With the mid-September announcement of Claude for Financial Advisors, wealth management is entering a new phase in how AI connects with financial data and technology.
In an interivew, Masttro's Domingo Viesca, Co-Founder and Head of Innovation speaks to what this shift means for complex wealth, why trusted data remains critical, and how Masttro is approaching an increasingly open AI ecosystem.
In Q4 2026, Masttro plans to launch a platform-agnostic MCP, giving clients a controlled way to make selected, read-only Masttro data available to external AI environments including Claude, OpenAI and others.
There’s suddenly a lot of attention on connecting AI platforms like Claude to wealth technology. What do you think is changing for family offices and wealth managers?
Clients will increasingly expect to interact with their information through different environments, whether that is Masttro Intelligence, Claude, OpenAI or other platforms that emerge.
As those interfaces become more flexible, what matters increasingly is the system underneath them that understands what a family owns, how everything is connected, which information is authoritative and who is entitled to see it.
That matters even more when the underlying ownership structure is complex.
Masttro is the system of record for the client’s complete estate, bringing together public and private investments, entities, trusts, real estate and other financial and nonfinancial assets, together with the documentation and ownership structures that give that information meaning.
We do not need to build another frontier model. Our role is to combine the reasoning capabilities of those models with Masttro’s deep understanding of complex wealth.
Clients should also be able to choose where they work. They can use AI natively through Masttro Intelligence or selectively make approved, read only Masttro data available to an external model when that better fits their workflow.
Why is complex UHNW wealth such a difficult problem for general purpose AI to solve on its own?
Because the numbers alone are not enough.
A family may own public securities, private companies, funds, real estate and other assets through trusts, partnerships, holding companies and accounts across multiple jurisdictions, currencies and generations.
Making sense of that wealth requires more than reading numbers. You need the ownership structure, relationships, documents, methodologies, valuations, commitments and permissions around them.
Models like ChatGPT and Claude are very capable reasoning tools, but they do not inherently know what belongs to whom, which information is authoritative, how a financial calculation should be performed or whether the person asking a question is entitled to the answer.
Masttro provides the wealth context that makes that reasoning relevant and precise.
Why can’t a family office simply upload its data into Claude or ChatGPT and start asking questions?
It can, and for certain use cases that may be entirely appropriate.
The difference is between giving a model a set of files and letting it work with information that is continuously maintained and governed.
An uploaded file is a snapshot. The model may understand its contents very well, but it does not automatically inherit the ownership relationships, methodologies, provenance, permissions and subsequent changes that give that information its full meaning.
For UHNW families, there is another critical dimension: security, confidentiality and continued control over where their information exists and how it is handled.
Once sensitive wealth information moves into another environment, the family office needs to know where it is processed and what may be retained. It also needs clarity on who can access it, how long it persists and how permissions are enforced.
The question is therefore not simply whether an external AI platform is secure. It is whether the family can maintain clear and continuous visibility and control over its most sensitive information as that information begins moving across multiple systems.
The more important question is not simply whether AI can read the data. It is where the intelligence should operate, what information it needs and what data the client chooses to make available externally.
We believe that decision should remain explicitly with the client.
Why has Masttro focused so heavily on building that trusted data foundation before opening it up to external AI platforms?
Because in the AI era, trusted data becomes even more valuable.
AI can reason at extraordinary speed, but the quality of that reasoning depends on the quality, precision and context of the information underneath it.
Masttro has spent years building a system of record capable of consolidating a client’s complete estate, including financial and nonfinancial assets, ownership structures, documents, normalized data, precise financial calculations and permissions, within a secure and confidential environment.
Without that foundation, better models do not solve the underlying data problem.
Masttro Intelligence works directly with that system of record, so it can use the ownership structures, permissions and financial context already maintained inside Masttro.
An important part of that architecture is that Masttro Intelligence is model agnostic.
We are not building our intelligence strategy around Claude, OpenAI or any single provider. We can use the models best suited to different capabilities while Masttro continues to provide the wealth context, permissions and workflows around them.
As frontier models become more capable, Masttro Intelligence can benefit from those advances too.
Clients can benefit from progress across the AI ecosystem without rebuilding their wealth infrastructure around whichever model happens to lead at a particular moment.
In practical terms, what does an MCP change? Why is this different from simply exporting Masttro data and uploading it into Claude or ChatGPT?
MCP provides a controlled way for an external AI environment to interact with selected Masttro data.
Our MCP, launching in Q4 2026, will primarily expose selected capabilities already available through Masttro APIs. It is a controlled, read-only data interaction layer, not the Masttro Intelligence environment itself.
Instead of exporting large datasets, clients can authorize an external model to request specific information through controlled interfaces.
The client controls whether that access exists and what information is available. The interaction remains read only, and the external model does not inherit Masttro’s complete intelligence context or the ability to modify information inside the platform.
This lets clients work in external AI environments without treating access to selected Masttro data as equivalent to having the full Masttro context.
Looking ahead, how do you see Masttro Intelligence, Claude, OpenAI and other AI platforms working together?
We see them as complementary.
Masttro Intelligence is our primary AI environment because it operates on top of the client’s trusted system of record, with the wealth context, documents, permissions, calculations and workflows required to reason accurately and act meaningfully.
Its architecture is deliberately model agnostic, so we can use different models for different tasks rather than building around a single provider.
That also means clients are not tied to the capabilities of one model.
The models will continue to change quickly. What needs to remain consistent is the underlying wealth record, the permissions around it and the client’s control over how that information is used.
Some clients will want to work directly in Claude, OpenAI or other AI environments. For them, Masttro can provide selected, read-only access to approved data through our MCP server.
So there are two ways of working.
They can use Masttro Intelligence, where it can operate with the richest context and support more complex workflows. Or they can bring selected Masttro data into the external AI environment of their choice.
For us, the important point is that clients can benefit as AI models improve without having to rebuild the underlying wealth infrastructure around each new model or interface.