Morpholog
The AI security and compliance core for regulated industries. Open source. An agent's proposed action becomes a business fact only when it meets the firm's written rules, the person's authority and the evidence in hand. Every decision leaves a record a third party can check.
Morpholog on GitHubWhy it exists
More and more of what writes to a trade book is software you did not write. Optimisers, models and AI agents book, amend and settle trades. With Morpholog, none of them writes directly. Each proposes a change, and the change is saved only if every rule still holds.
What it does today
Rules checked at the moment of saving. A limit breach or a booking in a closed period never reaches the book.
Refusals in business words. A refusal names the rule and the values that failed it, so an agent can fix its proposal.
A tamper-evident log. Who proposed each change, what changed and which rules were checked. A check proves the history has not been edited.
The books on any past date. A position or a P&L as it stood, under the rules in force then.
Evidence a third party can verify offline. An auditor checks the exported history on their own machine, with no access to your systems.
A new rule tested against history. See which past changes it would have refused before you enforce it.
What it is not
Not a trading system. Your trading system stays the system of record. Morpholog guards the small part that must be defensible: the trades, the books, the filings.
Not an AI agent. It does not decide what to trade. It decides what may be saved.
Not a substitute for your rules. It enforces the rules your risk and compliance people write.
Open source
Apache 2.0. Rust on PostgreSQL. Open source, in active development. Code and documentation are on GitHub.
Glasshouse is an open ETRM core for European power, built on Morpholog and in early development. It shows Morpholog at work on a real trade lifecycle.
Running it in production, or about to? Talk to us about support.