This practice serves three audiences. They look like different markets, and on a procurement-and-engagement basis they are. But they are answering different versions of the same question — what does it mean to put autonomous systems into financial markets — and the work travels meaningfully between them.
The builder’s question
The builder is asking: what does the system have to do, and what does it have to be able to defend? A prop desk standing up an analyst trinity, a family office wiring an MCP server, a trading firm narrowing the tool surface of its execution agent — they are all asking variants of an architectural question. Architecture is the easy part; supervisability is the hard part. The artifact a builder needs is a written description of the topology, the authorization granularity, and the decision log — not because the regulator is asking yet, but because the regulator will ask, and producing the artifact retroactively is harder than producing it concurrently.
The incumbent’s question
The incumbent is asking: what does our business look like in three to five years? A bank’s strategy group, a CIO office at an asset manager, a vendor whose client base is migrating from human-buyer to agent-buyer — they are all asking variants of a competitive-landscape question. The competitive map has three structural axes: where moats hold (distribution, supervisory standing, counterparty trust), where moats are eroding (commodity research, generic execution venues), and where new revenue surfaces are opening (agent-grade APIs, MCP-licensable products, machine-readable licensing). The artifact an incumbent needs is a written paper that names which of the firm’s revenue lines sit in which category, with the rationale to defend the categorization in front of a board.
The regulator’s question
The regulator is asking: what is the systemic-risk surface, and how is it written into supervisory frame? The Bank of England’s 2026 Financial Policy Committee record names herding as the operational primitive1. The CSA Staff Notice and CIRO Guidance Note are working the same surface from a Canadian registrant perspective. The SEC, CFTC, FCA, MAS, and IOSCO are tracking adjacent slices. Each authority is producing supervisory drafts faster than firms are producing the documentation against which those drafts will be applied. The artifact a regulator or institution needs is a working paper, a comment-letter draft, or an obligations-map that translates between the two clocks — the one supervision is on, and the one industry is on.
Why the work travels
The questions are different. The underlying analytic surface is the same. A working paper written for a regulator is read with profit by a builder; an architecture review written for a builder is read with profit by a strategy team; a competitive-landscape paper written for an incumbent is read with profit by a regulator’s research staff. The practice is structured so that a single analyst can hold all three views without losing the discipline appropriate to each.
This is the operational reason for the three-audience framing on the homepage: not because the work is generalist, but because the same well-built artifact can serve a builder, an incumbent, and a regulator if it is written for the question rather than the buyer.
What this means for an engagement
A first conversation with this practice will surface, briefly, which of the three questions is being asked. The artifact and the supervisory framing will be scoped to the one that is. The work will be aware of the other two — because they are not actually different questions — without being routed through them.
That is the entire premise. One artifact, scoped to a real question, written for the senior reader who has to act on it. Three audiences, one question, one engagement at a time.
Notes and citations
Bank of England Financial Policy Committee record, April 2026. See companion working note. ↑
On the regulator’s clock vs. industry’s clock: see CSA Staff Notice 11-348, CIRO GN-3300, and the FCA AI strategy materials.
On the agent-buyer market: see the companion note on incumbent disruption.