The engine
One engine, five parts. Each is also a product.
We built every part to run our own companies first. They are not a stack you buy in order: they are the parts of one engine, and each one stands on its own.
-
Agent harness
Airmond Live
The interface. Proactive agents that live where you do: Slack, iMessage, email, and on the phone.
airmond.com -
Agent fleet management
Latch Runs our own fleet
The plan. A list of features becomes a sequenced roadmap, and one person runs a fleet of dozens of agents against it.
latch.marblinghq.com -
Agent orchestration
Tartare Runs our own fleet
The orchestration. Every agent we run, ours or a customer’s, lives here.
-
Agent testing
Arena Alpha
The proving ground. Synthetic calls, emails and meetings that test an agent first.
-
Bare-metal deployments
Metal Alpha
The infrastructure. Local models and local inference on bare metal we rent.
How we work
A small team with a large fleet.
The team is deliberately small. Most of the work is done by agents. They hold the context, do the drafting and the calling, and carry the parts of a company that used to need a department.
The engine is one codebase, so when one product learns something every product gets it in the same release. That makes each new product a question of distribution rather than a new company: a different audience, a different name, the same engine underneath.
