Learn AI the practical way, and get ready for the new jobs it is creating.

Learn

Courses
Roadmaps
AI careers
Labs
Big map of AI
Certifications

Practise

Interview questions
Interview prep path
Prompt library
Calculators

Explore

AI tools
Models
Weekly digest
What's new

Company

Pricing
Contact
Verify a certificate
My plan

Privacy Terms Refunds Cookies

– visitors– online now

© 2026 Route to AI. For education, not professional advice. With your permission (or by default where the law allows), we count visits anonymously with a random ID stored in your browser. No name, email or IP address is saved for this. See cookies and storage.

Pilots, Scaling & the Operating Model · AI Transformation Strategist · Route to AI

Courses ✦AI careers My learning Roadmaps Labs Pricing

Sign in Start learning

Courses

Lessons and outlineLesson 6 of 10

Course0 / 10 done

AI Transformation Strategist

Overall completion 0%

Lessons

AI Strategy in Plain Words
Finding & Ranking AI Use Cases
The Business Case: Costs, Benefits & Payback
Risk, Rules & Governance
Change Management & Adoption
Pilots, Scaling & the Operating Model
AI Technology Literacy for Leaders
Data Readiness & Company Knowledge
Measuring Value & Reporting to Leaders
Strategy Interviews & Case Practice

On this lesson

In plain English
Design a pilot that teaches
Decide with evidence
Build, buy or configure
An operating model that lasts
Funding and the roadmap
Report in the leaders' language
Key takeaways
Hands-on exercise
Check yourself

Course0 / 10 done

AI Transformation Strategist

Overall completion 0%

Lessons

AI Strategy in Plain Words
Finding & Ranking AI Use Cases
The Business Case: Costs, Benefits & Payback
Risk, Rules & Governance
Change Management & Adoption
Pilots, Scaling & the Operating Model
AI Technology Literacy for Leaders
Data Readiness & Company Knowledge
Measuring Value & Reporting to Leaders
Strategy Interviews & Case Practice

On this lesson

In plain English
Design a pilot that teaches
Decide with evidence
Build, buy or configure
An operating model that lasts
Funding and the roadmap
Report in the leaders' language
Key takeaways
Hands-on exercise
Check yourself

AI Transformation StrategistLesson 6

18 min

Pilots, Scaling & the Operating Model

Run a pilot that teaches, decide to scale or stop, and set up the roles and funding that last.

In plain English

A pilot is a small, real test with real users, a baseline, a success number and a decision date. At the end, scale, change or stop. Buy for common needs and build for special ones. Set owners, a small central team, staged funding and a roadmap that you review each quarter.

Pilot: A small, real test with an end date and a decision.
Center of excellence: A small central team that sets standards and helps others start.
Roadmap: A time-ordered plan of what you will do next.

Picture it

The decision at the end of a pilot

Scale

Good results, risks understood

↓

0 / 10 finished0% · AI Transformation Strategist

Change

Mixed results, fix and retest

Stop

Poor results, learn and move on

A good strategy ends with a plan that can run. This final lesson shows how to design a pilot. It covers how to decide to scale or stop. It also covers building versus buying, and the roles and funding that keep AI work going.

Design a pilot that teaches

A pilot is a small, real test. It is not a demo. It uses real users, real data and real measures. It has an end date and a decision.

One team, one process, a few weeks to a few months.
A baseline measured before the start.
A success number and a stop rule agreed in advance.
A named owner, a named sponsor and a named reviewer.
A plan for what happens at the end: scale, change or stop.

Decide with evidence

At the end, compare the results to the plan. Look at the success number, the quality, the cost and what users say. Then choose.

Scale: results are good and the risks are understood. Plan the wider rollout.
Change: results are mixed. Fix a known problem and run a shorter second pilot.
Stop: results are poor, or the cost is too high. Write down what you learned and stop cleanly.

Stopping a pilot is not a failure. It saves money that would be spent on a weak idea. Celebrate learning, and share it.

Build, buy or configure

build-or-buy.md

BUY OR CONFIGURE an existing product when:
- the need is common (meeting notes, support replies, document search)
- speed matters more than a perfect fit
- the vendor passes your security, privacy and data checks

BUILD (or have built) when:
- the process is special to you and gives you an edge
- you need control over data, cost or behavior
- you can staff and maintain it for years

ALWAYS ASK
- What does it cost per year, at our real usage?
- How do we leave? Can we export our data and prompts?
- Who fixes it at 3 a.m. when it breaks?
- How will we test it when the vendor changes the model?

An operating model that lasts

AI work needs ongoing owners, not only projects. Many organizations set up a small central team, sometimes called a center of excellence. It does not do all the work. It sets standards, shares tools, reviews risk, trains people and helps teams start.

Central team: policy, shared platform, review, training.
Business owners: each use case has a person who owns the result.
Builders: in teams or shared, they build and run the solutions.
Reviewers: people who check output and report problems.

Funding and the roadmap

Fund in stages. Release a small amount for the pilot. Release more when results meet the plan. Keep a running budget for tools, models and support, because AI has ongoing costs, not only a one-time build.

Turn the plan into a roadmap for the next twelve months. Show quick wins first, then core bets. Review it every quarter. Models, prices and rules change, and your plan should change with them.

Report in the leaders' language

Give leaders a short update on a regular day. Show the success numbers against the plan. Show cost against budget. Show risks and incidents. Show what you will decide next. Use charts only when they help.

→Keep a public list of live AI projects, with owner, purpose, tier and status. It prevents hidden projects and makes audits easy.

Key takeaways

✓A pilot uses real users and data, a baseline, a success number, a stop rule and a decision date.
✓At the end choose scale, change or stop. Stopping with clear learning is a good result.
✓Buy for common needs and build for special ones. Always ask about exit, cost and who fixes problems.
✓Set up owners, a small central team, staged funding and a quarterly roadmap. Report in leaders' own terms.

Hands-on exercise

Write a pilot plan and a twelve-month roadmap

Choose one use case. Write the pilot plan: team, length, baseline, success number, stop rule and owners.
Write the decision rules for scale, change and stop.
Decide whether to build or buy, using the questions in the checklist.
Sketch a four-quarter roadmap with one quick win, one core bet and the review dates.

Check yourself

Q1What makes a pilot different from a demo?show answer

A pilot uses real users, real data and real measures, with a baseline, a success number, a stop rule and a decision date.

Q2What are the three decisions at the end of a pilot?show answer

Scale, change or stop. Each is based on the evidence against the plan.

Q3What does a central AI team do?show answer

It sets standards and policy, shares tools, reviews risk, trains people and helps teams start. It does not do all the building.

Up next

Lesson 7: AI Technology Literacy for Leaders

← Previous Course overview