Project Inquiry

I work with a small number of companies on AI strategy, workflow redesign, and the systems required to put those ideas into production.

The starting point is usually not a predefined tool. I first want to understand how your team works, where knowledge-intensive processes consume disproportionate time, and where AI can create meaningful leverage.

If there is a strong use case, I can support both the strategic design and the technical implementation.

If you think there may be a fit, tell me a little about your company and what you are currently looking to improve.

Timing: I keep my workload intentionally focused and only take on a limited number of engagements at the same time. Timing depends on scope and current availability.

Scope: I focus on AI-native operating models, agentic workflows, knowledge systems, and custom AI software. I am most useful where there is a meaningful business problem to solve rather than a predefined technical task.

FAQ

AI Strategy & Engineering

AI is useful when it changes the workflow, not when it adds another tab.

I help teams rethink how research, analysis, and decisions move through the business — then build the systems that make the change real.

An isolated tool leaves the work around it unchanged.
  1. 01Research
  2. 02Analysis
  3. 03Draft
  4. 04Review
  5. 05Decision
Isolated AI tool

01

Assess

Start with how the work happens today. Where does time go? Where does information move? Where do people stop to make a judgment?

A useful map follows the work, not the org chart.

02

Diagnose

Find the moments where AI can create real leverage. Separate repetitive synthesis from work that needs context, accountability, or expert judgment.

Not every task should be automated.

03

Prescribe

Define the target workflow before choosing tools. Set human and system responsibilities, the data required, the interfaces, the boundaries, and the order of work.

A target workflow is more useful than a feature list.

04

Implement

Build what the workflow needs: retrieval, structured interfaces, integrations, review states, and software that can hold up in day-to-day use.

Strategy only matters when it becomes a working system.

See Markets AI

Human judgment stays where it matters.

In high-stakes workflows, the goal is not maximum autonomy. The system should make its output easy to inspect, correct, and verify.

Models propose. Systems constrain. People verify.

Probabilistic generation

Retrieval and model output

Structured state

Schema, validation, evidence

Human review

Inspect, edit, verify

Missing evidence should be visible. Outputs should stay editable.

Independent research

Collaborative AI for Private Markets: Schema-Driven Document Generation

Independent research into a private-markets editor where model output is schema-bound, validated, and reviewed by people before it becomes part of a high-stakes document.

Read the paper (PDF)

Build the system only after the problem is understood.

Test the method on one workflow