The hedge fund had a research problem, but it wasn’t short of research. It was short of connections. Its applications worked awkwardly together. Analysts and portfolio managers struggled to share information across mobile devices, cloud services and other access points. As the firm expanded its research, manual processes became more expensive. The machinery around the thinking was getting in the way of the thinking.
That is the situation described in an RFA case study about an unnamed New York hedge fund managing $3.8 billion. It is also a useful introduction to a company whose work usually attracts attention when something stops working. RFA sells the infrastructure, support and security behind investment businesses. Its latest wager is that AI can make that work less reactive.
- The customer: investment firms that need specialist technology operations.
- The offer: IT, cloud, security, compliance and AI under one service relationship.
- The change: Titan ownership brings AI agents into RFA’s delivery model.
01 / The failure was between the tools
In the hedge-fund project, RFA began with an audit and a needs assessment. It documented the existing business process, designed a new information architecture and migrated the system. The implementation took six weeks, divided between discovery and rollout.
The ingredients were familiar: OneDrive, SharePoint Online, Office 365 and Microsoft Flow. RFA organized documents by ticker and connected emails, database subscriptions and other information sources into a shared library. Its case study describes easier collaboration and less manual work. The striking detail is the choice to improve how existing tools fitted together. The client’s problem lived in the joins.
For anyone commissioning a technology project, there is something to copy here. Write down the journey a piece of information takes before buying another place to store it. Otherwise, the next application can become one more island.
Six-week implementation · RFA’s published hedge-fund case study
02 / Who answers when everything connects?
Richard Fleischman founded RFA in 1989. The firm’s current customer list by sector includes hedge funds, private equity, asset managers, family offices and fund administrators. These businesses need people who understand the practical consequences of a failed connection, a missed access change or an investor asking for evidence of security controls.
RFA brings several jobs into the same relationship. Managed IT covers networks, servers, devices, patching and the service desk. Cybersecurity adds an in-house, round-the-clock security operations center, detection and response, phishing defense and incident handling. Cloud work spans Azure, AWS and Microsoft 365, including backups, migrations and cost reviews.
The less visible work matters too. Its compliance and advisory team provides virtual security leadership, policies, vendor reviews and help with investor due-diligence questionnaires. Development services build data warehouses, reporting and custom applications. The commercial proposition is ongoing managed service, with project work for changes such as migrations and development.
Consider the appeal from a buyer’s chair. A small investment firm may need a cloud architect, security analyst and experienced service engineer without building separate departments. RFA offers access to those capabilities through one provider. Responsibility becomes easier to locate, although the client still has to define what belongs in the contract.
Cloud recovery provides a concrete example of the expertise being sold. RFA describes business-impact analysis, written recovery runbooks, off-site replication and rehearsed failover. A recovery point defines how much recent data a business can tolerate losing; a recovery time defines how long a system can remain unavailable. Those choices belong to the business. Engineers translate them into an architecture and a test.
This is an established market. ECI and Agio also serve financial firms with overlapping technology and cybersecurity offerings. RFA’s case rests on its sector experience, breadth of delivery and now Titan’s platform. Buyers should test those claims against their own operating needs, rather than treating a long service menu as proof of fit.
03 / Then the help desk acquired an AI layer
On August 12, 2025, Titan announced two developments: a $74 million financing led by General Catalyst, and its acquisition of RFA. The money was Titan’s financing. It should not be read as the price paid for RFA.
The rationale was straightforward. Repetitive technical work consumes people’s time. Agents could collect information and execute defined workflows, leaving technicians more room for advice and complicated decisions. Titan co-founder Saurin Patel described its mission as empowering “the humans behind the screen.”
“Our mission is to empower the humans behind the screen”
Saurin Patel · Titan co-founder and CEO · August 2025
By August 2026, RFA was describing a rebuilt service model. Agents gather diagnostics, correlate signals and propose resolutions. Engineers retain responsibility for judgment and escalation. The distinction matters: assembling evidence and authorizing a consequential change are different jobs. RFA’s platform description includes human approval gates and reviewable action logs.
The management team reflects that combination of established service operations and new engineering. President David Andrade previously led ECI; chief operating officer Steve Schoener was its CTO. Bill Ercolano, formerly CTO of Portfolio BI, leads RFA’s technology. Their backgrounds put financial-services delivery experience beside Titan’s AI development capability.
It is also taking that experience to clients through Managed AI. The offering covers secure deployment, governance, spend visibility and a virtual Chief AI Officer. Optional products include managed agents, model routing, a governed knowledge base and application hosting. The intended buyer already has employees using AI and needs a program that someone owns.
04 / Borrow the sequence, then buy the service
The governance details are where this becomes concrete. RFA says its managed knowledge base preserves source permissions, so a model cannot retrieve a document a person cannot access. Its managed agents have named functions and approval boundaries. A buyer can ask to see those controls demonstrated with its own roles and documents before deciding how much work to delegate.
There is a practical way to evaluate all this. Begin with the work that hurts: repeated onboarding, scattered research, slow incident triage or unclear AI permissions. Ask who owns it, what evidence a fix produces and where a human must approve the next step. Then ask for the proposal to separate continuing operations from transition and development work.
For recovery, demand a rehearsal. For AI, agree access boundaries. For support, define response commitments and escalation routes. A firm seeking only a narrow tool may prefer a specialist product; one with substantial internal engineering may want a more limited engagement. Integrated services earn their place when coordination is a material part of the problem.
RFA’s research project offers the clearest lesson. The useful intervention began with understanding how people worked. Its AI chapter faces the same test: whether better coordination gives the people doing the thinking fewer reasons to stop.