Keep your logs. Lose the bill.

Your team shouldn't have to choose between a 7-day retention window and next month's cloud bill. Now they don't.

Request early access

Private beta. Built for teams who outgrew their log bill.

Integrations

Drop-in compatible

If you already forward logs to a collector or SaaS, you are probably minutes from pointing the same pipeline at Logfold.

OpenTelemetry

Ship OTLP logs with standard exporters, nothing proprietary wrapped around them.

Fluent Bit

Point existing forwards here: the same pipelines you already run.

Loki / Grafana

Use Logfold as a Loki datasource; panels and Explore carry over.

Ingest
Logfold
compress index query

Same wires, new termination point.

Where it runs

Your infra.Your buckets.

  • VM
  • Serverless
  • Kubernetes

Logfold runs on a VM, a serverless container, or Kubernetes. It writes compressed blocks to S3-compatible buckets (AWS S3, Cloudflare R2, MinIO, GCS with interop).

Look things up without unpacking the whole archive. Logs stay useful, not stuffed in a folder you never search.

The problem

Why log costskeep growing

  • Usage-based bills
  • Shrinking retention
  • Cheap backups

Most hosted log tools charge for volume and for how long you store data. Traffic rises. The bill rises with it. Your configs are fine. The logs just never stop.

So teams cut retention to placate finance. Thirty days becomes two weeks. Then someone needs last quarter for audit. Or you're chasing a bug from last month's ship.

The usual workaround is dumping older logs to cheap cloud storage. Saving money feels great until you need answers buried in those files.

That's a safety net, not something people can search when something breaks.

Architecture

How Logfold works

  1. 1
    Ingest

    Send logs over paths you already use. OpenTelemetry logs and Fluent Bit forwarding work without custom adapters.

  2. 2
    Compress on write

    Ingest compresses streams into a columnar block format in Rust. Blocks land in any S3-compatible store: AWS S3, Cloudflare R2, MinIO, GCS via interoperability. No storage vendor lock-in.

  3. 3
    Query compressed data

    Reads scan blocks without decompressing the full dataset first. Filters and time ranges hit structured storage, not a pile of raw lines on disk.

  4. 4
    Grafana as-is

    Logfold exposes a Loki-compatible API. Point an existing Grafana datasource at Logfold. Dashboards and Explore behave like they did against Loki.

Storage

Save spaceand keep search

Zipped backups often win on file size alone. Searching them usually means unpacking a lot of data first. Logfold keeps logs compact and lets you look inside without opening the whole pile each time.

74% smaller than raw uncompressed logs, indexes included. Still fully searchable.

Zipped archive Logfold
Finding answers Slow: unpack large files to search. Search while data stays compressed.
Where it lives Cheap cloud folders. Your buckets: same idea, your keys.
How logs arrive Often manual dumps. Streams in live from your pipeline.
Years later Affordable, easy to ignore. Still affordable, and still findable.

AI & retrieval

Built for the AI era.

  • Agents
  • RAG
  • Long context

LLM agents and RAG pipelines need enormous working context. Logfold lets you query terabytes of historical logs to ground those workflows—without paying active-indexing premiums on every byte you might someday ask about.

The diligence story is already familiar: compound context from real production history, priced like archive storage, reachable when agents and humans need it.

Early access

Want in? Say where to write.

We onboard testers in small waves. One email, no booking link circus.

Join the next wave

No pitch decks or sales calls. We only email when there is a seat.

You are on the list. We will write when the next batch opens.