Start here
Start here: how an AI holding company works
A five-minute introduction to buying businesses, changing their workflows with AI, and the execution questions that determine whether the idea works.
Engraved illustration of a rough sea. The picture is decorative; the site works without it.
Read the arguments shaping this field: original letters and essays, independent reporting, interviews, and podcasts. Every item says whether the text is available here or should be read at its publisher.
Start here
A five-minute introduction to buying businesses, changing their workflows with AI, and the execution questions that determine whether the idea works.
Announced August 12, 2025, from New York. Titan, an AI holding company incubated by General Catalyst, announced $74 million in funding led by General Catalyst and, in the same release, its acquisition of Richard Fleischman & Associates (RFA), a New York-based managed service provider with roughly 250 employees serving more than 400 financial services clients across multiple countries. RFA becomes Titan's first MSP platform, into which Titan layers its augmented AI stack. The stated thesis is that traditional MSPs are constrained by legacy tooling and manual, ticket-based workflows, and that agentic AI can automate routine support, predict failures, and free technicians to act as strategic advisers rather than reactive support. Co-founders Saurin Patel and David Heffernan said Titan intends to acquire additional IT services firms. Yohan Kim is chief executive of RFA; Marc Bhargava, managing director at General Catalyst, commented on the investment. The deal was one of several executions of General Catalyst's AI-enabled roll-up program, for which the firm allocated $1.5 billion beginning in late 2023.
Argues for owning distribution while acknowledging the difficulty of combining software development with acquisition integration.
Links better document processing, legacy-system automation, and conversational agents to owning and executing service workflows. Discusses Sequence and Arcos.
Curates frameworks and operator interviews. Pushes back on treating software multiples as a necessary condition for the thesis.
Revisits the thesis through capital structure, real product advantage, and the time required for compounding. Separates platform-building dollars from asset-purchase dollars.
Supports building a strong product before acquisitions, progressing toward larger assets, and using capital efficiently. Warns that buying businesses is costly and difficult.
Visible introduction records the strategy spreading across major venture firms and multiple services categories.
Announced June 18, 2025, from Singapore. Multiplier Holdings publicly launched with $27.5 million in seed and Series A funding. The Series A was led by Lightspeed Venture Partners, with Ribbit Capital, which had led the seed, plus EDBI and SV Angel participating. The model is to acquire specialized professional services firms — tax and accounting practices in particular — and integrate custom AI automation, using an in-house library of AI and workflow components configured to each acquired firm. Multiplier states that portfolio firms keep operational independence while incentives are aligned across Multiplier leadership, firm leaders, and staff, producing what it calls hybrid technology firms. The release cited a global professional services market above $6 trillion and a forecast 11.2 percent CAGR in tax advisory services. Noah Pepper is chief executive. Multiplier subsequently raised a $35 million Series B led by The General Partnership in August 2026, with Ribbit and Lightspeed participating, and hired former Slack CFO Allen Shim as president and CFO.
Kaul discusses evaluating roll-up opportunities.
Introduces a permanent-capital vehicle to invest in, acquire, and operate businesses over long periods.
Explains the hybrid of technology development, serial acquisition, and permanent-compounder inspiration.
Organizes applied AI around intelligence, infrastructure, and workforce enablement, with emphasis on integration and organizational redesign.
Lists fragmented, knowledge-intensive, relationship-based markets as candidates and asks when owning customers beats selling vertical software.
Explains how embedded engineering and detailed workflow models can become reusable products. Extends the argument to integrated AI service providers.
Explores vertical SaaS buyouts as a response to technology adoption problems. Traces the argument to earlier Yoni Rechtman and Will Manidis social posts.