AI Adoption Advisory

Get your people to use the AI you already bought.

I help leadership teams get employees to use the AI they already paid for, and I tell them whether it is fit to stay. In a short diagnostic we look at what people are avoiding, what the tool is getting wrong, and what has to change before the company trusts it. You leave with a clear call on what to keep, what to drop, and how you will know it is working.

I'm taking on a small number of new projects. Email me what you're working on and I'll reply.

Scott Mastin

The Enterprise AI Adoption Curve

From frontier hype to sovereign small models.

Evaluation is the gate at every step of AI adoption.

AI Eval Corporationaieval.org
Each gate is an AI Eval service
  1. 01

    Manual Work

    People and existing software do the work.

  2. Gate 1Mission metrics before seats

    02

    Frontier Subscriptions

    Enterprise seats or API access, with people told to use AI.

  3. Gate 2Independent quality measurement

    03

    Token Waste

    Frontier-scale models on routine work.

    Spend peaks here
  4. Gate 3Right-sizing log audits

    04

    Workflow Automation

    Pipelines and light routing: smaller models for easy steps.

  5. Gate 4Workflow and agent harnesses

    05

    Right-Sizing

    A traffic audit shows which calls do not need frontier-scale reasoning.

  6. Gate 5DoW Bench + SLM leaderboard

    06

    Small Models

    Targeted sub-tasks move after the same acceptance tests.

  7. Gate 6ATO-aware continuous T&E

    07

    Sovereign In-House AI

    Domain-specific small models on data and hosts the organization controls.

Corrections flow back from every step

Each correction is handled one of three ways.

  • One-offFix it once.
  • Test caseSave it to test and train.
  • Fix the systemChange the process.
  • Held-out test set
  • Fine-tuning data

Feeds the small in-house model

Adoption problems usually sit at stage 2: seats are bought, people are told to use AI, and nobody has measured whether the work got better. The diagnostic is Gate 2, independent quality measurement, applied to your team's real work.

Tell me what you're seeing. I'll reply with the stage I think you're at and the next test.

Email me

Get in touch

How to reach me

What to put in your email

  • What your team was given, and who actually uses it
  • What people are avoiding, and why
  • What the tool gets wrong on your real work
  • What you've already tried
  1. 1. Email me.

    Tell me what you bought, who is using it, and where it falls short.

  2. 2. I reply.

    I'll tell you where I'd look first: the tool, the training or the workflow.

  3. 3. Decide together.

    If it makes sense to keep going, we work out what that looks like. If it doesn't, you still leave with a clearer picture.

You'll work with Scott Mastin, who manages AI test and evaluation programs for the Pentagon.