Bringing AI to your team

Up to now the course has been talking to you. This module is about everyone else: the five, fifteen or forty people who work with you and who already use AI assistants. What you learned in nine modules (context, instructions, the data traffic light, projects, automations, verification) is not worth much if it lives only in your head. Here we turn it into a team method.

Five questions: the starting point, the 30, 60 and 90-day plan, training, what to measure, and how to talk about the changes. It is the last module of Part 2; at the end we tell you what comes next.

What does AI use look like in a company that "already uses AI"?

Almost always the same. The manager uses ChatGPT for emails. Someone in sales builds quotes with Gemini on a personal account. The bookkeeper pastes bank statements into a free chat because "it’s faster". Nobody knows what customer data has left the company, and when something goes wrong nobody knows where it came from. We describe it in AI for small businesses in Colombia: where to start: there is usage, but no method.

The symptoms you will find if you ask:

• Personal accounts mixed with work. Company information sits in accounts the company does not manage, under consumer terms where there may be training on your data.
• Everyone with their own tool and style. The same task (answering a complaint, summarizing a meeting) comes out differently depending on who does it, because nobody shares prompts or context, and the person who found an excellent way to draft proposals has nowhere to leave it for the others.
• Nobody knows what data went out. Customers’ or employees’ personal data has passed through chats without anyone deciding it should. Not out of malice: because nobody said no.
• Quiet fear. Some people use AI in secret because they do not know whether it is frowned upon; others will not touch it because they think it is cheating, or that it will replace them.

Before planning, measure that starting point with an anonymous eight-question survey or a 20-minute conversation with the whole team. The first prompt in this module builds it for you.

I need a short, anonymous survey to find out how my team uses artificial intelligence today. We are [NUMBER] people at a [WHAT IT DOES] business in [CITY]. Areas: [SALES, ADMIN, OPERATIONS...].

Design a survey of at most 8 questions that takes 5 minutes to answer, in everyday language and without an audit tone. It must cover: which AI tools each person uses (including the option "none"), on what kind of account (free personal, paid personal, company, not sure), for which tasks, how often, what kind of information they have pasted in (options from least to most sensitive), what has worked well for them, what worries them, and what they would like to learn.

Give me: (1) the questions ready to drop into a form, with their answer options; (2) a three-line message to send the survey to the team, making clear it is anonymous and there are no "wrong" answers; (3) a reading guide: which patterns to look for in the answers, which ones are data-risk signals, and how to spot the two people who could be the team’s AI champions.

When I paste the answers, help me summarize them in one page with three findings and one recommendation on where to start.

How do I build a 30, 60 and 90-day adoption plan?

The most common mistake is starting with the flashiest tool or the most ambitious process. The plan that works in a small company has few pieces, in this order: one main tool on a managed account, a one-page policy, one or two people who push, two or three concrete workflows, and the shared documents that feed them. Ninety days is enough to have it running without paralyzing the operation.

Days 1 to 30: foundations

• Pick one main tool and buy it on a team or business plan. For the admin controls and data terms from module 3: accounts the company creates and removes, connectors an admin approves, and terms that by default do not use your conversations for training. If there are three tools today, do not ban them on day one: name one as the main tool and move the work that involves internal information there.
• Write the AI use policy. The one-page version from module 5: approved tools, data traffic light, anonymize first, verify before sending, a person signs. You draft it with that module’s prompt, discuss it with the team (not impose it by email), and sign it yourself first.
• Name one or two champions. Not the most technical person: the most curious and respected one, who already uses AI well and whom the others ask. They answer questions, collect what works and try every new workflow first. Give them real time, even if it is two hours a week.
• Run the diagnosis. The survey above, before the new tool changes the answers.

Days 31 to 60: the first workflows

• Pick two or three workflows, not ten. The candidates are the ones from module 7 and module 8: tasks that repeat every week, follow rules that fit on one page and can be measured.
• Build the company context document and one per workflow. The one from module 2, but now shared: who we are, what we sell, to whom, in what tone, what we never promise. It lives in a project or a shared folder, and it has an owner who maintains it.
• Open the team prompt library. The one from module 4, in a shared document: name, what it is for, the prompt with its blanks, and who tested it. Start with the prompts for the chosen workflows.
• Measure the baseline of each workflow before touching it: how long it takes today, how many errors or redos it causes. Without that number, on day 90 you will not be able to say anything serious.

Days 61 to 90: extend and decide

• The champions bring each workflow to the rest of the team with one short session per workflow (next section).
• Revise the policy with what you learned. Which rule nobody understood, which case was not covered, what got pasted that should not have been.
• Compare against the baseline and decide, workflow by workflow: keep, adjust or drop. Dropping a workflow that did not work is a result, not a failure.
• Pick the next two or three workflows and repeat. The plan does not end with a finished system; it ends with a team that knows how to bring in the next one.

Help me build a 90-day artificial intelligence adoption plan for my company. I want a realistic plan for a small team, not a corporate slide deck.

My company: [WHAT IT DOES, HOW MANY PEOPLE, CITY]
Diagnosis results (summary): [WHICH TOOLS PEOPLE USE TODAY, ON WHICH ACCOUNTS, WHAT WORRIES THEM]
Main tool we chose and plan type: [TOOL, TEAM/BUSINESS PLAN]
Champion candidates: [NAMES OR ROLES]
Repetitive tasks that could be the first workflow: [LIST OF 4 TO 8, WITH HOURS PER WEEK EACH TAKES]
People’s data we handle: [CUSTOMERS, EMPLOYEES, GUESTS...]
Hours per week I can dedicate, and the champions can dedicate: [HOURS]

Structure the plan in three blocks (days 1 to 30, 31 to 60, 61 to 90). In each block: objectives, concrete actions with an owner, which document gets written (policy, context document, prompt library, measurement log) and what signal tells me the block ended well.

For the workflows: pick the best two or three from my list based on how often they repeat, whether they have clear rules and whether they can be measured, and explain why you discarded the others. For each chosen workflow, say which baseline to measure before starting and who reviews the result before it reaches a customer.

Finish with the three most likely reasons this plan would not get done in my company, and how to prevent each.

How do I train my team without losing a week?

Not with a two-hour talk on "what AI is". What works is short sessions, hands on the tool, on real tasks belonging to the people in the room:

• 45 minutes, at most eight people, each with their own laptop and their company account already created.
• One real task per session, from one of the chosen workflows. For example, answering three customer complaints from last week, with names anonymized.
• The champion does one live, with the context document and a prompt from the library, making mistakes and correcting them in front of everyone. That teaches more than any slide.
• Each person does their own with a task of theirs and compares the result with the person next to them. The differences open the conversation about context and verification.
• A five-minute close: what goes into the prompt library, which question stayed open, and which rule of the policy came up today.

For the conceptual foundation you do not have to invent anything. Anthropic Academy (academy.claude.com) offers free courses: AI Fluency for Small Businesses (about three and a half hours) applies the 4D framework to the tasks of a small company and ends with a repeatable workflow and a use policy; Building Effective Human-Agent Teams (about 45 minutes) is for people leading a team that already uses AI individually and wants to move to shared work. And this course, free and bilingual: Parts 1 and 2 are onboarding material for anyone new to the team.

Prepare the script for a 45-minute, hands-on training session on our AI tool for my team.

Tool we use: [TOOL, TEAM PLAN]
Workflow we are teaching: [FOR EXAMPLE: answering customer complaints by email]
Participants: [NUMBER AND ROLES]; current level: [NEVER USED AI / USE IT ON THEIR OWN / MIXED]
Material that already exists: [CONTEXT DOCUMENT: YES/NO; PROMPT IN THE LIBRARY: YES/NO]
Rules from our policy that apply to this workflow: [FOR EXAMPLE: customer data anonymized, a person reviews before sending]

I want a minute-by-minute script with: a 5-minute opening that explains why this workflow and what changes for them; a 10-minute demonstration I do live, with a realistic example but no real personal data, including a deliberate mistake and how to fix it with feedback; 20 minutes of individual practice with exact instructions for each person; 5 minutes of comparing in pairs; and a 5-minute close.

Include three questions they will probably ask me, with a short answer for each, and a list of what I need to have ready before walking into the room.

What should I measure to know whether AI is working?

Few things, concrete, per workflow. Three questions are enough:

• Time. How long the task took before (the baseline) and how long now, including review. If moving an invoice into the spreadsheet took twelve minutes and now takes four with review, that is a data point. If "it now takes two" but the review got skipped, that is not a data point: it is a risk.
• Quality. How many errors or redos came out before and how many now: corrected quotes, replies the customer did not understand, figures fixed afterwards. It is the measure that protects you from "faster but worse".
• Adoption. How many people use the workflow as designed (company account, context document, shared prompt) rather than a version of their own. A workflow only the champion uses is not adopted.

What is not worth measuring, even if it looks good on a slide: messages sent to the AI, people who "have an account", prompts in the library. These are vanity metrics: they go up on their own and say nothing about whether the business changed. And if the hours saved come from "we think it saves about half", they are not figures; ask for the log.

The log can be a spreadsheet with five columns: workflow, date, who, time before and after, errors caught in review. It takes a minute to fill in, and after 90 days it gives you a story with your own numbers: the only source you should take any figure about AI in your company from.

How do I talk to my team about how roles change?

Truthfully and without adjectives. Some tasks change shape: the person who wrote every reply from scratch now edits drafts and decides the hard cases. That needs to be said out loud, before people work it out on their own. Two things kill adoption: the speech about how "this changes everything" (nobody believes it, and it creates anxiety) and silence (everyone fills the gap with the worst case). What works is describing, workflow by workflow, what the AI does, what the person keeps doing, and why that part matters more than before.

The second thing is making clear who answers for the output. That is Diligence, the fourth skill of the AI Fluency framework, applied to the team: every workflow has a named person who reviews and signs before anything reaches a customer, a supplier or a public body. "The AI did it" is not a valid explanation, just as "the intern did it" never was. That person is the human in the loop from module 9, and naming them is part of designing the workflow. If you have a development team, the same principle applies to code: How to adopt Claude Code or Cursor in a development team is the version for programmers.

The third thing is knowing when to ask for help. This plan is designed so you can run it without anyone from outside, and most small companies can. It is worth bringing someone in when a workflow touches systems your team does not control (billing, a CRM, high-volume WhatsApp), when the data risk is high (health, minors, third parties’ finances), or when you have tried twice and the team is still in the same place. If you get there, 77Rentals offers AI consulting for Colombian companies at 77rentals.com/ai; but the diagnosis, the policy and the first workflow you can do this week.

What comes after this module?

This is the end of Part 2 and, with it, of everything that requires no code. If you made it here, you have the complete toolkit: understanding the model, giving it context and instructions, protecting data, using projects, connectors and automations, verifying what it produces and bringing it to your team with a method. You can stop here. That is a valid decision and, for many companies, the right one: the value is in applying well what you already know, not in learning the next tool.

If you want to build your own tools (a workflow the apps do not offer, an integration with your systems, an agent that works while you sleep), continue to Part 3. Module 11, "Claude Code: first steps without being a developer", starts from zero with the terminal, so a business person can build something real or at least understand what they are asking of whoever builds it.