “We are always rebuilding instead of advising.”

The deeper insight for us was that this failure repeats in cycles. Every few years, a new platform claims to have solved FP&A. Then the pace of business accelerates again. New tools, new metrics, new stakeholders, new planning cadences. The system bloats and breaks under change.

That realization shaped our direction. We did not want to build a static solution for a single operating model. We wanted a platform that would remain relevant as the business evolves, which becomes even more critical in the AI era. Being admitted to YC later reinforced that this was a global problem, not a niche one.

Abacum now supports real-time forecasting, scenario modeling, and headcount planning. At what point did artificial intelligence move from a future concept to a foundational part of the platform’s architecture?

AI was never an afterthought for us, but we were very deliberate about when and how to apply it. Finance is a trust business. You cannot put intelligence on top of chaos and expect credibility. If the data is messy, definitions are inconsistent, and the model is fragile, AI will not fix it. It will simply scale the confusion faster.

So, we started with fundamentals: a strong data layer, reliable integrations, and modelling primitives that reflect how businesses actually work. From day one, the AI strategy was to embed intelligence where it creates real leverage.

That meant applying AI to high volume, low judgment work that historically consumes time and creates errors. d. Cleaning and normalizing incoming data. Reconciling mismatches across systems. Classifying and tagging at scale. Surfacing anomalies early, not at month end.

Once that foundation is in place, AI changes the economics of planning. Scenario exploration becomes economical. Trade-offs can be tested in the moment instead of scheduled for a follow-up meeting days later.

That is when AI becomes foundational. Not when it can generate a nice chart or a summary, but when it allows finance to apply rigor fast enough to influence a decision while it is still open.

In fast-growing companies, finance data often lives across many systems and updates constantly. What were the hardest technical or organizational challenges in turning that fragmented data into a reliable, real-time planning system?

Moving data is not the hardest problem. Most modern systems are API-driven, and integration tooling exists. The real challenge begins after the data arrives. Raw data is inconsistent. Revenue in CRM does not match revenue in ERP. Headcount in HR systems does not match payroll. Even basic metrics like ARR or runway mean different things to different teams.

If definitions are not explicit and agreed upon, every forecast turns into a debate. The challenge is threefold. First, a data foundation that updates continuously as source systems change. Second, a governed business model where calculations are explicit and consistent Third, organizational alignment, because real-time planning requires shared definitions and a shared operating rhythm.

Governance is what makes speed safe. Real-time planning only works when trust is built into the workflow, not bolted on as a policy document or a spreadsheet checklist.

From what you’re seeing across hundreds of customers, how is continuous, real-time planning changing how CFOs make decisions week to week, not just at board or budget time?

The role of finance has shifted from periodic review to continuous decision support.

First, the volume of decisions has exploded. Finance is now involved in hiring, pricing, GTM investments, renewals, product bets, and operational trade-offs on an ongoing basis.

Second, data is never “done.” New tools, new metrics, and new stakeholders mean the dataset is always in motion. The business cannot wait for a perfect close to move forward.

In this context real-time planning changes finance from reporting and explaining to actively shaping direction. Runway becomes a living constraint, not a quarterly metric. Scenario planning becomes a frequent trade-off conversation rather than an annual exercise.

The best finance teams are not becoming less rigorous. They are becoming more rigorous earlier. That is the shift.

How does Abacum apply AI differently from traditional rule-based automation, and which financial decisions still require strong human judgment?

Most AI in finance today starts at the end of the workflow. It assumes your data is already clean and governed, then adds a chatbot to query it or summarize insights. That can be helpful, but it skips the hardest part of FP&A.

We start at the beginning. We apply AI where humans add the least value and make the most errors, such as cleaning, reconciliation, classification, anomaly detection, and assisting with model logic. Intelligence lives inside the workflow, not in a separate chat interface.

AI also reduces the complexity tax that holds teams back. Many platforms require specialized consultants or experts, creating a dependency on “system owners” AI should lower that barrier. Finance teams should be able to express intent and have the system help construct the logic correctly.

This is also where our middle path standpoint matters. Historically, finance teams had to choose between tools that were flexible but fragile or platforms that were powerful but heavy to manage. AI is now forcing the same false trade-off: copilots that are easy but shallow, or orchestration systems that are powerful but also require you to learn a new way of working. Whereas we believe the right answer is AI that disappears into the workflow, improving planning without changing how teams operate.

As for judgment, the boundary is clear. AI can accelerate analysis and exploration, but decisions involving capital allocation, hiring trade-offs, pricing, and strategic prioritization still require human context and accountability. The CFO owns the call.

As models become more predictive, how do you think about trust and explainability for finance leaders who need to stand behind the numbers?

In finance, “directionally correct” is not good enough. Finance leaders are accountable for the numbers they present. If you cannot explain a forecast, you cannot use it in a decision conversation.

Trust starts with a deterministic foundation. Consistent definitions. Reconciled data. Transparent logic. Predictive intelligence only works when it is built on something solid.

Explainability is what turns insight into action. CFOs need to answer quickly what changed, why it changed, which drivers moved, and what assumptions are responsible for different outcomes.

Governance cannot live in static controls anymore. It must be embedded in the workflow, so assumptions are visible, logic is traceable, and every scenario leaves a clear record. The goal is not to remove humans from the loop but to help them exercise judgment earlier, with more confidence.

You’ve gone through both early acceleration and later-stage growth funding. How did those phases influence how aggressively you invested in AI versus core product fundamentals?

Early funding forced discipline. We could not chase shiny objects. We had to earn trust by building the fundamentals: reliable integrations, strong data models, and a planning engine that did not break when the business changed.

AI was always part of the strategy, but we refused to treat it like a marketing layer. If AI did not create real leverage inside the workflow, we did not ship it.

As we grew, the market shifted. AI became table stakes. Every vendor could demo a chatbot and be “AI-powered.” So, the bar had moved from optics to outcomes.  Does AI help finance make better decisions faster, with traceability, or does it just produce impressive-looking outputs?

Later-stage growth raised the efficiency standard, too. Teams were expected to do more with less. That reinforced our focus on AI that delivers measurable leverage, not narrative appeal.

You’ve been doubling down on U.S. expansion. How does the U.S. market differ in its readiness to adopt AI-native finance platforms compared to other regions?

U.S. companies move fast, and investor expectations are high. CFOs are expected to be deeply operational, not just accurate. They are continuously guiding hiring plans, GTM investments, spend decisions, and prioritization.

That makes the pain of slow planning more acute. When decisions happen weekly or daily, finance cannot afford to operate on a monthly rhythm. Embedded intelligence becomes less of a nice-to-have and more of a requirement.

The U.S. market is also more receptive to the idea that finance systems should be dynamic, not static. The expectation is not just reporting accuracy, but decision support at the pace the business needs to move.

Looking ahead to 2026, which parts of financial planning do you believe will become largely automated by AI, and where will human judgment remain essential?

The layers that will become largely automated are the repetitive, low judgment tasks that consume disproportionate time today. Data consolidation, cleaning, normalization, reconciliation, anomaly detection, and baseline reporting should run continuously.

Forecasting and scenario generation will be accelerated dramatically, but they will not be fully delegated. AI will make it cheap and explore options and stress test assumptions, but context risk, and accountability still matter.

Human judgment will remain essential wherever stakes are high. Capital allocation. Hiring strategy. Pricing decisions. Board narratives. AI changes whether finance can keep up with the pace of decisions. It does not change who is responsible for the outcome.

Thank you for the great interview, readers who wish to learn more should visit Abacum.