Running your own AI? Archive it.

Self-hosted AI doesn't mean unregulated AI. Arc Bridge connects to your internal AI infrastructure and archives every interaction - same compliance controls as enterprise platforms.

Get Started Request Demo

Self-hosted integrations

Arc Bridge adapts to your infrastructure. Proxy pattern for API-based tools, DB connectors for UI platforms.

LiteLLM

Live

Proxy integration - Arc Bridge sits behind LiteLLM and captures all model calls across providers.

Ollama

Live

Proxy pattern - intercept local Ollama API traffic for archiving without touching your model setup.

Open WebUI

Connected

DB connector pattern - archive conversations directly from Open WebUI's data layer.

vLLM

Coming Soon

Proxy pattern - transparent interception for high-performance vLLM inference servers.

Text Generation Inference

Coming Soon

Proxy pattern - archive HuggingFace TGI model interactions.

OpenRouter

Coming Soon

Gateway integration - capture routed model calls across multiple providers.

Two integration patterns

Proxy pattern

For API-based tools (Ollama, vLLM, LiteLLM). Arc Bridge acts as a transparent proxy - swap the endpoint URL and all traffic is captured.

Client → Arc Bridge → Ollama → Model

DB connector pattern

For UI platforms (Open WebUI). Arc Bridge connects to the application's database and exports conversation records to the archive.

Open WebUI → DB ← Arc Bridge → Archive

Frequently asked questions

How does Arc Bridge archive LiteLLM traffic?
LiteLLM has a callbacks system - Arc Bridge ships a LiteLLM callback that fires on every completion. The callback captures the prompt, response, model, user, and timing data and writes it to the Arc archive. Configuration is a single line in your LiteLLM config.yaml.
Does Arc Bridge work with Ollama?
Yes. Arc Bridge sits in front of Ollama as a transparent proxy on its OpenAI-compatible endpoint. Point your clients at arc-bridge:11434 instead of ollama:11434, and Arc Bridge forwards every request to Ollama while archiving the traffic. No Ollama config change required.
What about Open WebUI?
Open WebUI talks to OpenAI-compatible backends (Ollama, LiteLLM, vLLM, OpenAI). Point Open WebUI at Arc Bridge instead of those backends directly. Same proxy pattern as Ollama - archiving is transparent to Open WebUI and to your users.
Does any data leave my network?
Not unless you point it to. Arc Bridge runs entirely inside your network. Archived records are written wherever you configure - your S3 bucket, your Postgres instance, your filesystem. The hosted Comma archive is one option; self-hosted is another. Same encryption either way.
How is this different from just turning on Ollama / LiteLLM request logs?
Native logs are unstructured, mutable, and rotate. Arc Bridge writes structured records with user identity, conversation linkage, hash-chained tamper-evidence, and tenant-isolated encryption. The difference is "we have logs" vs. "we can prove what happened to a regulator."

Your models, your infrastructure, our archive

Keep full control of your AI stack. Arc Bridge just makes sure the records are there when you need them.

Get Started Request Demo