04 · BUSINESS AUTOMATION
Automation that removes the handoff, not just the typing
Most automation projects fail because they automate a process nobody mapped first. We start with the map, which routinely shows that the expensive step is a handoff between two teams rather than the work either of them does.
- Process Mapping
- Document Intelligence
- Voice & Chat Agents
- Systems Integration
What we build
Four layers, usually in this order.
Process mapping
What actually happens, as opposed to what the documentation says happens. This step regularly changes the scope of everything after it.
Document intelligence
Extraction and classification over the forms, invoices and contracts that currently move by hand, with uncertain cases routed to a person rather than guessed at.
Voice and chat agents
Front doors that can answer, qualify and escalate. Built to hand over to a person cleanly, because the escalation path is what determines whether anyone trusts it.
Systems integration
Durable workflows across the tools you already run, with an audit trail for every step and retries that are safe to retry.
A person stays where judgement matters
Every system we ship keeps a human in the loop at the points where being wrong is expensive. Automation that removes the reviewer is usually automation that moves the cost somewhere less visible.
2 of these are live right now
Not a portfolio of screenshots — every one of these opens.
Maiku AI
Real-time interview copilot
A desktop overlay that transcribes the interview live with Groq Whisper and drafts structured answers with LLaMA while the call runs on Zoom, Meet or Teams. The window is registered with the Windows Display Affinity API, so it is physically absent from captured frames rather than hidden behind another window. Core is open source and runs on the user's own Groq key.
maiku.app ↗Cartesia Assistant
Low-latency voice assistant
A call-shaped assistant built on Cartesia speech synthesis with streaming turn-taking, so a reply starts speaking while the rest is still being generated. A live waveform shows who holds the turn, and users choose from a roster of voices before the call starts.
cartesia-assistant.vercel.app ↗
What we build on
- Temporal
- n8n
- Twilio
Halyx is not religious about the stack — it picks what your team can maintain — but these are the defaults it is fastest and safest in.
Questions we get asked first
- Where does automation usually pay back fastest?
- At handoffs between teams and at any step where a person retypes data that already exists in another system. Process mapping finds these; guessing does not.
- Will this replace roles?
- Our position is that AI should amplify people rather than replace them, and we build to keep a person in the loop where judgement matters. If the goal of an engagement is purely headcount reduction, we are probably not the right studio.
- Can you integrate with our existing tools?
- Usually. Durable workflow engines and the vendors' own APIs cover most of it; the honest answer for anything exotic comes after a short assessment.
The other four practices
Talk to the people who would build it
The first conversation is a scoping call, not a pitch: what you are trying to move, what already exists, what the constraint is. It ends with either a range and a proposed first step, or a straight answer that Halyx is not the right fit.