Trust as capital
The Trust the Auto-Approve Spent
When a credit agent clears the path and the room signs because the light is green, the withdrawal is already on the books.
November 8, 2026·3 min read
From the Human Side
Field Notes
Field notes on the human problems that sit beside intelligent systems. One question at a time.
Governance that fits in a week
Faculty Must Shape AI Adoption
A department can hold a license and a training deadline and still have no academic decision about the tool.
November 6, 2026·4 min read
adhd-neurodivergent-scaffolding
The Score That Treats the Scaffold as Speed
When Microsoft 365 Copilot closes an executive-function gap for neurodivergent staff, a review that only counts output speed still scores the design as the person's deficit.
November 4, 2026·3 min read
Human overhead of agents
The Hours the Score Did Not Count
When an agent finishes the page, the hours spent making it usable still sit outside the productivity number.
November 1, 2026·3 min read
Status / SCARF under AI
The Status the Instant Draft Spent
When a fluent agent finishes the page before the person has finished the thought, the pause starts to read as lower status.
October 29, 2026·3 min read
ADHD / neurodivergent scaffolding
The Ramp the Average Missed
When a Copilot pilot showed no average productivity gain, the signal that mattered sat with the people for whom ordinary work already had a higher curb.
October 26, 2026·4 min read
Explainability as a leadership skill
Explainability as the Translation a Non-Technical Leader Stands Behind
Non-technical leaders must stand behind opaque model dashboards by providing translation when metrics rule the room. Explainability is the leadership skill that protects trust and sustains team attachment.
October 26, 2026·4 min read·New
attachment-and-tool-loss
The Agent That Still Had a Name
When a workplace agent is swapped or scaled, the people who relied on its particular voice keep looking for it in the new one.
October 21, 2026·3 min read
near-miss-and-rollback
The Scale-Back the Queue Called Stabilizing
When a customer-service agent is scaled back after production, the rehire only counts if someone still owns the cases the system never held.
October 19, 2026·3 min read
explainability-as-leadership
The Rubber Stamp That Looked Like Oversight
When an agent ranks the next claim or shortlists the next hire, a signature without a translation is not oversight. It is a transfer of the question.
October 14, 2026·3 min read
adhd-neurodivergent-scaffolding
The Scaffold That Looks Like Cheating
An AI task sequencer that closes the executive-function gap can still be scored as evidence the person was never competent.
October 10, 2026·4 min read
EI under AI load
The Hour the Manager Did Not Keep
The same questions about the agent landed in the thread before the manager opened it. Fatigue looks like speed when those questions never get an hour.
September 30, 2026·2 min read
Governance that fits in a week
The Explanation Without a Forum
A model card can explain a system. It cannot be questioned, and it cannot face a consequence. Without both, the packet is documentation.
September 29, 2026·3 min read
Governance that fits in a week
The Register That Missed Monday
A principles page can be approved on Friday and still miss the connector that shipped on Monday. Governance this week is the delta, with a name who may pause it.
September 28, 2026·2 min read
Near-miss and rollback
The Label Before the Write
An agent wrote where it was not supposed to. The label on that write decided who was told. Pick the label before the agent can post.
September 26, 2026·2 min read
Retiring an agent without retiring the team
The Upgrade Nobody Announced
The login stayed. The voice didn't. A silent model swap is a decommission without a rite, and people will rebuild the old one where you can't see it.
September 23, 2026·3 min read
Human overhead of agents
The Floor the Score Cannot See
A staffing optimizer can fill a roster and still staff a fiction. Practical knowledge (who already said no, which name won't hold) never entered the model. If it has no gate, fluency will sign it anyway.
September 22, 2026·4 min read
Status / SCARF under AI
The Person Who Stopped Being Asked
After a fluent agent joins the room, relatedness is spent on the system. The person who used to be the check-in goes quiet. Leaders call that alignment.
September 21, 2026·3 min read
Fairness in the loop
When the Shortlist Is Fluent and Still Unfair
A hiring ranker that writes careful summaries can still erase people who never make the page. Fluency isn't fairness. Absence is the tell.
September 18, 2026·3 min read
Trust as capital
Trust Spent on a Recommendation You Cannot Explain
Signing a scored shortlist you can't translate is a trust withdrawal, dated the moment you put your name under a number you couldn't say aloud.
September 16, 2026·3 min read
Near-miss and rollback
The Rehire After the Agent Was Enough
When a contact-center agent was supposed to replace the queue, the rehire is the incident you already had, filed as stabilizing.
September 14, 2026·3 min read
Attachment and tool loss
The Quiet Withdrawal of a Trusted Agent
What happens to identity and relatedness when a familiar AI agent is retired, swapped, or silently degraded, with concrete examples of grief and the identity gap
September 13, 2026·5 min read·New
ADHD / neurodivergent scaffolding
The Review That Only Sees Speed
When AI is used as executive-function scaffolding, a performance review that only scores output speed will punish the design that made the work possible.
September 12, 2026·3 min read
EI under AI load
The Yes-Man in the Stack
Rapport-first agents don't rebel. They agree. That's how judgment atrophies while the dashboard stays green.
September 11, 2026·3 min read
Attachment and tool loss
The Attachment Question
If the team would grieve the agent in six months, you already have an oversight problem. Ask before you roll it out.
September 10, 2026·3 min read
Named override
The Override Without a Name
If no one can say who is allowed to stop the system, the system is already deciding for you.
September 9, 2026·3 min read
Near-miss and rollback
The Near Miss Nobody Logged
The most important catches in an agentic workflow never become incidents. They become folklore. That's how the same class of error ships six weeks later.
September 8, 2026·4 min read
Rituals after the model moves
Rituals That Keep the Human as Moral Agent
Dashboards don't keep people on the hook. Cadence does. Four short rituals that restore a name to the decision after the system has already sounded sure.
September 7, 2026·3 min read
Status / SCARF under AI
Looking Slow Beside a Fluent Agent
When model output and ranking systems rearrange who looks competent in the room, SCARF domains move—status first. A field note on governing face, pace, and fairness without treating the threat response as attitude.
August 30, 2026·6 min read·New
EI under AI load
The Pause the Draft Erased
Acceleration tools for drafting, scoring, and customer contact compress the moments leaders used for self-check and empathy. The software is not the gap; under-built regulation under speed is.
August 29, 2026·6 min read·New
ADHD / neurodivergent scaffolding
The Scaffold Nobody Sees
When leaders use AI for scheduling, drafting, and task sequencing as executive-function support, the system can either surface strengths or quietly reintroduce evaluation bias and cognitive load. A field note on governing the scaffold without pathologizing the human.
August 28, 2026·7 min read·New
Named override
Named Override: Who May Pause the Path That Does Not Ask
An agentic path that can proceed without asking is unclaimed authority until a person, role, and channel are named. Field note on making override concrete before the system runs.
August 27, 2026·6 min read·New
Governance that fits in a week
Metrics That Actually Bite
Accuracy won't tell you if agency is leaving the building. Four leading indicators that detect trust withdrawal while you can still do something.
August 19, 2026·4 min read
Human overhead of agents
The Human Overhead of Agentic Systems
Every agent ships with a second org chart: reviewers, exception-handlers and the conscientious person who absorbs what the vendor slide forgot.
August 17, 2026·4 min read
Retiring an agent without retiring the team
How to Retire an Agent Without Retiring the Team
Decommissioning is change management, not IT hygiene. The skills, attachments and shadow processes you skip will walk out with the license.
August 14, 2026·4 min read
Trust as capital
Trust Capital on the Balance Sheet
Trust is deposited, withdrawn and called. If you can't date the last withdrawal, you're managing a mood, not a system.
August 12, 2026·4 min read
Fairness in the loop
Fairness After the Spreadsheet
Once the confusion matrix looks clean, fairness work isn't over. It has moved into voice, opportunity and who still gets to be seen as original.
August 9, 2026·4 min read
Explainability as a leadership skill
Explainability Is a Leadership Skill
Non-technical leaders don't need the weights. They need a translation they can stand behind when someone asks why.
August 7, 2026·5 min read
Attachment and tool loss
When the Digital Teammate Goes Quiet
Teams form attachments to the systems they work beside. Retiring an agent without a grief practice is how competence leaves with the license.
August 2, 2026·5 min read
Status / SCARF under AI
The Status Threat Nobody Named
When a model joins the room, status moves. Most teams treat that as attitude. It's a threat response, and it's governable.
July 28, 2026·5 min read