Zeplyn shipped an agent that completes Schwab's own account-opening workflow, aimed at the rejection-and-rework loop rather than the paperwork, and the distinction is the whole story.
The client already said yes. That part is finished. What happens next is that somebody at the advisory firm spends the following week or two turning that yes into an account that actually exists, and the reason it drags is almost never the form. It is the rejection. The custodian sends the packet back because a signature is missing, an address does not match the statement, a beneficiary field contradicts the trust document. Wealth management has an acronym for that moment, NIGO, short for Not In Good Order, and the person who cleans it up is usually a paraplanner or a client service associate. The New Planner Recruiting 2025 Salary Report, cited in Kitces' compensation benchmarking, puts the average salary for a planner with zero to two years of experience at $73,808, which is about $35 an hour before employer taxes. On August 12, Zeplyn shipped an agent that does not help that person fill out the packet faster. It goes and fills out Schwab's workflow instead.
That preposition is the entire story, and it is worth sitting with before the feature list.
What actually shipped
The August 12 release describes a chain that starts with one request from an advisor and ends with a finished draft. Zeplyn's agent pulls client information out of meetings, emails, CRM records, and documents, opens Schwab's digital account-opening workflow, completes the relevant fields, and hands back a draft for a human to review and submit. Finovate confirmed the launch the same day, and the integration also pushes live Schwab holdings and transactions into Zeplyn's client briefs.
Co-founder and CEO Era Jain framed it as "enabling AI to complete work that has traditionally remained manual." The numbers attached to that claim are Zeplyn's own early pilot figures: Not-In-Good-Order submissions down roughly 80 percent, and more than 12 hours a week saved on meeting preparation. The platform carries SOC 2 Type II compliance.
Notice what the agent is not doing. It is not generating a prettier intake form on the firm's side of the wall. It is not summarizing the client file so a human can retype it into somebody else's portal. Almost every tool sold into intake work over the last decade has lived on the near side of that wall, producing a cleaner draft that a person still has to transcribe into the counterparty's system. The transcription step is where fields get dropped, and the dropped field is what comes back as a NIGO two weeks later. An agent that operates the custodian's own workflow is a different category of product, because it removes the handoff that generates the error rather than making the error easier to fix.
What it costs against what it replaces
Zeplyn does not publish a price list. Its pricing page is a contact form, which tells you something about who the intended buyer is. The figures the company has put on the record, reported by WealthManagement.com in July, start at $120 per month per advisor seat for the Meeting Assistant, with the agentic Agent Nexus tier starting at $195 per month per advisor seat. Enterprise pricing scales with deployment model, integrations, volume, and firm size.
Run the arithmetic at $195. That is $2,340 per advisor per year. Against $35 an hour of paraplanner time, the seat pays for itself at roughly 67 hours a year, which is about an hour and twenty minutes a week. Zeplyn's pilot claim is more than 12 hours a week, and that claim is unaudited vendor data from a small sample, so treat it as marketing until a firm you know reproduces it. The useful thing about the math is how much room it leaves. Discount the vendor's number by ninety percent and the seat still clears its own cost.
The savings that matter, though, are not on the timesheet. A NIGO does not just consume staff hours. It restarts the clock on a client who has already committed, in the exact window where a new relationship is most fragile, and it does it while the assets are still sitting at the previous firm.
Who it is wrong for
Start with the obvious constraint: this is Schwab. The release details one custodian, and a firm running Schwab alongside Fidelity or Pershing gets the benefit on a fraction of its account openings while paying per seat across every advisor. The release also describes early pilots without naming a general availability date, so anyone treating this as shipped-and-proven is reading ahead of the evidence.
Volume is the second filter. Sixty-seven hours a year is a real threshold, and a two-advisor shop opening a dozen accounts a year will not cross it. At that size the seat cost plus the work of rewiring how the firm handles intake is worse than the problem.
The third one is the honest one, and it applies well outside wealth management. If a firm's NIGO rate is high because its CRM is full of stale addresses and half-finished client records, an agent that reads that CRM will fill out Schwab's forms with the wrong data faster and more confidently than any human would. The rejection rate does not improve. It just arrives sooner, with better formatting. Automating an intake process on top of bad source data does not fix the rework loop, it industrializes it. Any firm looking at this should measure where its rejections actually originate before buying seats, because the agent inherits whatever it reads.
Every intake process in every industry has a wall in the middle of it, and the expensive work has always been on the far side. The question worth asking about any agent sold into that work is not whether it can fill out your form. It is whether it can fill out theirs.