Every company keeps the same knowledge in its own systems, at its own cost. Nobody asked for a grid until there was one.
Before the electricity grid, every factory generated its own power. The grid made power something you connect to.
Companies still keep the same knowledge in separate systems, at great cost. Legra is the grid: knowledge becomes something you connect to.
The shipping container standardized trade. Knowledge still travels as cargo did before it: copied into every system that needs it.
1
At Legra's core is a graph database. But the world's knowledge cannot live in one database or one system. Silos exist for good reasons: ownership, regulation, accountability, and limits of size and speed.
So Legra divides knowledge into workspaces. A workspace is not a database in one place. It can be anywhere, replicated across thousands of nodes in the Legra network, and it stays under its owner's control.
Under control means encrypted. A workspace works like a Signal or WhatsApp group: everything in it is encrypted end to end, and only its members hold the key. The nodes that store your data never see it unencrypted, and neither does Legra. The one exception is the host that answers your queries: it must decrypt, so it is a member of the workspace, and you choose it: your own hardware, or an operator you trust. Legra's own Messages app works this way: every message of every participant lives, encrypted, in one shared workspace.
2
Tim Berners-Lee, who invented the web, also described a Giant Global Graph: all the world's data, linked. Legra is that graph, divided into workspaces that each have an owner and rules.
3
Many believe that once knowledge is captured as data and linked, AI can take it from there. It cannot. Much of what a business knows lives in code, and when the people who wrote it leave, nobody knows how it works.
A Legra workspace holds data, documents, and the use cases that run on them: a query, a piece of code, an API to call. Others can run a use case in your workspace without your data leaving it.
4
Coding agents work within a practice built over decades: shared repositories, version history, automated tests, security checks, peer review, and continuous delivery. People and agents contribute to the same body of work, check their changes, and build on accepted results. That discipline helps turn AI’s speed into useful, dependable output.
Legra’s mission is to bring those practices to law, consulting, and every other industry. One or more workspaces hold the shared body of work: structured data, documents, rules, decisions, tests, and deliverable artifacts. People and agents propose changes, validate them, review them, and merge accepted work.
Inherit, work, compound. Each session starts from accepted knowledge. Its reviewed contributions become the starting point for the next. Improvements accumulate instead of disappearing into separate conversations.
The ambition is to achieve the same step change in productivity and quality beyond software development.
5
Transactions with legal consequences need more than good data. They need proof that nothing was changed after the fact, and a clear record of who did what, when, and why. Legra provides both: Merkle trees, which make tampering detectable, and a full history and journal of every change.
6
Storage, replication, encryption, backups across regions, recovery, high availability: in Legra these are standard. A two-person company and a bank get the same infrastructure, so they can do business as equals.
7
AI will never be fully predictable, and neither are people. What makes it safe is context and precise instructions. Legra works with any AI service and brings its own harness: an agent works as a persona, on one use case, through stories that say exactly what to do, with exactly the data that job needs, every step on record. Thousands of them can work at once.
Deduction helps too. Legra's own reasoner derives what the facts and rules imply, every time, with proof: deductive AI, beside whatever model you use.
8
Together, the workspaces form one economy. Data products are the supply; use cases and apps are the demand. Apps connect to the data products they need and pay in tokens. Whoever offers good data or services earns tokens. The value of data is found the way any value is found: by what people pay for it.
Each does one part. None covers the whole exchange, from agreement to proof. Legra does not fit an existing category. See the comparison.
Legra uses open web standards throughout. Nothing is locked in. Each workspace is also a pod.
The plus marks where Legra goes beyond the published standard.