COMPANY
Alef Labs
Alef Labs builds Alef, a decision intelligence platform for enterprises. Alef records what an organisation knew at the moment a decision was made, who had the authority to act, what was approved, what actually executed in the systems of record, and what resulted. It is built for teams whose decisions carry money, customers, or regulatory consequence.
WHAT WE BUILD
One governed record, instead of five systems that disagree.
In most companies the evidence behind an important decision sits in one system, the authority to act in another, the action itself in a third, and the result in a fourth. Reconstructing what happened, and why, is manual work that arrives too late to matter. AI agents make more of these decisions, faster, which makes the gap more expensive.
Alef puts those pieces in one place. Permissions are enforced in storage before anything is retrieved, so what a person or an agent can see is decided by policy rather than by hope. Facts stay separate from predictions. Every claim carries its source. Actions require human approval, and every approval leaves a durable receipt with a stated reversal condition.
The product runs inside the customer's own network, including on premises, so sensitive evidence does not have to leave their boundary.
OPEN SOURCE
The storage engine is public.
Alef is built on ArcGraph, our own graph and vector storage engine written in Rust, with permissions, time-aware history and provenance held in the engine itself rather than bolted on above it. The core is open source under Apache 2.0 and can be read, built and run by anyone.
COMPANY DETAILS
Where we are today.
- What we do
- Decision intelligence software for enterprises
- Founded
- 2026
- Location
- India, working with customers remotely
- Stage
- Early stage. The platform is in active development and we are running technical evaluations with prospective customers.
- Open source
- github.com/aleflabs/arcgraph, Apache 2.0
- Website
- aleflabs.ai
- Contact
- hello@aleflabs.ai
CONTACT
Talk to us.
Email hello@aleflabs.ai, or book a call directly. If you would rather bring a real problem than a question, request a technical evaluation from the home page.