Capability Gets You a Demo. Trust Gets You Deployment.

At a glance

• Who: Oscar Casanova, Founder of Eververdant Growth Partners
• Firm focus: Strategic finance, value creation, and transactions for founder led and middle market companies
• Background: Goldman Sachs investment banking, then strategic finance leadership at private equity backed and public companies
• Why Nera: Every time an advisor puts client data into an AI tool, they make a risk decision on that client's behalf. Nera keeps the data encrypted even while the AI analyzes it, so the model never sees a client's numbers.

Background and experience with AI

Can you introduce yourself, your background, and what you do at Eververdant Growth Partners?

I am the Founder of Eververdant Growth Partners, where I advise founder led and middle market companies on strategic finance, value creation, and transactions.

Before Eververdant, I worked in the investment banking division at Goldman Sachs, where I advised on more than $30 billion of transactions, and later held strategic finance leadership roles across private equity backed and public companies.

At Eververdant, our work sits at the intersection of finance and strategy. We help management teams understand the economic engine of their business, allocate capital intelligently, improve cash flow and profitability, and make better decisions around growth, investment, and ultimately enterprise value.

You've built a career advising finance leaders. What drew you to the intersection of AI and finance specifically?

Finance is fundamentally an information and decision making function.

Historically, highly trained finance professionals have spent an enormous amount of time gathering information, cleaning data, reconciling systems, building analyses, and preparing materials before they ever get to the highest value part of the job: judgment.

AI has the potential to radically change that equation. The opportunity I see is not to replace the CFO. It is to dramatically increase the amount of time a CFO can spend thinking, advising, challenging assumptions, and making decisions. If AI can compress ten hours of mechanical work into one or two, the real economic value is not simply cost savings. It is increasing the amount of senior financial judgment available to the business.

You've talked about how differently in-house CFOs and fractional advisors think about handling client data in AI tools. Can you explain that distinction?

The distinction is fiduciary in nature. An in-house CFO is handling the financial information of one organization. A fractional CFO or advisory firm may be handling highly sensitive information across ten, twenty, or fifty different companies. That creates a fundamentally different risk profile.

When I upload my own company's information into a system, I am making a risk decision on behalf of my own organization. When I upload a client's information, I am making that decision on someone else's behalf. That requires a much higher standard.

For advisory firms, security, data segregation, confidentiality, retention policies, model training policies, and access controls are not technical footnotes. They are part of the trust relationship with the client.

"When I upload a client's information, I am making that decision on someone else's behalf."

Oscar Casanova, Eververdant Growth Partners

Trust and AI

You contributed to the Rillion AI in Finance report. What did that research tell you about where the industry stands on AI trust, and why does trust, more than raw capability, determine whether finance leaders actually adopt AI?

I contributed specifically to the Trust pillar of Rillion's research, which incorporated perspectives from more than 250 CFOs and finance leaders. My biggest takeaway was the tension between enthusiasm and institutional readiness. Finance leaders clearly understand the potential of AI. The harder question is whether they trust the infrastructure enough to place sensitive financial information inside it.

Capability gets you a demo. Trust gets you deployment. A CFO can be incredibly impressed by what a model can do and still decide that the risk of exposing sensitive information outweighs the productivity benefit, especially while questions around confidentiality, governance, accuracy, or data handling remain unresolved.

Finance deals with payroll, customer economics, pricing, M&A, forecasts, board materials, liquidity, banking relationships, and sometimes information that could materially affect a transaction or valuation. The standard cannot simply be, "Does the AI work?" It has to be, "Can I responsibly use this system with information that my CEO, board, investors, employees, and clients expect me to protect?" That is a much higher bar. In finance, adoption ultimately moves at the speed of trust.

"In finance, adoption ultimately moves at the speed of trust."

Oscar Casanova, Eververdant Growth Partners

What does "trustworthy AI" actually look like in practice for someone handling sensitive client financials?

To me, trustworthy AI starts with architecture. I want to understand what happens to the data from the moment it enters the system. Who can access it? Is it retained? Is it used for model training? Can it be reconstructed? How is it encrypted? How are different clients segregated? What happens when information moves between systems or agents?

The second component is auditability. Finance professionals need to understand where an answer came from. If AI produces an output that informs a capital allocation decision, forecast, or transaction analysis, I need the ability to interrogate the assumptions behind it.

The third is human accountability. AI can accelerate analysis tremendously, but in sophisticated finance environments, the machine should inform judgment, not replace accountability.

Oscar's take on Nera

What was it about Nera's approach, never decrypting data even during processing, that stood out to you compared to how other tools handle sensitive data?

The architecture immediately caught my attention because it addresses one of the core tensions in enterprise AI. Historically, we have often accepted a tradeoff between computational usefulness and data exposure. To process information, the system generally needs access to the underlying information in a usable form.

The idea that sensitive information can remain protected throughout the computational process changes that risk equation. For someone like me, who may eventually want AI working across multiple client environments, that is particularly interesting. The limiting factor is not whether AI can analyze a financial model. We already know it can.

The harder problem is whether I can responsibly allow it to interact with highly confidential financial information at scale. That is the problem Nera is solving.

"The harder problem is whether I can responsibly allow [AI] to interact with highly confidential financial information at scale. That is the problem Nera is solving."

Oscar Casanova, Eververdant Growth Partners

You've been testing Nera closely for months. What convinced you this was worth your time as an early user?

The problem is important enough that I wanted to understand the technology before the market fully matured.

I run an advisory firm. My clients trust me with some of the most sensitive information inside their businesses. If AI becomes deeply embedded in how firms like mine operate, privacy architecture is going to become part of the operating model. What interested me about Nera is that it is not simply another AI interface.

I have also appreciated the team's willingness to take feedback seriously. Early-stage technology is never perfect, so for me the question is whether the underlying problem is important, whether the architecture is differentiated, and whether the team learns quickly. Those are the reasons I have continued spending time with the product.

Broader market view

Across your network of fractional CFOs and advisors, how common is the concern about putting client data into standard AI tools?

Very common, although the level of sophistication around the issue varies considerably. Almost every finance professional understands instinctively that uploading sensitive client information into a general-purpose AI system creates questions.

What varies is how deeply they have investigated those questions. Some firms have formal policies. Others simply avoid putting identifiable client information into AI tools. Others are using the technology aggressively without fully understanding the underlying data architecture.

I think that gap will close quickly. As AI becomes a core part of professional services delivery, clients are going to ask much more sophisticated questions about how their information is being handled.

Where do you see the biggest opportunity for AI in the fractional CFO and advisory world over the next couple of years?

A great CFO advisory relationship requires context. You need to understand the company's economics, history, strategy, people, KPIs, cash flow, forecast, and decision-making patterns. Today, a significant amount of that context lives fragmented across accounting systems, spreadsheets, emails, board decks, CRM systems, meeting notes, and the advisor's own memory. The future is an AI-native advisory environment where that institutional knowledge becomes continuously accessible.

Imagine an advisory firm where AI can prepare the initial variance analysis, identify changes in unit economics, reconcile forecast assumptions, summarize operating trends, surface anomalies, prepare meeting materials, and maintain institutional context across every client. The CFO then spends substantially more time on the questions that actually matter:

• What should we do?
• What are we missing?
• Where should capital go?
• What needs to change?
• What does this mean for enterprise value?

That is where I think the profession is going.

Looking forward

What would you tell another finance professional who's AI curious but hasn't taken the leap yet?

Start with a real workflow, not with AI. Ask yourself where highly capable people spend time on repetitive work that does not require their highest level of judgment. Pick one process. Understand the risk. Establish clear guardrails. Test the technology. Measure the result.

I see mistakes at either extreme. Some organizations deploy AI everywhere without sufficient governance. Others become so concerned about risk that they do nothing. Neither is particularly sophisticated. The better approach is controlled experimentation.

"Start with a real workflow, not with AI."

Oscar Casanova, Eververdant Growth Partners

What excites you most about where this is headed?

I think we are approaching a period where very small teams will be able to operate with capabilities that historically required significantly larger organizations. That is particularly exciting for the middle market.

A $10 million or $20 million company historically could not afford the same financial infrastructure, analytical horsepower, or specialized talent available to a Fortune 500 company. AI begins to compress that gap.

For advisory firms like Eververdant, that means the opportunity to bring institutional quality thinking, analysis, and decision support to businesses that historically could not access it. That democratization of financial capability is what excites me most.

About Eververdant Growth Partners

Eververdant Growth Partners advises founder led and middle market companies on strategic finance, value creation, and transactions.

About Nera Systems

Nera lets organizations use AI on sensitive data without exposing it. Data is encrypted before it leaves the customer's environment, stays encrypted even during processing, and only the customer holds the key. Nera's ChatApp brings this to finance teams and advisors, so they can use Claude, Gemini, or ChatGPT on client financials without the model provider, or Nera, ever seeing plaintext.