Question 1

What is Engram?

Accepted Answer

Engram is a decentralized AI memory layer built on Bittensor subnet 450. It lets you store text, images, and PDFs as content-addressed vector embeddings that are replicated across a network of incentivized miners and permanently archived on Arweave.

Question 2

How does Engram differ from Pinecone or Weaviate?

Accepted Answer

Unlike centralised vector databases, Engram has no single point of failure or central authority. Data is stored across multiple miners on the Bittensor blockchain, each of whom must cryptographically prove they hold your data. There is no monthly subscription — miners are paid in TAO tokens by the network.

Question 3

What is a content-addressed CID?

Accepted Answer

A CID (content identifier) is a SHA-256 hash derived from the vector embedding and metadata of your stored content. The same text always produces the same CID, regardless of which miner stores it — making storage verifiable and tamper-proof.

Question 4

How do I store data on Engram?

Accepted Answer

You can use the Python SDK (pip install engram-subnet), the CLI, or the web playground at theengram.space/playground. Each method returns a permanent CID you can use to retrieve the data later.

Question 5

What is Bittensor subnet 450?

Accepted Answer

Subnet 450 is Engram's slot on the Bittensor network. Bittensor is a decentralized machine learning network where miners and validators earn TAO tokens for running useful AI services. Engram's subnet uses those incentives to ensure permanent, verifiable vector storage.

Try Engram

Store Text

Semantic Search