Research

Open work on the questions agentic systems are forcing institutions to answer.

Working papers, technical notes, and short essays from the DSI research programme.

Areas of work
// 01

Agent accountability

Decision records, declared intent, and the burden of proof when an autonomous system acts on behalf of an institution.

// 02

Data liquidity

How meaning becomes the unit of access across legacy estates. The economics of cross-system retrieval under an ontology.

// 03

Cross-boundary trust

Portable credentials, key-as-principal trust, and the cryptographic primitives that let agent authority travel between organisations.

// 04

Regulatory infrastructure

What the AI Act, DORA, and adjacent frameworks imply for the structure of agent deployment in regulated estates.

Publications
FORTHCOMING

On dynamic ontologies for agent access in regulated estates.

A note on routing agent intent across legacy systems by meaning, and the consequences for retrieval, governance, and proof.
// data liquidity IN PREPARATION

Agent accountability under AI Act Article 12.

On reading Article 12 as a requirement that shapes the runtime itself. What the record-keeping obligation implies for the system that produces the record.
// regulatory infrastructure IN PREPARATION

The ledger as evidentiary record.

On producing a regulator-legible record as a property of the runtime itself.
// agent accountability IN PREPARATION

Metadata, not copies.

Why an agent runtime in a regulated estate should hold decisions and metadata while leaving the underlying data in place.
// cross-boundary trust