Last post we said the structured database exists today as labor. This post is about why it has to exist as something else — and why the dashboard, the thing every wealth-tech company is fighting over, is actually the least defensible part of the stack.
Bucket one is the rendering. Bucket two is the data underneath. Most of the industry has confused the two. We want to walk through why that’s a problem and where it ends up.
The Snowflake observation
Bear with us on an analogy from outside our category.
For a long time, the database business looked monolithic. The big data-warehouse incumbents — Oracle, Teradata, IBM — sold storage and compute tightly coupled, locked together by architecture and by contract. If you wanted to scale your compute, you bought more storage. If you wanted to query your data with a different tool, good luck. Vendor lock-in by design.
Snowflake’s breakthrough was separating storage from compute. Your data sat in one place — structured, owned by you, addressable. Compute scaled independently. The BI and analytics tools — Tableau, Looker, Power BI, your own SQL — could all read from the same canonical data without fighting each other or paying a tax to the database vendor. The data layer became the durable asset. The query and visualization layers became a competitive open market.
The companies that thought they owned the customer by owning the rendering eventually got commoditized. Tableau was acquired into Salesforce. Looker into Google. The data underneath stayed put. The customer didn’t leave the data. The customer rotated through dashboards.
Wealth tech is roughly where the data-warehouse business was in 2014.
Why dashboards keep losing clients to other dashboards
Here’s a brutal reality. Every wealth-tech founder we know has the same complaint about churn: a client picks up a competitor’s deck, sees a chart they like, and switches platforms. The migration is painful — six months, lots of reconciliation — but the family does it anyway, because the data underneath was never really theirs in the first place. It was the platform’s job to assemble it. So when they switch platforms, somebody assembles it again.
This is not a hypothetical. Addepar itself, despite being one of the strongest in the category, has reportedly penetrated only about 2.7% of the SEC-registered RIA universe according to Sacra’s analysis. The remaining 97% are running on something else, or on a quilt of somethings else, or on a spreadsheet. Movement between platforms is normal. The data work gets redone every time.
That is a category failure. Not a product failure. A category failure.
What “client owns the data” actually means
The phrase “the client owns their data” has been kicked around the industry for ten years. Mostly as marketing. Here’s what it would actually mean, operationally:
(1) The structured representation of every position, transaction, entity, ownership relationship, and document lives in one canonical place.
(2) That canonical place is controlled by the wealth owner, not by the vendor.
(3) Any rendering tool — dashboard, client portal, CPA software, estate-planning tool, AI assistant — can read from that place with permission.
(4) Switching renderings is a setting change, not a six-month migration.
(5) There is no such thing as selling the data. Not anonymized. Not anywhere. And you don’t need to jump through a bunch of hoops to extract your data, and it doesn’t kick out to complex, confusing Excel tables — it seamlessly ports. In fact, someone should just port it for you.
That is the Snowflake model applied to wealth. We don’t see it done the way we would want it architected. Most platforms in our category are architected the opposite way — to be the system of record, with rendering and storage bundled, and migration friction baked in.
We’re not making a moral argument. The vendors are doing the rational thing for their business. We’re making a category observation: the architecture that wins ten years from now is not the architecture that wins today. No technical debt in 2026 is a big deal.
Why AI accelerates this
McKinsey calls what’s coming for wealth tech the “SaaSpocalypse” — markets repricing wealth-tech equities because investors are figuring out that the rendering layer is about to become a commodity (Wealth Management). When a portfolio manager can ask Claude or ChatGPT “show me my exposure to commercial real estate across all entities, net of the leverage held at the LLC level,” and get a clean answer, the chart-and-PDF business loses its premium.
What doesn’t lose its premium is the structured database that the AI has to read from to answer that question. The AI is rendering. The data is the moat.
Anthropic’s Model Context Protocol, released in November 2024 and since adopted by OpenAI, Google, and most of the major enterprise vendors, is a tell here. MCP is essentially a protocol that says: “AI agents will read from your data wherever it lives, in a permission-aware, auditable way.” It is the open-data future for AI the same way HTTP was for the web. Moody’s is already an early adopter for credit and reference data. The wealth industry will follow.
When that future arrives, the question every wealth-management firm is going to be asked by its UHNW clients is, in effect, “Can my AI read my data?” And if the answer is “only through our dashboard,” that’s not going to fly.
What this means for buyers right now
If you’re a family-office principal, RIA founder, or COO evaluating wealth tech in 2026, here are three questions to ask any vendor before you sign:
(1) If we leave you in three years, who owns the structured data — us or you? In what format? With what fidelity? Was it ever sold to anyone, even anonymized?
(2) Can a third party — our CPA, our estate lawyer, an AI agent we authorize — read directly from our data layer without going through your UI?
(3) When a discrepancy exists between your dashboard and the underlying data, which one is the source of truth?
The reframe
The platform layer in wealth has spent fifteen years trying to be the system of record. The next decade will reward the firms willing to invert that — to be the system of record underneath, and let the rendering happen anywhere.
The dashboard is the thinnest part of the stack. Making it modular, and quickly able to incorporate your specific views, has immense benefits — because let’s be honest, do you really want to deal with the infrastructure management inside your family office, or SOC 2 audits within your RIA? But the data underneath it is the durable part.