Skip to content

The platform

What every module runs on.

Three things come with the platform: a workspace for your team, a data hub for your sources and a governed knowledge base. You activate them once, and every module is built on them.

How it fits together

Three foundations, then modules.

The foundation is the same for every customer and is included. What you commission sits on top of it.

The platform as a stack of four parts. Decision modules, which you commission, sit on top of the knowledge base and the data hub. Those two sit on top of the workspace. The knowledge base, the data hub and the workspace are included.

  • Commissioned

    Decision modules

    Each one prepares one named decision for the person who makes it.

  • Included

    Knowledge base

    What your business knows: objectives, policies, limits, decisions and their reasoning.

  • Included

    Data hub

    What your systems hold, connected one data domain at a time.

  • Included

    Workspace

    Where your team works: users, roles, permissions and the audit trail.

Modules read from the knowledge base and the data hub. What your team learns from each decision goes back into the knowledge base.

The knowledge base

Every change is a proposal until a person approves it.

The knowledge base is where your business's intent is written down and kept current. People and AI assistants can both propose changes. Nothing becomes part of it without review.

Reformance
Inbox
3 changes waiting for review
  • Edit
    Treasury policy: liquidity buffer
    Raises the minimum buffer from three to six months of capital calls.
    Review
  • New
    Decision: Manager C placed on the watch list
    Records the committee's decision and the reasoning behind it.
    Review
  • Edit
    Mandate: emerging markets equity
    Adds the limit on single-country exposure agreed in September.
    Review
Nothing is searchable until a named person approves it.
The review Inbox. Illustrative content.

Proposed, then reviewed

A new entry or an edit arrives in the Inbox as a proposal, with a summary of what changed and a word-by-word comparison with the approved version.

Approved by a named person

A reviewer approves or rejects each proposal, and gives a reason when rejecting. A proposal that changed after the reviewer opened it has to be read again before it can be approved.

Versioned and audited

Nothing is edited in place. Every change is a new version, earlier versions are kept, and every approval and rejection is recorded.

Read by people and by AI

Your team searches the knowledge base by meaning as well as by keyword. An AI agent starts each task with the same approved knowledge, in the AI assistant your team already uses.

Knowledge health

The knowledge base checks its own health.

Automated checks run on a schedule and after every approval. They look for missing details, broken references, stale content, entries that nothing links to and duplicates. The results combine into one health score.

Loop
Vision → intent → capability
View loop graph
Vision
6
strategic anchors
Intent
28
planned work
Capability
19
shipped in production
In flight19 intents in flight to capability
Knowledge health
Last maintenance: 9m ago — completed
93EXCELLENT
  • Completeness100
  • Integrity99
  • Freshness72
  • Connectivity99
The knowledge base dashboard in Reformance's own workspace.
Completeness
Every entry carries the details that make it findable.
Integrity
References between entries lead somewhere.
Freshness
Content that has gone a long time without review is flagged.
Connectivity
Entries are linked to the ones they depend on.

Workspace and data hub

Your team, your systems, one place.

Workspace

A web application for your team on your own domain.

  • Users and roles, with permissions per module and per action
  • A full audit trail of access and actions
  • The same permissions apply to AI agents

Data hub

Connects to the systems you already use and keeps the provenance of every record.

  • Accounting, portfolio and booking platforms, reports and document stores
  • Grows one data domain at a time, each paired with the module that uses it
  • Your existing systems stay in place

How modules are made

Assembled from proven building blocks.

A module is specified in your knowledge base and assembled from building blocks that already run in production. Reformance's engineers direct AI agents that do the assembly, and review what they produce.

That is what we mean by self-writing: the system is produced from your business's own knowledge, and a change to the system starts as a change to that knowledge.

Operations

Data management, dashboards, metric tracking and workflow tools.

Reporting and documents

Reporting packs, generated documents and notifications.

Integrations

One integration layer connecting the systems you already use.

Access control and audit

Role-based permissions per module and per action, with a full audit trail.

Intelligence

Your institutional knowledge, structured, governed and readable by people and by AI.

Portals

Interfaces for boards, partners or investors, when the need arises.

Privacy and deployment

Where it runs is your choice.

Sovereignty comes from where the platform runs and which models it uses. Both are yours to choose.

Your options
Where the platform runsOur cloud, your private cloud, or your own servers, including air-gapped
Which AI models it usesA model provider we manage, your own provider, or models hosted privately for you
What stays the sameYour knowledge base, your modules and your permissions. Moving between options needs no re-architecture

Our access to your knowledge

When the platform runs on our cloud, our own access is controlled the same way yours is.

  • Access needs a grant

    Reformance staff, the founder included, need a time-limited access grant to open your knowledge base. Each grant states its reason.

  • You see it and can end it

    Every grant is visible to you, and you can revoke it.

  • Every read is logged

    Each time our staff read your knowledge, it is recorded in a log you can open.

Which decision would you improve first?

Schedule a 30-minute conversation. Bring one decision your team makes on incomplete or late information. We will walk through what the module for it would need: the data, the knowledge, and what stays with your people.