Deployed Intelligence
A practical way to design the partnership between people and AI—one task step at a time.
Note: Our methodology evolves as our research progresses. This page previews the core concepts.
What is Deployed Intelligence?
Deployed Intelligence treats human and AI capabilities as a unified pool of cognitive resources. Instead of asking “Can AI do this job?”, we ask “Where should intelligence—human or artificial—be deployed across the steps of work?”
From Jobs to Micro‑Delegation
Most work isn’t all‑or‑nothing. By decomposing tasks into a simple six‑step cycle, we can delegate specific moments to AI—drafting, parsing, checking—while keeping human judgment where it matters. This is how real productivity gains happen without losing control.
Three Modes of Work
A simple lens for understanding where AI fits—and where humans lead.
Operational (ODI)
The realm of process
Predictable, rule‑based, and repeatable. High leverage for automation and speed.
Strategic (SDI)
The realm of optimization
Managing trade‑offs and adapting patterns to goals. Best as human‑AI collaboration.
Evolutionary (EDI)
The realm of innovation
Ambiguous and novel. Human‑led with AI as a catalyst for ideas and exploration.
These modes describe the nature of work—not job titles. Most roles blend all three.
The Universal Task Cycle
A simple loop to describe how thinking work gets done. We analyze each step to decide how to partner with AI.
1. Perceive
Gather signals and inputs (e.g., an email, a dataset, a sensor reading).
2. Interpret
Make sense of inputs—what do they mean in context?
3. Determine
Generate possible options or next moves.
4. Select
Choose the best option using priorities, values, and constraints.
5. Act
Execute—draft, code, calculate, update, or perform a procedure.
6. Observe
Check outcomes and feed learning back into the next cycle.
Why this matters
Micro‑delegation lets you automate the routine while preserving human expertise where it’s valuable. It’s not about replacing jobs—it’s about redesigning workflows.
Four Lenses We Use
To decide who should do what, we look at the thinking involved, the human sensitivities, the context/data needs, and how rigid or variable the process is.
Cognitive Demands
How complex is the reasoning? Does it require creativity or deep expertise?
Judgment & Sensitivity
Where ethics, empathy, and values matter, humans lead—with AI assisting.
Context & Data
What information is required? Is historical case memory essential?
Operational Profile
How standardized is the process and what are the consequences of errors?
Built on Proven Ideas
The Deployed Intelligence framework stands on the shoulders of giants. It synthesizes and extends decades of world-class research to offer a sophisticated model for the modern workplace.
Hierarchical Task Analysis
Cognitive Ergonomics
The long-established practice of breaking a complex goal (e.g., "cook a meal") into a hierarchy of sub-tasks and actions.This informs our core principle of moving beyond job titles to analyze the specific tasks that constitute real work.
The OODA Loop
Military Strategy & Cognitive Science
John Boyd's Observe-Orient-Decide-Act loop is a classic model for making decisions under pressure.This classic cycle inspires our own DI-Cycle, viewing every task not as a single event, but as an iterative process of perception, interpretation, and action.
Dual-Process Theory (System 1/2)
Cognitive Psychology
Kahneman's famous model of fast, intuitive (System 1) and slow, analytical (System 2) thinking.This provides a powerful analogy for our ODI/SDI/EDI modes, helping us map the type of "thinking" a task requires.
Decision Intelligence
Applied Data & Management Science
A modern, multi-disciplinary approach focused on improving organizational decision-making by integrating data science and behavioral science.We see DI as a practical application of these principles, providing a structured way to enhance decisions at every level.
Why Architecture Matters
Humans are stateful: we remember histories, relationships, and responsibilities. AI systems are often stateless. DI recognizes this gap and designs around it.
Human Statefulness
- Accumulated Experience: Tacit, contextual knowledge.
- Persistent Relationships: Trust and nuance over time.
- Accountability: A stable identity that owns outcomes.
AI (Often) Stateless
- Short Context: Limited memory per interaction.
- Re‑establish Context: Needs state systems to remember.
- Identity by Design: Responsibility via workflow, not persona.
Where Things Break
Plugging a stateless model into a stateful role causes:
- Context Fragmentation: Decisions without history.
- Repetition Fatigue: Users re‑explaining constantly.
- Accountability Gaps: No clear owner for outcomes.
DI Designs Around Reality
We assess each step’s inherent statefulness and recommend the right blend of human oversight, memory, and guardrails—so AI augments responsibly.
- Automate low‑state steps with confidence.
- Keep humans in the loop for high‑stakes judgment.
- Use memory and audit where continuity matters.
What You’ll Get
As our research and product evolve, expect practical outputs—not just a score.
AI Impact Snapshot
A fast read on where AI can help—without overstating what should be automated.
Micro‑Delegation Ideas
Tangible suggestions for steps to automate or co‑pilot—grounded in your actual work.
Guardrails & Memory
Where to keep humans in the loop and what state or audit you’ll need to deploy safely.
We’re actively calibrating our methodology to balance practicality, safety, and transparency.
See Deployed Intelligence in Action
Explore how micro‑delegation can reshape your daily work. Start small, move fast, and keep control.