Coming soonRAG Pipeline & Agent Toolkit

    Turn object storage into
    an agent knowledge layer.

    S3 object storage with a built-in RAG pipeline. No stitching required, no per-query fees.

    Start storing for freeTalk to an expert

    No credit card required · No per-request fees · Connects in minutes

    You don't just pay to store your data.
    You pay for every step that makes it usable.

    Building a knowledge base on standard object storage providers means stitching together multiple services, each billed separately. Then your agents start running, and every query, memory read, and retrieval call adds to the tab. The stack is expensive to build and even more expensive to run.

    The architecture cost

    A typical RAG-enabled agent requires several services duct-taped together. A vector database, a retrieval layer, a compute service, and an object store, each with its own integration, its own failure point, and its own bill. You're paying for complexity before your agents run a single query.

    The usage cost

    A single agent task triggers dozens of retrieval calls: context lookups, memory reads, state writes. Multiply that across thousands of tasks per hour and the meter runs constantly. Every action your agents take is a billable event.

    A different model

    Fil One collapses the stack and the bill

    One platform for agent memory, RAG corpus, and retrieval. No glue code required. You pay for what you store, not what your agents do with it.

    Memory. Corpus. Traces. One endpoint.

    Standard S3 PutObject and GetObject cover every agent storage pattern. No new SDK, no per-query pricing. Flat storage for the full knowledge layer.

    agent_storage.py
    import boto3, os

s3 = boto3.client(
    "s3",
    endpoint_url="https://eu-west-1.s3.fil.one",
    aws_access_key_id=os.environ["FIL_ACCESS_KEY"],
    aws_secret_access_key=os.environ["FIL_SECRET_KEY"],
    region_name="eu-west-1",
)

# Agent memory — persist across restarts
s3.put_object(
    Bucket="agent-store",
    Key=f"memory/{agent_id}/session-{session_id}.json",
    Body=json.dumps(memory_state),
)

# RAG corpus — store chunked documents
s3.put_object(
    Bucket="agent-store",
    Key=f"corpus/{namespace}/doc-{doc_id}.txt",
    Body=chunk_text.encode(),
)

# Retrieve for augmentation — $0 egress
obj = s3.get_object(
    Bucket="agent-store",
    Key=f"corpus/{namespace}/doc-{doc_id}.txt",
)

# Inference outputs — archive completions for eval
s3.put_object(
    Bucket="agent-store",
    Key=f"traces/{run_id}/{step}.json",
    Body=json.dumps(trace),
)

    Storage that works the way agents do.

    S3-compatible storage built for agents is live today. Be the first to try native RAG pipeline and AI agents integrations.

    Agent memory & state

    Persist conversation history, task queues, episodic memory, and checkpoint files across agent restarts. Standard PutObject/GetObject — the agent writes, the agent reads.

    Flat cost for loop traffic

    Agents write frequently and read back their own outputs. Per-request billing makes loops expensive. $4.99/TB flat — no PUT fees, no GET fees, no egress.

    RAG corpus storage

    Coming soon

    Store raw documents, chunked text, and embeddings backing a retrieval pipeline. Reads are included in flat storage — no per-retrieval egress counter.

    Join the waitlist

    AI toolkit integrations

    Coming soon

    LangChain, LlamaIndex, and Haystack connectors for direct corpus management, plus agent memory integrations.

    Join the waitlist

    One rate. $4.99/TB/month.

    Storage only — no PUT fees, no GET fees, no egress. Try it free with 1 TB: one place to store your data, retrieve it, and put it to work.

    Start for freeTalk to an expert

    No credit card required · No per-request fees · Connects in minutes