Knowledge your AI can use. In a format you control.

Score encodes what your organisation knows into structured files that any AI can read.

The skill file

The building block of Score is a skill file.

A skill file captures one piece of organisational knowledge: how to handle a particular type of request, what standards to apply in a given context, how to represent your organisation in a specific situation. Each skill file has two parts: a short header describing what the skill is for and who approved it, and a body written in plain language telling the AI what to do.

Skill files are plain text, stored in a system your organisation controls, readable and editable by anyone on your team without specialist tools. Every change is tracked and every version is recoverable.

How skills are used

When a request comes in, the right skill is identified and provided to the AI alongside it. The AI reads the skill and responds according to the knowledge and standards it contains.

Your organisation decides what the AI knows. The skill defines the boundaries: what to say, what to avoid, what standards apply, what the response should look like. The AI executes within those boundaries.

Because the skill file is separate from the AI tool, the same knowledge works across any model. Change the model and the skill travels with you. The knowledge your organisation built belongs to you, not the vendor.

Governance built in

Every skill file carries metadata: who authored it, who approved it, when it was last updated, and what it is permitted to do. That information travels with the skill wherever it is used.

The question of what your AI was permitted to say on a given date has a verifiable answer: a record, in a file, in a version-controlled system. For organisations in regulated industries, that is the governance most AI deployments are currently missing.

Designed to scale

A single skill file is useful. A library of skill files, organised and governed, is infrastructure.

Organisations typically start with one or two skills covering their highest-value or highest-risk AI interactions. As confidence grows, the library expands. Skills are authored, approved, and retired through the same process, governed consistently as the organisation's needs develop.

Score is designed to support a governance layer on top of this process, managing approvals, routing requests, and maintaining the audit trail, without requiring it from the start. Organisations can begin simply and add structure as their needs grow.

Two paths