Pontera pitches an AI trust layer as Schwab prepares its Charley agent

Pontera spent years making the case that the recordkeepers standing between it and clients' 401(k) accounts should let it in, and now the New York company is arguing that the AI agents arriving in financial services will need a gatekeeper—and that it is built to be one. The pitch came in a Sept. 24 blog post titled "Agentic AI reframes the debate over financial access," written by Ben White, Pontera's senior director of public policy, and drawn from what he described as the takeaways of the 2026 FDATA summit in Toronto earlier that month.

White's argument is that agent-to-agent finance needs a trust layer satisfying both the humans behind the agents and the institutions on the other side, and that much of the ecosystem is not yet built for it: many firms lack the plumbing to tell a person from an authorized agent, while plenty of institutions have no way to govern automated activity running on their own systems. Pontera, he argues, can hold both jobs—a trust layer for the consumer and a protection layer for the institution—because a trusted intermediary can confirm a user's identity, limit an agent's permissions, avoid sharing more data than necessary, and keep records of activity.

Within days the argument had a live test: Schwab announced it is preparing to deploy "Charley," an AI agent foundation aimed at its millions of clients, arriving on the heels of Meta's Muse agent. Since Schwab is one of the recordkeepers whose cooperation Pontera has spent years courting, a custodian-built agent is the setting where the trust-layer pitch will be judged.

For advisors, the client-facing half of Schwab's announcement may matter as much as the data question, because a custodian telling millions of clients it is preparing to put AI agents on the job sets an expectation that account questions can be answered by a machine; firms managing held-away retirement money will field that expectation from clients whose plans sit somewhere else. Pontera's answer is that the intermediary validates and scopes what an agent may touch, a governance promise rather than a feature, and one that is only as good as the recordkeeper agreements behind it.

The dispute Pontera wants to reframe is concrete: advisors who manage held-away 401(k) assets depend on reaching client accounts that sit at recordkeepers, and Fidelity's ban on credential sharing closed one route in. The year-long battle includes an open letter from Pontera's chief executive that accused Fidelity of a "power grab" and called it the No. 1 plan provider; Pontera and Fidelity have not found common ground to date, and the practice has been framed as a privacy threat that could run afoul of the Labor Department rules plan sponsors live by. Pontera has floated ways it might work with Fidelity, but the coverage describes no agreement.

Not everyone treats the trust-layer pitch as settled strategy: Joel Bruckenstein, who sponsors the T3 events, said that "this is still an emerging field," so "any assumptions made today might be upended in months." He offered no verdict on motive—"I have no specific thoughts on whether Pontera is being clever or desperate"—and said he hopes Fidelity continues to monitor the situation and makes adjustments as technological advances permit.

For an advisory practice, the question comes down to data access: held-away accounts are only manageable if someone can authenticate the client, pull the data, and place trades inside a plan, and today that access turns on whether a recordkeeper grants it. Pontera's bet is that agentic AI turns sanctioned access from a favor an institution can withhold into infrastructure it needs, since once clients run agents of their own, every recordkeeper will have to sort legitimate automated requests from illegitimate ones and decide who vouches for them. A trust layer run by a third party is one answer, a trust layer run by the custodian is another, and Schwab now sits on both sides: an agent for its clients, and an AI seat for RIAs at $240 a year, reported in September.

Advisors who manage plan assets have a narrower stake than the futurism suggests: the case made in September—that the 401(k) savings gap is a plan-design problem, and that getting paid inside the plan, not at the rollover desk, is a fight over access and product—depends on exactly the account permissions Pontera is arguing for. Technology choice is already becoming a custody decision; Pontera's post pushes the same logic one layer down, where the software a practice uses to see and trade held-away accounts depends on agreements with the institutions holding them rather than on the software's own merits.

Bruckenstein's caveat is the honest one: the field is young enough that a framework built on this year's assumptions may look different in a year, and the two-roles pitch is just such a framework. The standoff remains, and it now has a new variable: Schwab is building agents of its own, Fidelity and Pontera have not reached terms, and the next recordkeeper to follow will decide whether a third-party trust layer is part of the access layer or a step the institution can skip.