There is now broad agreement that artificial intelligence is becoming part of the technology infrastructure of wealth management. But while the direction of travel is increasingly clear, a more important question remains unresolved: what will the AI architecture of the industry actually look like?

One possible answer is already taking shape. Large wealth-management technology providers are building their own AI solutions.

At CGL Group, we have been analysing this market and have identified several examples of this approach.

Orion has developed Denali AI, which it describes as an enterprise intelligence platform for advisory firms. Denali connects Orion products with third-party systems and is designed to automate workflows across the firm.

Another major player, Addepar, has taken a similar approach with Addison, its native AI system integrated directly into the Addepar platform. Its proposition is closely connected to one of Addepar's strongest assets: investment data. Addison operates on a unified, permission-aware data foundation and produces results grounded in actual portfolio data.

Another example is SS&C, which has launched Black Diamond AI, bringing AI assistance, contextual insights and natural-language interaction into its wealth-management platform. Envestnet, meanwhile, is developing what it calls an “AI-native advisor experience”, moving from individual analytics tools towards AI-supported decision-making and workflows.

The logic behind this model is easy to understand. These companies already have clients, financial data, permissions, workflows and specialised software. Adding their own AI allows them to make their existing platforms more intelligent without requiring advisers to move to a different system.

The arrival of frontier AI

In September 2026, Anthropic launched Claude for Financial Advisors.

Instead of creating another portfolio-management, CRM or financial-planning system, Anthropic is connecting Claude to the systems that financial advisers already use. Through integrations with leading financial software providers, Claude can work across information held in custodial systems, CRM, portfolio reporting, financial planning, estate planning and meeting records.

Josh Brown, CEO of Ritholtz Wealth Management, described Claude for Financial Advisors as something that “sits on top of the whole stack.”

Some technology companies have already chosen to work with this model. Wealthbox brings its CRM context directly into Claude, allowing a frontier model to work with client and adviser data. iCapital is taking a similar approach, connecting its specialised investment marketplace, data and infrastructure to Claude.

Others appear to be keeping both options open. SS&C Black Diamond is perhaps the clearest example. On the one hand, SS&C has developed its own Black Diamond AI. On the other, it has released a Model Context Protocol (MCP) server that allows advisers to access Black Diamond data from their “AI enterprise platform of choice.”

In effect, SS&C is preparing for two possible futures: one in which intelligence remains within specialised wealth-management platforms, and another in which those platforms provide their data and functionality to an external AI.

It is also important to understand that Anthropic is not simply developing another AI product. It is building an ecosystem around Claude. In March 2026, Anthropic committed $100 million to the Claude Partner Network to support partner training, technical assistance and joint marketing. By June, more than 40,000 firms had applied to join and more than 10,000 consultants had received Claude certification.

This suggests that Anthropic's strategy extends well beyond building the underlying AI model. It is also creating a network capable of bringing Claude into individual industries and companies.

This matters because enterprise AI adoption is not simply a technology problem. Companies need people who can identify use cases, redesign workflows, connect existing systems, implement AI, train employees and manage adoption.

What do industry studies tell us?

There is no clear answer yet as to which architecture will prevail. However, several recent industry studies point towards increasingly integrated combinations of AI, data and workflows.

Capgemini's World Wealth Report 2026 describes an emerging “intelligence layer” capable of orchestrating workflows and connecting the expanding ecosystem of products and specialists within wealth management.

Deloitte reaches a related conclusion from a different direction, arguing that progress depends less on individual tools and more on strengthening data foundations and redesigning end-to-end processes. McKinsey, meanwhile, expects wealth-management technology to rely increasingly on modular, API-based architectures connecting legacy systems with new digital platforms, with firms combining internal development, acquisitions and external partnerships.

One particularly interesting example of how such a situation can develop comes from New Zealand. Soul Machines was founded around the idea of creating intelligent “Digital People” — human-like digital characters designed to interact with users. The company attracted substantial investment. By 2022, it had raised approximately $135 million, with investors including SoftBank Vision Fund 2, Temasek, Salesforce Ventures and Horizons Ventures.

But the technological environment around the company changed rapidly. As general-purpose AI became significantly more capable, the conversational intelligence behind a digital human no longer necessarily had to be developed by the digital-human provider itself. By 2025, Soul Machines allowed its Digital People to connect to external systems including OpenAI's GPT, Microsoft Copilot, Google Dialogflow, Amazon Lex and IBM Watson.

This raises a fundamental strategic question: if a general-purpose AI can provide an increasing share of the intelligence, how much value remains in a separate specialised AI layer between the user and that intelligence? In February 2026, Soul Machines entered receivership.

There is no evidence that the rise of general-purpose AI was the direct cause of Soul Machines' failure. But the case illustrates how quickly the competitive position of a specialised AI product can change when capabilities around it become available from much larger general-purpose platforms.

What does this mean for wealth management?

It is still too early to know which model will ultimately prevail. However, the experience of other AI markets suggests that the growing role of frontier AI companies such as Anthropic and OpenAI should not be underestimated. We believe particular attention should now be paid to how these companies move into wealth management — and how specialised wealth-management platforms respond.