On building the AI platform
for investment advisors in Brazil.

Brazil has 27,721 credentialed assessores de investimento as of early 2026 — a figure that grew 502% in a decade. They are not bankers. They are not portfolio managers. They are something closer to what an independent financial advisor is in the US or UK, but with structural characteristics that make the Brazilian version distinct in two ways that matter for this story.

First: the indicação model. In Brazil, high-trust professional relationships run on personal referrals. Clients don't find their assessor through a platform or a search engine. They get a phone number from a friend, a family member, a colleague. The assessor's book is not a user base. It is a network of personal obligations. When things go wrong, the call goes to the assessor — not to XP, not to BTG, not to anyone else. The relationship is the product.

Second: the volume problem. A productive assessor manages between 80 and 150 clients. Each client has a different risk profile, tax situation, investment horizon, and set of expectations. The assessor is expected to be proactive — to call before something matters, not after. To explain macro moves in terms the client understands. To produce documentation that protects the advisor legally and builds the client's confidence simultaneously. To do all of this, continuously, across a hundred different people, with no research team and no operations support.

This gap between what the role demands and what a single person can produce is not a new problem. What is new is that the gap is now visible — and measurable — to the client.

In February 2023, the CVM issued two resolutions that together constitute the most significant restructuring of investment advisory in Brazil in a generation. They took full effect in late 2024. The consequences are still unfolding.

CVM 178 ended the exclusivity requirement. Before, an assessor operated as an exclusive preposto of a single intermediary — tied to XP, or BTG, or Clear. After CVM 178, assessores can work across multiple intermediaries simultaneously. This sounds like a technical change. It is a structural one. The captivity that kept assessors inside a single platform is gone. Independence is now legally possible. And wherever independence is possible, clients start asking why they're paying for a service that is constrained by a distribution arrangement they never chose.

CVM 179 ended opacity. Since March 2025, every client receives a quarterly extrato de remuneração — a detailed statement showing exactly what the advisor earned from their portfolio during the period. Every rebate, every distribution fee, every basis point of commission paid by any fund or product in the portfolio, disclosed in a format the client can read and share with someone else.

The question that statement generates is now the defining question of the profession: is my advisor's judgment worth what I am paying for it? The advisors who keep clients are the ones who can answer yes — with evidence. Research they produced. Reports they sent. Decisions they made and documented. Not just a relationship that felt good in better markets.

The direction of travel is also unmistakable. The consultor CVM model — fee-only, fiduciary, CVM-registered, fully transparent — has more than doubled in five years. In the most recent year tracked, more new consultorias registered with CVM than new gestoras. The fee-based, fiduciary model is not a niche. It is the direction the regulation is pointing every advisor toward, whether they move voluntarily or are pushed.

XP alone operates with approximately 18,000 assessores — 3,000 internal and 15,000 autonomous. BTG, Ágora, Clear, NuInvest, and dozens of independent offices account for the rest. These are not retail consumers. They are professionals with recurring revenue, an existing client base they are trying to protect, and a clear economic incentive to buy tools that help them do that.

The easing Selic cycle adds a second layer of urgency. The decade of high fixed income rates created a generation of clients whose portfolio strategy was essentially “Tesouro Selic until further notice.” That strategy is now actively losing ground. Every assessor with a fixed income-heavy book is having the same conversation simultaneously: where does this go, and what do I recommend? The advisors who can answer that question with research and conviction are the ones who deepen the relationship. The ones who answer with a morning note forwarded from the platform are the ones who lose clients to someone who can.

The intuitive move in fintech is to go direct to the investor. Build the robo-advisor, the AI chatbot, the self-directed portfolio app. Cut out the advisor and capture the economics directly. Dozens of companies have tried this in Brazil in the last decade. None have succeeded at scale, and the reason is structural, not executional.

The indicação network does not transfer to a platform. The client who trusts their assessor does not trust a product that claims to replace them. The “someone to blame” dynamic that defines high-stakes financial relationships in Brazil is not a problem that better UX solves. When the portfolio drops 20%, the client calls a person — not a dashboard. That person's phone number, their relationship, their accountability: that is the distribution mechanism for financial services in Brazil, and it cannot be disintermediated by an app.

Velna's bet is the opposite of disintermediation. We are building for the assessor, not against them. The assessor keeps the relationship, the accountability, the indicação chain. Velna provides the research infrastructure, the portfolio surveillance, the documentation — the things that made it impossible for one person to serve 120 clients at the level clients now expect.

This is the same model that won in legal AI, in investment banking research AI, in medical documentation AI. The professional who already has the client relationship gets dramatically more capable. They don't need to be replaced — they need to be leveraged. The tools that make them 5x more productive don't face adoption resistance; they face pull demand, because the advisor's income depends on keeping clients, and keeping clients now requires a level of service that is impossible without infrastructure.

The result is B2B2C: Velna sells to advisors, advisors serve clients better, clients stay and grow their AUM with the advisor, the advisor pays for Velna because their economics depend on it. The end client never sees Velna. They see a better advisor.

Velna is an AI research and productivity platform for investment advisors. In its current form, it does four things that previously required a team: research generation, portfolio surveillance across an advisor's full client book, client-facing report production, and compliance documentation.

The research layer pulls from a purpose-built data warehouse: live market data across US and Brazilian equities, CVM fund registry and daily NAV data for the full Brazilian fund universe, BCB macro series, FRED US macro, SEC and CVM filings, earnings transcripts, Tesouro Direto rates, and Brazilian tax rules. This is not a generic web search. It is structured financial data, indexed for the kinds of questions an advisor actually asks about a client's portfolio.

The AI agent layer runs on top of this warehouse. An assessor can ask: “What should I tell a client with 60% Tesouro Prefixado 2029 exposure, given today's COPOM communication and the current IPCA trajectory?” Velna does not return a generic answer about fixed income. It synthesizes the specific macro context, the specific instrument characteristics, the client's stated objectives and tax situation, and produces a reasoned recommendation that the advisor can review, edit, and send.

The portfolio layer gives the advisor a view of their entire book simultaneously — not one client at a time. Which clients have positions maturing this month. Which portfolios have drifted from target allocation by more than a threshold. Which clients hold something that a macro event just made relevant. The advisor sees the priority. The client gets the call.

The documentation layer generates compliant suitability analysis, investment policy statements, and rebalancing proposals — not from templates, but from the actual client context: their risk profile, their holdings, their objectives, their documented history with the advisor. Every recommendation is covered. Every decision is on record.

This is not a feature set. It is the infrastructure that makes the difference between an assessor who can serve 80 clients well and one who can serve 200. That 2.5x multiplier on book size, at the advisor's existing AUM fee rate, is the economic argument for adoption. It does not require a behavior change. It requires giving the advisor what they already need.

The assessor market is the wedge, not the ceiling. Brazil's wealth management ecosystem has layers that the assessor layer connects to: consultorias CVM managing family office mandates, multi-family offices serving ultra-high-net-worth clients, gestoras running discretionary mandates for institutional accounts. Each of these layers has the same structural problem — too many clients, too complex a market, too little research infrastructure — and each is willing to pay substantially more per seat for a tool that solves it.

The data advantage compounds in one direction. Every advisor who uses Velna enriches the models with the signal of real professional judgment — what questions get asked, which recommendations prove right, what documentation patterns protect advisors most effectively. This is not generic training data. It is the accumulated professional knowledge of Brazil's advisory ecosystem, encoded into a platform that becomes harder to replicate as it grows.

The regulatory trajectory also continues in one direction. The CVM is not done. The pressure toward transparency, toward fiduciary standards, toward documented and defensible advice, will only increase. Every advisor who uses Velna today is building the compliance infrastructure that the regulation will require of everyone eventually. The early adopters are not just getting a productivity tool. They are getting ahead of a requirement.

The vision is the operating system of Brazilian wealth management — not a product that individual advisors use, but the infrastructure layer that the profession runs on. If the platform that every serious assessor and consultor CVM uses is Velna, and every advisor interaction with every client produces data that flows back into the research and compliance layers, the network becomes the moat. The data advantage becomes permanent. The switching cost becomes the cost of leaving behind the documentation history of every client relationship you ever managed.

We are not trying to build a better CRM. We are not trying to build a robo-advisor that cuts the advisor out. We are building the infrastructure that makes the best advisors in Brazil more capable than any investor's money can currently access — and then making that infrastructure available to every advisor who is serious enough to use it. That is a different kind of ambition. It requires a different kind of company.