AI budget share doubles as firms wait for proof

Three wealthtech vendors used the week before Future Proof to ship AI upgrades aimed at the manual work of finding problems inside client accounts and portfolios, and the batch says more about the next software renewal than about any one product. Vanilla, SS&C and FinTurk rolled out capabilities spanning autonomous monitoring agents and AI-assisted analytics, a spread InvestmentNews reads as evidence that agentic systems — software acting on an advisor's behalf rather than answering questions — are moving out of pilots and into core platform features.

The money is arriving ahead of the proof: a benchmark study released this week by Cerulli Associates and Vista Equity Partners projects AI spending roughly doubling as a share of technology budgets in 2026, from approximately 8 percent to approximately 15 percent, while F2 Strategy separately reports that firms are spending more on the technology even as translating that spend into measurable returns has proved harder than expected. Held together, the two findings describe what a principal carries into the next budget cycle — a line item set to double against a payoff the buying firms have not documented.

A share can double without a dollar being added, and that is the least comfortable reading of the Cerulli-Vista projection. An AI line climbing from 8 to 15 percent of the technology budget can get there by shrinking everything else in the stack — reporting, data, CRM — and calling the reallocation a strategy, an arithmetic that principals who have been through a platform consolidation or two will recognize.

For an RIA operator, the question is less whether an agent can find the unfunded trust than who clears the finding, and what gets displaced to make room for the clearing.

Vanilla's announcement is the concrete one: VAI, its estate planning assistant, will now scan an advisor's entire client roster instead of only the largest accounts, flagging unfunded trusts, outdated beneficiary designations and insurance gaps measured against a client's estate tax exposure. The capability builds on a partnership struck earlier this year with Carson Group that put Vanilla's tools in front of more than 600 advisors across the network, and it follows a March deal that added AI-native tax intelligence provider Worthy to Vanilla's roster of integration partners.

The company says its extraction engine has already processed more than 60,000 estate documents and modeled more than 32,000 estates, and that the deterministic calculations underneath those models are unchanged even as agents extend the platform's reach. Client-identifying data, in Vanilla's account, never reaches a language model — a de-identification approach the firm has protected by patent since VAI's original launch in 2023.

Chief executive Gene Farrell frames the shift as moving work off the advisor's desk rather than adding to it. "We're evolving the platform to be agentic so AI can do the work advisors actually need done," he said, describing a system that returns the underlying reasoning behind each finding so the advisor can act and leaves the response to the person Farrell calls the one "with the relationship and the license."

That last clause is the design decision worth watching, and it cuts against the loudest version of the agentic pitch: software that reads a full book and returns every unfunded trust, stale beneficiary form and coverage gap produces a queue rather than a resolution. Detection has gotten cheap; the hour spent with a client amending the document, and the follow-through that keeps the amendment from being re-filed and forgotten, have not. Advisors who treat a client's estate plan as static risk the relationship itself — 68 percent of advised clients are open to switching, as this publication has reported — and software that flags everything raises the premium on that discipline rather than standing in for it.

Carson is the useful test case, because the platform breadth that makes its Vanilla rollout land is a function of how fast the firm grows by acquisition: Carson logged advisor-team liftouts in August, including one carrying $1.8 billion. Every team that arrives brings a book of estate documents with designations and trusts written years ago, which is precisely the inventory a whole-roster scan is built to read. That is a real operational gain, and it is also why the follow-up problem gets bigger rather than smaller.

AI's consequential arrival in advisory is in the client meeting rather than the back office, where the plan becomes a byproduct of the software and the fee migrates toward judgment and liability, and Vanilla's architecture is a bet on the second half of that migration: keep the deterministic math, keep identifying data out of the model, leave the signature with the advisor. The autonomy-first framing that sells demos points the other way, and the firms that buy it will find they have purchased a longer to-do list.

The arithmetic F2 Strategy flags is the part this week's releases skip: at roughly 8 percent of the technology budget, an agent that generates follow-up work can be absorbed by staff who already exist, but at 15 percent it has to displace something — a hire, an incumbent platform, a data contract — and nothing in this batch says which one a principal should give up. The platform arms race has already turned on adoption and retention, and unresolved findings are how adoption actually gets measured.

Future Proof opens next week and the demos will show the finding: the unfunded trust, the stale designation, the insurance gap a client's estate tax exposure has outgrown. The number that decides whether the category earned its doubled share is the proportion of those findings an advisor clears, and this week's announcements do not carry it.