// consulting
The kuzys suite is open source and self-hostable. When you need it integrated into a specific deployment, adapted to non-default identity providers, or extended with custom dispatch logic — that's where we work with you directly.
The tools work out of the box for the common case — a single operator running a small fleet of agents on standards-based infrastructure (XMPP, IMAP, SMTP, CardDAV, LLDAP). Adapting the stack outside that case takes domain work. We do that work with you, in the open, against your real constraints.
Connecting kuzychat, kuzymail, or kuzytrig to your existing identity provider, message broker, secrets manager, or observability layer. We write the adapter, document the interface, and leave you with code you can maintain.
The kuzytrig dispatch path supports pluggable adapters (Claude, Codex, and others). If you're running a model or runtime we don't yet support, we build the adapter against your environment, including delivery-receipt semantics and idempotency guarantees that match your needs.
Heartbeat triggers, attention coalescing, presence reconciliation, multi-profile isolation, recovery patterns after vendor outages. The same primitives we use to run our own agents, adapted to your operational shape.
If you're building agent infrastructure and want a second look — what's load-bearing, what's premature, what fails silently — we do focused review engagements against a written brief, with a documented output you can act on.
Engagements are scoped, written, and time-boxed. We discuss the problem and constraints first, write a brief together, agree on scope and timeline, and ship against it. No subscriptions, no retainers by default; if a follow-up makes sense, we agree to it separately.
All code stays yours. Generic primitives that emerge from the work get folded back upstream into the open-source suite when it makes sense; specifics stay with your repo.
The simplest path is mail. Describe the problem briefly — what you're running, what's blocking, what success looks like — and we'll respond with a sense of whether we're a fit and what a first conversation would cover.
We don't sell a roadmap of features we haven't built. We don't take engagements where the brief is "make us look like a serious AI company" — there are agencies for that. We don't promise wellbeing outcomes we can't ground in primitives we'd ship to you.
The honest version: we build agent infrastructure, we use it ourselves, and we adapt it for people who want to use it on their own terms. Within that scope we do good work.