AIRCLERK / ENGINES

Choose your engine.

Company Brain runs the same abilities on the engine that fits your firm. Pick fastest-to-value, or the most powerful, regulated-ready memory. Here’s an honest, side-by-side comparison.

gbrain

Fastest to value

Hosted, set up in minutes.

Browser-only, multiplayer team brain with a markdown-first memory. Strongest at scheduled “watch” reports and writing in your voice. Hosted by the vendor, or self-host the open-source engine.

Setup: Minutes — hosted, browser-only
Memory model: Markdown files in git
Best at: Scheduled watches + writing
Multiplayer: Yes — shared team brain
Hosting: gbrain.io hosted, or self-host OSS
Data residency: Vendor-hosted by default
Pricing lineage: ≈ US$200/mo hosted; OSS free + your LLM cost

Highlights

Up and running in minutes — browser only, no infrastructure
Multiplayer: one shared brain the whole team reads and writes
Markdown-in-git source of truth — fully portable, nothing locked in
Rich abilities catalogue: scheduled watches, research, writing
Tiered enrichment for people and companies

Worth knowing

Single-operator by design — isolate one brain per client for regulated work.
Hosted by default, so weigh data-residency needs before connecting sensitive sources.
Benchmarks are vendor-published; validate on your own data during a pilot.

Hindsight

Most powerful / regulated-ready

Structured memory that forms beliefs.

Retain, recall, reflect. Hindsight forms observations and beliefs, detects contradictions, and answers temporal “what changed” queries with four-way retrieval (semantic + keyword + graph + temporal). Run it on Hindsight Cloud or self-host in your own environment.

Setup: More involved — cloud or self-host
Memory model: Observations + beliefs (graph + temporal)
Best at: “What changed”, contradictions, multi-client
Multiplayer: Yes — with tag isolation
Hosting: Hindsight Cloud, or self-host (MIT)
Data residency: Own-cloud / self-host option
Pricing lineage: Token-based: Retain US$15/M, Recall US$0.75/M, Reflect US$3/M

Highlights

Forms beliefs and observations automatically as it learns
Contradiction detection — flags when new facts conflict with old
Temporal queries: “what changed since…”, point-in-time recall
Four-way retrieval: semantic + keyword + graph + temporal
Self-host path (MIT, SOC 2, tag isolation) for data residency

Worth knowing

More setup than gbrain — best when belief-formation or residency justify it.
Token pricing means cost scales with volume; model it during a pilot.
Benchmarks are vendor-published; validate on your own data.

Either engine is model-agnostic

Company Brain doesn’t replace the AI assistant your team already uses — it connects to it over MCP. Whether your people work in Claude or ChatGPT, the brain reads and updates itself as they go, so it gets smarter with use no matter which engine sits behind it.

Which should I choose?

Answer three quick questions and we’ll point you at the engine that fits. You can always change it later.

1.How fast do you need to be live?
2.Do you need data to stay in your own environment?
3.What matters most day to day?

Still deciding? Start anyway.

You can switch engines later — your sources and abilities come with you.