A $36 billion hybrid funds the data foundation before the AI framework

Signature Estate & Investment Advisors has consolidated enterprise data from multiple custodians, its advisory and broker/dealer businesses and its in-house turnkey asset management platform into one governed foundation, working with the data platform Invent. The $36 billion hybrid's stated destination is SEIA Brain, an internal AI framework meant to let employees across marketing, sales, finance, operations and compliance pull information in plain language, automate routine workflows and reach insights they would otherwise wait on a report to see.

The line a firm principal should underline in that description is not the AI framework but the dynamic reporting underneath: authorized users can now create, publish, filter, sort and export dynamic reports themselves, without relying on third-party reporting specialists, according to Advait Kulkarni, SEIA's senior vice president of technology operations. For a business running an advisory practice, a broker/dealer and a TAMP across several custodians, that is hours coming out of the operations budget every month — the least glamorous item in the announcement, and the one most likely to reach the margin line.

Kulkarni's account of where the firm started will read like a mirror to anyone who has inherited a growing hybrid: before he joined, the work sat across four or five different systems, some homegrown and some off the shelf, held together by a few connectors, with layers shoe-horned into jobs they were not designed for. The homegrown data architecture was difficult to maintain and leaned on legacy connectors with problems of their own, so interoperability and integration underperformed. "When I started, I wanted to identify a single source of truth and build off it," he said.

The shape of the firm is what made that expensive: a single-custodian practice with one book can reconcile at month-end with a spreadsheet and a patient analyst, but add a second custodian, a broker/dealer carrying its own reporting obligations and an in-house TAMP, and the same household arrives in several formats under several sets of identifiers. Cleaning, reconciling and preparing that data for reporting was significant manual work at SEIA, according to Kulkarni.

Invent, which launched in 2019, sells the other half of the arrangement, and its argument is about what it does not own. Oleg Tishkevich, the founder and CEO, describes the platform as an independent ecosystem: because it does not compete with the applications sitting on top of it, a firm keeps whatever third-party technology it already runs — the key difference, he says, from the all-in-one agentic platforms now arriving in the market that ask a firm to trade its stack for the integration.

The fragmentation is not SEIA's alone: wealth management firms of every size, hybrid and pure RIA alike, are wrestling with the same pile of custodial feeds and point systems, even as those all-in-one agentic platforms come to market to address it. SEIA took the other trade — keep the systems, buy the layer that governs the data between them, and build the intelligence on top itself.

Reconciliation latency belongs to the advisor as much as to the operations desk, because when a reporting cycle opens with a manual cleanup pass across several source systems, the figures that reach a client review are a reconstruction and the person defending them is the advisor. Consolidating the feeds shortens the distance between what the custodian reports and what the client hears. The reporting on the project does not put a figure on the hours saved, which is the number a principal would ask for first.

As this publication has argued, the AI premium in wealth management has moved from the quality of the model to the governed copy of the client record — a reconciled, permissioned version of the household that a language model can safely read. SEIA's build sits on that thesis: the firm bought the foundation first and put its own framework on top, which is why SEIA Brain carries the name and the data layer carries the invoice.

That is the right order of operations, and it is the order most firms will get wrong. An assistant answering a service question from unreconciled custodian feeds does not fail loudly; it answers confidently and incorrectly, and the advisor discovers it in front of the client. The reporting and workflow gains Kulkarni describes are real, and they are also the cheap half of the program; the expensive half is governance — which employee may see which household, and whether the model's answer can be traced back to a custodian statement — and none of that shows up in a demo.

Whether the arrangement compounds into an advantage is the open question, and the answer turns on the part that cannot be bought: SEIA Brain is the firm's own, but the governed layer beneath it is a product any competitor of similar size can license. The mapping work is what resists duplication — settling what a household is when it is custodied in several places, reported through a broker/dealer and managed inside a TAMP, then holding that definition in production while everyone's systems change. That effort is invisible in the announcement, and it is the asset.

SEIA's complexity also makes it a revealing early reference for the platform it hired: a $36 billion hybrid with multiple custodians, two business lines and an in-house TAMP presents close to the hardest data problem a wealth firm can have, and the same sprawl that sells the story is the sprawl that has to run on a Tuesday.

The framework's first permission is internal reporting, the safe first deployment because a wrong answer stays inside the building; the premium, and the exposure, arrive when plain-language access reaches client-facing questions, the direction the industry's AI budgets have been moving. For now the tell is where the money went: under the framework rather than in it, and firms that fund the demo before they fund the data will find out which half was expensive.