North Star of UtahLeadership
Cody Johnson is CEO of North Star Intelligence and leads its decision intelligence application, North Star Vision. He works from North Logan, Utah.
Cody builds software for owner-led service businesses where decisions get lost between a phone call and a finished job. North Star Intelligence reads the signals a business already produces, scores the decisions that matter, gives each one an owner and checks the outcome.
He also runs EchoCody AI, the first system attached to North Star Intelligence. EchoCody handles the customer side: calls, follow-up and the record of what happened. North Star Intelligence decides what matters and verifies whether it got done.
Founding engagements are handled directly by Cody. There is no sales floor and no handoff to an account team.
Every decision has one owner
A decision with no named person behind it tends to slip. One owner and one due date come first.
Done is not the same as verified
A finished task only counts once the outcome is checked against what was expected.
Show the evidence
Every score and every action should point back to the signal that justified it.
Hand off when unsure
Software should escalate uncertainty to a person, not hide it.
Intelligence first, systems attach
North Star Intelligence is the layer that decides and remembers. EchoCody AI attaches to it and carries the work into live conversations.
Cody, who is based in North Logan, Utah, was quoted in this coverage in his role as founder of EchoCody. The article does not cover North Star Intelligence.
Founder coverage · LA Voice ·
Businesses Set New Rules for Responsible AI Use
By William Jones, LA Voice
LA Voice includes Cody Johnson, founder of EchoCody, in its coverage of responsible AI adoption. His contribution focuses on setting automation boundaries, preserving human accountability and being transparent with customers.
Research direction
North Star Intelligence is the intelligence layer. EchoCody AI is the first system attached to it, carrying decisions into live customer conversations. The research question is how to keep that control layer trustworthy in real time.
Keep one shared state in two languages
A conversation can move between English and Spanish. The goal is one record of what was said and decided, whichever language was used.
Tie every action to evidence
Each step the system takes should point back to the signal or statement that justified it, so a person can check why it happened.
Verify the work actually happened
A task marked done is not proof. The goal is to confirm execution against the record before an outcome counts as verified.
Escalate when unsure
When the system is not confident, it should hand the decision to a named person instead of guessing.
Do this without slowing live voice
These checks should run alongside a live phone conversation without adding noticeable delay for the caller.
Status: these are design goals and research hypotheses. The mechanisms are not independently verified, and no latency or accuracy results are published. This is not a claim of funding, endorsement, patents or certification.