Hudson Valley Future Summit 2026 · AI Readiness

Deep Learning, Deeper Listening

Relearn how organizations actually work.

George Jensen  |  Qualitative Systems Research & Operational Ontology
Group 3 Session · November 6, 2026
From the Field · Harvard Business Review

How AI Agents Orchestrate Work Across Silos

Kris Johnson Ferreira and Jordan Tong · September–October 2026

A Little History of AI + ML

What led up to the deep learning movement ?

thomasosborn.org
Background : My Path

My focus on Information Science

Planned vs Actual: How do large companies close the gap?

Maturity · AI Practice, 2022 to Today

Six Stages of AI Practice

1

Magic Prompt

Late 2022

You can’t say where an answer came from.

2

Prompt Craft

2023

Quality depends on who wrote the prompt.

3

Referenced

2023–24

You can click through to the source.

4

Delegated

2024–25

You review work, not answers.

5

Owned Knowledge

2025–26

When someone leaves, their knowledge stays.

6

Accountable Orchestration

2026–

Every decision has an owner and a trail.

Stages 5 and 6: where this talk lives · A teaching framework, not a measured scale · Staircase format inspired by Nielsen Norman Group’s UX Maturity Model
On Staying With Problems
“It’s not that I’m so smart, it’s that I stay with problems longer.”
— Albert Einstein
Portrait of Albert Einstein
From Practice
AI doesn’t make us smarter.
It just makes us stay longer and deeper with a problem. So, in essence, it is more intense to work with AI.

Expect this before you expect any other benefit from AI.

Introducing: Deeper Listening Practice

  • Organizations were mostly dependent on the vertical HR org chart
  • That produces a problem of mission creep.
  • It's time to relearn the real processes behind knowledge work.

Deeper listening, paired with AI, can regenerate truly functional organizations.

Case Study: Deeper Listening Practice

Group 3 · About This Session

The Culture Shift Behind AI Readiness:
From Individual Contributors to Entire Organizations

AI readiness requires more than learning new tools, it requires a new way of thinking. The future belongs to those who don’t just use AI, but think bigger with it.

The common promise of AI is speed: faster work at lower cost. What gets less attention is what happens to the people doing the work. Thinking well together starts with deeper listening.

This session shows how teams can build that habit by relearning how their own organization really works to align with the organizational mission. We’ll explore the evolution from task-oriented AI users to strategic thinkers who challenge assumptions, connect ideas, and uncover opportunities.

From the Field · Harvard Business Review

About Workflows

Untangle Before You Automate

“Organizations cannot use AI to break coordination bottlenecks if their underlying decisions are opaque, entangled, and idiosyncratic.”

Redesign the Flow

“Companies will have to redesign how higher-level decisions flow from lower-level ones, how human-AI interactions are structured, and how workflows are integrated.”

Value Lives in the Links

“The most consequential decisions aren’t made within a single task; they emerge from linking together many narrowly scoped tasks.”

No One Holds It All

“No single individual’s brain can hold all the relevant information, constraints, and scenarios.”

Kris Johnson Ferreira and Jordan Tong, “How AI Agents Orchestrate Work Across Silos,” Harvard Business Review, Sept–Oct 2026
From the Field · Harvard Business Review

What People Know That AI Doesn’t

Knowledge Surfaces Late

“Each function knows things that others do not, and often people cannot articulate what will matter until it matters.”

Ask What AI Can’t See

“Do I know something important—context, constraints, or tacit knowledge—that the AI can’t access or infer on its own?”

People Stay Accountable

“Authority and accountability remain firmly with humans.”

Kris Johnson Ferreira and Jordan Tong, “How AI Agents Orchestrate Work Across Silos,” Harvard Business Review, Sept–Oct 2026
Define · Ethan Mollick, One Useful Thing

Define: Agent Swarms

In Mollick’s words

“a group of thousands of agents powered by an advanced model”

People set the goal and a light structure. The agents split up the work, pass ideas back and forth among themselves, and organize the rest on their own. Today’s coding tools already do a small version of this, spinning up teams of agents as a task needs them.

Ethan Mollick, “The Dot and the Swarm: Benefitting from the Bitter Lesson,” One Useful Thing, October 1, 2026
Scope · What This Talk Leaves Out

Not Covered Today

  • Choosing models and AI tools
  • Prompt writing and prompt libraries
  • Coding and building apps with AI
  • Marketing, content and AI search
  • Image, audio and video generation
  • Customer-facing chatbots
  • Fine-tuning, data pipelines and infrastructure
  • Compliance and regulation

My focus: information architecture and process knowledge

How work really flows, and what people know that nobody wrote down.

The Assistant That Went Quiet

It could tell you who owns a process. It could not tell you how the work flows.

  • Built a retrieval assistant for a small international nonprofit.
  • Ingested every formal policy, role document, and job description.
  • Ask it who owns a process: immediate answer.
  • Ask it how the work actually flows: silence.

This Week: Document the Baseline

Ask one person to walk you through their actual Tuesday.

  • Find the one rule nobody wrote down.
  • That is how you document the baseline for Week 1.
  • Then straight into questions.
The Frontier Unknown · Capabilities vs. Reality
“Well, there’s never been a product that’s less understood in terms of what its capabilities are than AI.”
— Bill Gates
The Ezra Klein Show · The New York Times (at ~11:10)
The First Unknown: What the model can do  |  The Second Unknown: What the organization actually does
HVFS 2026

The Routine Moves to the Machine

What is left is judgment, context, knowing why.

  • The promise: Speed, lower costs, throughput.
  • The reality: Removes repetitive tasks, leaving humans with the hardest edge cases.
  • The shift: Knowing why we do things this way raises the bar for everyone.

The Silence Is a Map

What a machine can retrieve is what is written down. Roles get written down. Workflows do not.

Documented Artifacts

  • Job descriptions
  • Org charts
  • Vendor contracts
  • Standard policies

Unwritten Tacit Knowledge

  • Who talks to whom on Tuesdays
  • Hidden approval bottlenecks
  • The "fish ladder" governance rules
  • Informal workarounds

"Walk me through your actual Tuesday."

Deeper listening: Not the process document. The tabs they really have open.

  • One person at a time.
  • Capture the reality, not the theory.
  • Build the picture on the board while they talk.
  • Ask, draw, check.

Thank you

← All Hudson Valley presentations at geo.run