Week 1
Agree the one-page rules. Everyone takes the diagnostic and completes one lesson.
AI training for a small business works best when it starts with a few real, repeatable tasks, gives each person practice in their own role and keeps the manager close enough to review the work. It does not need a course-design project or an L&D department.
Start with one repeatable task per role that someone already knows how to review. A marketer might revise an email from a fictional brief; an operations colleague might check a handoff summary; a support agent might draft a reply from the returns policy. Each task has a useful output and an obvious boundary, which makes it easy to see whether the AI helped.
For a team of five to twenty, a manager can start with a small group, review the work and decide whether the routine deserves more time. That is a practical rollout suggestion, not evidence that a particular cohort size produces engagement.
Here is a worked example from a fictional eight-person design studio, Fieldnote. The owner wants everyone using AI for meeting follow-ups. The supplied notes from a client call say: the client approved the logo direction, wants two colour options by Friday, and asked whether the website can launch before the end of the month.
The prompt: "Write a short follow-up email from these notes. Include only what was agreed. List any open question separately. Do not promise dates that are not in the notes." A weak draft confirms the website will launch before the end of the month, a promise nobody made. The reviewed email confirms the logo approval and the Friday colour options, and lists the launch date as an open question for the project lead. The task takes ten minutes and teaches the habit that matters most: the assistant drafts, the person decides what is promised.
Use a common language for the checks: context, evidence, information boundaries and human judgement. Let the examples vary by role so a finance colleague does not practise a sales email simply because the team shares one account.
Tosheo.ai is being built for exactly this mix. The manager assigns each person a role from a preset list, such as Marketing, Sales, Ops, Support, HR and Recruiting, Finance, Product or General, and each person takes a five-minute diagnostic before getting an adaptive path of 10 to 15 minute lessons. Tasks are graded by AI against a visible rubric, and a score can be flagged for review. Role pages for marketing, sales, HR and finance show sample tasks.
Write down three things on one page: which AI tools and plans are approved, what information must never go into them, and who reviews AI-assisted work before it reaches a customer. Small teams often skip this because everyone sits together, and then discover that two people use personal accounts and one pastes client contracts. The AI policy guide for small business has a template.
Keep it to one short session a week per person, plus one review conversation. The sequence below is a suggestion, not a measured result.
Agree the one-page rules. Everyone takes the diagnostic and completes one lesson.
Each person practises their role's first task on fictional material.
Apply the same routine to one real, low-risk piece of work and show the manager the draft and the edit.
Review together: which tasks saved effort, which need more checking, and which should stay manual.
The planned Team offer is priced per seat per month with a five-seat minimum, with a cheaper annual option, and includes role paths, practice scenarios, manager visibility and a recognised certificate after a final assessment. A 7-day free trial for up to five seats is planned, with a card required. The pricing page records the figures in one place; they are a starting hypothesis and may change before launch.
If your team is three people, paying for five seats may not make sense. The free guides on this site and a written one-page policy may be enough until you grow.
Look at the quality of an edited output and the learner's explanation. Check whether people have time to return to practice and whether someone is available to review it. Tool access and course completion are inputs, not proof of improved job performance.
The planned manager view shows skill progress and completion, not raw chat logs, and a weekly digest summarises activity. Tosheo.ai is pre-build. Everything described here is planned launch scope, not a record of customer training, assessment results or outcomes. The launch offer is intended to be self-serve; SSO, LMS integrations and enterprise contracting are not part of the initial scope.
It is not right if you are a sole trader learning on your own, because plans are built for teams. It is not right if you need lessons in a language other than English, a mobile app, or training on building automations and AI agents. And because the product is pre-build, it cannot help a team that needs training before launch; start with the 30-day rollout guide instead.
Planned lessons take 10 to 15 minutes, so one or two sessions a week per person fits most small teams. The initial diagnostic takes about five minutes.
No. The planned setup is designed for a manager or founder to complete in about ten minutes: choose roles and tools, optionally upload documents on Growth, and invite the team.
The planned Team plan has a five-seat minimum, so a three-person team would pay for five seats. The free guides on this site may be a better start until the team grows.
No. The planned manager view shows skill progress and completion, not raw chat logs.
The planned recognised certificate follows a final practical assessment and has a verifiable URL and a LinkedIn share. It shows that a person completed the path and assessment; it is not proof of business results.
Explore the planned role-specific lessons, practical tasks and manager view.