The expert should know which device you clicked.
Why local BAS specialists need system context, useful tools, and a place in the technician’s existing workflow.
FROM THE FIELD. INTO THE CODE.
AI. MEET THE
MECHANICAL
ROOM.
I’m Russ. I put AI to work in building automation. Engineering tools. Field notes. Developing local experts for the systems behind the screen.
CONTROLS ENGINEER / AI BUILDER / @FSTACKED
THE TERRITORY
BACnetNiagaraDesigo CCLONHVAC controls
01 THE WORK
MORE SIGNAL.
LESS GUESSWORK.
AI earns its place when it helps with the actual job. This is where I share the tools, questions, and experiments behind that work.
[01][02][03]
Engineering, amplified.
Station analysis, documentation, diagnostics, and the practical work of putting AI into a controls workflow.
The work, in the open.
Technical notes on what works, what falls apart, and what I’m building next.
Expertise, kept close.
Developing focused local models that can work with building context, right where the engineer needs them.
02 LOCAL EXPERTS
SMALL MODELS.
DEEP BUILDING
KNOWLEDGE.
I’m developing focused subject-matter experts that can work with the technician’s existing tools, using building context and compute kept on site.
Inside the local expert programSPECIALIST / 001Research focus
BACnet
Device discovery. Communication faults. Configuration. The next check that actually moves the job forward.
- INTENDED RUNTIME
- Local workstation / site computer
- WORKFLOW DIRECTION
- Investigate. Explain. Prepare.
- PLATFORM CONTEXT
- Existing BAS tools
- RELEASE STATUS
- In development
Follow the research as the models and integrations take shape.
03 THE FIELD NOTES
FROM THE WORKBENCH.
Offline is a test you run.
A local model is only one part of an offline BAS assistant. The entire declared workflow has to survive without public internet access.
Show the evidence. Then name the next check.
A synthetic BACnet dropout scenario shows how a useful diagnostic answer separates observations, hypotheses, and the test that comes next.