Back to News
IndustryAI Understanding briefing

GeekWire reports Pro.com co-founders launch OnTrade, an AI wealth-management startup

GeekWire reports that Pro.com co-founders Raji Subramanian and Matt Williams have launched OnTrade with former Bank of America investment chief Zachary Harl. The startup connects existing wealth-management software and uses AI agents for portfolio reviews, tax-loss harvesting, account monitoring, and…

By 5 min readRead the primary source
AI-generated editorial illustration: Source-provided photograph of OnTrade co-founders Zachary Harl, Raji Subramanian, and Matt Williams. Created for “GeekWire reports Pro.com co-founders launch OnTrade, an AI wealth-management startup.”
The short version

GeekWire reports that Pro.com co-founders Raji Subramanian and Matt Williams have launched OnTrade with former Bank of America investment chief Zachary Harl. The startup connects existing wealth-management software and uses AI agents for portfolio reviews, tax-loss harvesting, account monitoring, and…

What happened

GeekWire reports that Pro.com co-founders Raji Subramanian and Matt Williams have launched OnTrade, a Seattle-based startup focused on AI software for wealth-management firms. They founded the company with former Bank of America chief investment officer Zachary Harl. According to GeekWire, OnTrade has operated privately since 2024 and raised an undisclosed amount from General Catalyst, Madrona, and angel investors.

GeekWire reports that OnTrade connects software already used by financial advisors, including customer-relationship management, portfolio-accounting, trading, and compliance systems, through a single interface. The company then deploys AI agents to perform tasks that advisors traditionally handled manually. Examples cited by GeekWire include scanning portfolios for tax-loss harvesting opportunities, flagging accounts that have moved away from target allocations, and drafting client proposals and reports. The source says humans approve the work before it reaches a client.

According to GeekWire, OnTrade has been operating under the radar since 2024 and has raised an undisclosed amount from General Catalyst, Madrona, and angel investors. The report identifies Jean Bredeche as chief technology officer; he previously co-founded algorithmic-trading platform Quantopian and later served as a director of engineering at Robinhood. The company’s founders are Subramanian, its chief executive; Williams, its president; and Harl, its chief investment officer.

GeekWire reports that OnTrade’s technology is being used by firms ranging from boutique advisory practices to large national firms. Williams told the outlet that one customer used the platform to win a billion-dollar family-office account and that another recovered the platform’s full annual cost in less than 30 days. These are company-provided claims reported by GeekWire. The source does not identify the customers, disclose the platform’s price, or provide independent documentation of either result.

The report places OnTrade’s emergence shortly after Vanguard agreed to acquire wealth-management platform Altruist. GeekWire says the deal was reportedly valued at $4 billion and that OnTrade’s founders view it as evidence of growing interest in wealth-management infrastructure. OnTrade differs from Altruist, as described by GeekWire, because it integrates with existing systems instead of requiring firms to move client assets onto a new custodian or replace their technology stack.

Source details: geekwire.com

Why it matters

OnTrade represents a vertical-AI approach aimed at automating specific, regulated financial workflows while leaving firms’ existing systems and client-asset custody arrangements in place. If the platform performs as described, it could help advisory firms handle more clients without requiring a complete technology replacement. The source does not independently confirm the company’s performance claims or customer outcomes.

GeekWire’s reporting frames OnTrade as an example of AI being built around a specific professional workflow rather than offered as a general-purpose model. Harl told the outlet that raw foundation models become more useful when connected to a firm’s portfolios, policies, compliance rules, and client relationships. That distinction matters because wealth management combines repetitive administrative work with decisions that can affect people’s savings, taxes, and financial plans.

The company’s integration strategy could lower the organizational barrier to trying AI. GeekWire reports that OnTrade does not require firms to replace their existing software or transfer client assets. In principle, that could make adoption easier for advisory practices that depend on established CRM, portfolio, trading, and compliance tools. Whether that advantage holds will depend on the quality of the integrations, the data-access permissions required, and the ability to audit the agents’ actions.

The potential public benefit described by the company is broader access to advice. Subramanian told GeekWire that wealth management has an access problem rather than a demand problem, while the report says the founders believe automation could help advisors serve more clients. That is a plausible business rationale, but the source supplies no evidence that OnTrade has expanded access for lower-wealth households, reduced fees, improved advice, or changed the distribution of financial services.

The financial setting also raises a higher standard for reliability than ordinary office automation. An agent that misreads account data, applies a tax rule incorrectly, drafts an unsuitable recommendation, or fails to escalate an unusual case could create financial or compliance consequences. GeekWire reports that humans approve outputs, but it does not explain the approval process, the controls around trading or tax-related actions, the system’s error rate, or how firms assign responsibility when an agent’s work is wrong.

What to watch next

The key questions are whether OnTrade’s integrations work reliably across different advisory firms, how its agents are supervised, and whether human approval is sufficient for tax, trading, compliance, and client-communication risks. GeekWire reports that the company has customers and cites results supplied by the company, but the article does not provide independent testing, customer names, financial results, pricing, regulatory details, or evidence that the reported benefits are representative.

The most important near-term issue is evidence of performance outside company-selected examples. GeekWire reports customer claims about winning a family-office account and recovering platform costs, but does not independently verify them. Useful follow-up reporting would identify the types of firms using OnTrade, describe the workflows actually enabled, and compare results with conventional advisor processes across a broader customer sample.

Readers should also watch how much authority OnTrade’s agents receive. The source describes agents that scan portfolios, flag drift, and draft documents, with humans approving the final work. It does not say whether the agents can place trades, alter portfolios, contact clients, or change compliance records. Clarifying those boundaries would show whether OnTrade is primarily an analytical and drafting tool or a system capable of taking consequential financial actions.

Data governance will be central to adoption. GeekWire says OnTrade connects multiple categories of financial software, but the report does not detail where data is processed, how customer information is retained, whether model providers receive it, how access is logged, or how firms can review and delete agent-generated records. Those questions are especially relevant when the platform handles portfolio information and client relationships.

Competition and market structure will provide another test. GeekWire names Orion Advisor Solutions, Envestnet, and Addepar as established players and reports that OnTrade views them as integration partners rather than direct rivals. Follow-up coverage should establish whether those companies permit the necessary integrations, whether they build comparable agent features themselves, and whether firms ultimately prefer an overlay such as OnTrade or a single integrated platform. The source does not provide pricing, revenue, funding totals, customer counts, regulatory filings, or independent assessments of availability.

Related guides & quizzes

AI AgentsAI Models ExplainedAI EthicsFuture of AITest what you know — try a free AI quizLook up an AI term in our glossary
Found this useful?