The Monday Test: how to choose an AI workshop for contractors, and how to judge it after

This is the checklist from the Cortex video on the Monday Test. It follows the video's three points, in the same order:

1. Your software. The session works inside the tools your group already runs, such as your field software and phones.
2. A short list. It ends with a few jobs AI should take on first, and what stays with a person.
3. A name on each item. Every item on the list has an owner who starts on it.

If a workshop delivers those three, the morning was worth it. If it delivers a pile of prompt tips and a recap deck, it was entertainment.

Why tools are the wrong yardstick

It is easy to judge a workshop by how many tools it shows. The numbers say that is not where owners get stuck. In the Federal Reserve Banks' 2026 report on employer firms, 54% of firms planning to use AI named finding tools that meet their business needs as a challenge, and 37% named the time to implement and train employees. The same report found 46% of small employer firms use AI, but only 7% have fully integrated it.

Adoption is still early across the board. The Census Bureau reported 19.8% of businesses used AI as of May 2026. A workshop that adds ten more tools to the pile does not help. One that narrows the pile to the right two or three does.

Before you book

Ask the presenter these questions, and note the answers:

• What do you need from us beforehand? A good answer asks which software your members run and which office jobs keep slipping. No questions back is a warning sign.
• Will we look at our own tools? You want examples in your field software, CRM and phones, not a generic chatbot demo.
• Who is it for? Owners, general managers and office leads get the most out of it. A room of people who do not run the work will not produce a usable list.
• What do we leave with? The answer you want is a short written list per company or per role, not slides.
• What will you not cover? Someone who can say "we won't promise savings" is more trustworthy than someone who can.

During the session

Use this as a scorecard while you sit in the room.

Check	What good looks like
Your software on screen	Examples in the field software and phone systems people in the room actually use
Plain jobs, not "AI"	Specific tasks: missed-call text back, quote follow-up, unbilled jobs, morning summaries
Workflows versus agents	Known steps get a fixed workflow; only open-ended work gets an agent
What stays human	Pricing, refunds, promises to customers and anything safety related named out loud
Approval	Every example shows where a person approves before a customer sees anything
The list	Time set aside at the end to write each company's short list

The workflow point is not ours alone. Anthropic's engineering team writes in Building effective agents that workflows suit well-defined tasks and agents suit open-ended problems. Most office work in a trades business is the first kind.

The approval point has a security reason too. The OWASP GenAI Security Project lists excessive agency as a named risk and recommends human approval for high-impact actions. A workshop that shows agents acting alone near your customers is teaching the risk, not the fix.

The short list, with names

By the end, each company should have a list like this one. The company and names are examples.

#	Job	AI's part	Stays with a person	Owner
1	Missed calls during jobs	Drafts a text back within minutes	Price and arrival time	Ana, office manager
2	Quotes with no reply	Drafts a follow-up that mentions the job	Any discount	Ben, sales
3	Pricing changes	Nothing yet	All of it	Cal, owner

Three rules for the list:

• No more than three items. More than that and nothing starts.
• At least one line in the "stays with a person" column says "all of it". That shows the room drew a line.
• Every row has a name, and the first row has a start date.

The Monday after

This is the test itself. On the Monday after the workshop, ask each company in the group three questions:

1. Did item 1 start?
2. Does the person named on it know what "done" looks like?
3. What is the first thing they will check with the software they already have before buying anything new?

If most companies can answer all three, the workshop passed. If nobody can, the session taught ideas but did not change the week, and that is the feedback to give the presenter.

When a workshop is the wrong format

A workshop works for a group: a trade association, a franchise group, a peer group or a software user group. A single company that wants its own workflows reviewed in depth usually gets more from a one-on-one review than from a room of peers. For that, your free AI Operations Audit looks at one company's software, phones and books.

If you run a group and want to see how Cortex runs these sessions, read about AI workshops. For what comes after the list, see AI implementation.