Your AI Action Plan: 5 Challenges, 20 Minutes
You've got the theory. Now put it to work. Five real business challenges, each with a guided AI exercise. Twenty minutes total.
Knowing about AI is worthless without doing. These five timed challenges take 20 minutes total and produce real outputs: a competitor brief, a branded email, a workflow map, a security quick-start, and a 30-day action plan. Do them now, not "sometime this week."
You've read six posts. You understand how AI works, how to prompt it, how to make it sound like your brand, how to build agents, how to research competitors, and how to handle security. That knowledge is worth exactly nothing if you don't use it.
This post is your action plan: five timed challenges that give you real, usable outputs and get you moving before you close the tab. Each one produces something you can use immediately. Set a timer. Do them in order. When I ran these live in our ZeroShot Studio workshops, the 20-minute constraint was what made them stick. People who "planned to try it later" rarely did. People who did it right then kept using AI daily.
Challenge 1: Can you build a competitor snapshot in 4 minutes?
What: Use the competitor analysis template from Post 5 to analyse one competitor right now.
How:
- Open ChatGPT, Claude, or Gemini (1 minute)
- Paste the competitor analysis template and fill in your business details and one competitor name (2 minutes)
- Scan the output for [UNVERIFIED] flags and bookmark 2-3 claims to check later (1 minute)
You now have: A structured competitor brief you can share with your team today.
Speed matters more than perfection here. A 4-minute draft you actually produce beats a thorough analysis you never start.
Challenge 2: Can you upgrade a bad prompt in 4 minutes?
What: Combine your brand persona (from Post 3) with the RCTFC framework (from Post 2) to write one real email.
How:
- Pick an email you actually need to send today (30 seconds)
- Set your persona in custom instructions, or paste it at the top of a new chat (1 minute)
- Write a RCTFC-structured prompt for that email (2 minutes)
- Review the output. Does it sound like you? If not, give one round of feedback (30 seconds)
You now have: A ready-to-send email that sounds like your brand, produced in 4 minutes instead of 15.
If you haven't done the tone extraction exercise yet, do it now instead. Paste 3-5 samples of your writing and ask AI to analyse your style. That 10-minute investment pays off every time you use AI for writing from now on.
Challenge 3: Can you map your first automation target in 4 minutes?
What: Identify the top 3 workflows in your business that are ripe for AI automation, based on the criteria from Post 4.
How:
- List 5 tasks you or your team do repeatedly every week (1 minute)
- Score each on: repetitiveness (1-5), risk if AI gets it wrong (1-5, lower is better for starting), and whether it's text-based (yes/no) (2 minutes)
- Pick the top scorer. Write one sentence describing what the automated version would look like (1 minute)
You now have: A clear first automation target with a one-sentence brief for building it.
Here's a quick scoring example:
| Task | Repetitive | Low risk | Text-based | Score |
|---|---|---|---|---|
| Reply to common customer questions | 5 | 4 | Yes | Top pick |
| Weekly team status summary | 4 | 5 | Yes | Strong |
| Invoice processing | 3 | 2 | Partially | Hold |
| Social media scheduling | 4 | 4 | Yes | Strong |
| Quarterly financial reporting | 2 | 1 | No | Not yet |
Challenge 4: Can you lock down your AI security basics in 4 minutes?
What: Implement the first 2 of the 5 quick wins from Post 6.
How:
- Check which AI plan your team is on right now. Free or paid? (1 minute)
- If free: start the upgrade process to a team plan for your top 3 users (2 minutes)
- Write your "never paste" list. Five categories of data that should never go into any AI tool. Send it to your team right now via Slack or email (1 minute)
You now have: The beginning of an AI security posture. Not perfect, but dramatically better than nothing.
Security doesn't need to be comprehensive on day one. It needs to exist on day one and improve every quarter.
[IMAGE: The five challenges shown as a progress tracker with time estimates]
- Type: diagram
- Filename: five-challenges-tracker.png
- Alt text: A horizontal progress tracker showing five challenges with 4-minute time estimates each, totalling 20 minutes
- Caption: 20 minutes. Five outputs. No excuses.
Challenge 5: Can you build a 30-day AI plan in 4 minutes?
What: Write a simple 30-day plan for your AI adoption, using AI to help.
How:
- Open a new chat and paste this prompt (1 minute):
"I run [YOUR BUSINESS]. I want to adopt AI tools over the next 30 days. I've identified these priorities: [paste your workflow audit top 3 from Challenge 3]. Create a simple 4-week plan with one specific action per week. Keep each action achievable in under 2 hours. Format as a table with columns: Week, Action, Tool to use, Expected outcome."
- Review the output and adjust any actions that don't feel realistic (2 minutes)
- Put Week 1's action in your calendar right now, with a specific day and time (1 minute)
You now have: A concrete 30-day plan with the first action already scheduled.
The single biggest predictor of whether someone actually adopts AI after a workshop? Whether they scheduled a specific next action before leaving the room. Not "I'll try it sometime." A date, a time, a calendar entry.
What are the eight principles to remember?
These are the core ideas from the entire course. Pin them somewhere visible:
- It predicts; it doesn't think. It's a very good autocomplete. Stop expecting wisdom from it and start treating it like one. (Post 1)
- Being specific matters more than which model you pick. A detailed prompt on a cheap model beats a lazy prompt on the expensive one. Every time. (Post 2)
- Show it your writing, don't try to describe it. Paste three examples of something you've written and let it figure out your style. Describing "warm but professional" gets you nowhere. (Post 3)
- Start with the dull stuff. The boring, repetitive, text-based tasks are the easy wins. Get those running before you get ambitious. (Post 4)
- Let AI handle the structure; you handle the facts. It's brilliant at organising things. It makes up numbers. Keep that straight. (Post 5)
- Free plans are not for business data. If your team handles customer information, pay for the team plan. It's not expensive and it matters. (Post 6)
- Start this week. Seriously. The stuff you learn in week one makes week two easier. Waiting until you "have a proper strategy" just means not starting.
- Nothing works first time. That's fine. Your first prompt, first agent, and first policy will all need tweaking. That's not failure, that's how it works.
Where do you go from here?
You've finished the course. Here's how to keep building:
Free resources we recommend:
- Anthropic Prompt Library -- curated prompts for common business tasks, excellent starting points for your own templates
- OpenAI Cookbook -- practical recipes and examples, useful even if you use Claude or Gemini since the principles transfer
- Building Your First AI Agent -- our technical guide for when you're ready to go beyond no-code
Keep practising:
- Use AI for at least one real task every day for the next 30 days
- Save your best prompts in a shared document
- Review and update your tone profile every quarter
- Run the competitor analysis template whenever a new player appears
Share what you learn:
- Teach one colleague the RCTFC framework this week
- Run a 15-minute AI show-and-tell at your next team meeting
- If you found this course useful, share it with another founder who could benefit
The course is done. The practice starts now. Open an AI tool and do Challenge 1 before you close this tab.
FAQ
Your workflow audit winner from Challenge 3. Build a basic version using Zapier or Make.com. Don't over-engineer it. Get something running, even if it's rough, and improve it over time.
Start with a 15-minute demo. Show them one task done manually, then done with AI. The "before and after" is more convincing than any presentation. Share this course with them and suggest they each pick one challenge to try.
Healthcare, finance, legal, and education all have additional AI compliance requirements beyond general GDPR. The framework in Post 6 gives you the foundation, but consult a specialist in your industry's regulations before deploying customer-facing AI.
The principles stay stable. The specific tools and pricing change every few months. We update the course periodically, but the core framework (predict, prompt, personalise, automate, verify, secure) will serve you well regardless of which models dominate next year.
Start with Anthropic's prompt engineering documentation and OpenAI's prompt engineering guide. Both are written for non-developers and cover advanced techniques. When you're ready for code-level integration, the OpenAI Cookbook has production-ready examples.
That's the course. Seven posts. Seven skills. One action plan. The only thing left is to do the work. Open ChatGPT or Claude right now and knock out Challenge 1. Four minutes. Timer starts now.
This is Post 7 of 7 in the AI for Business free course. Previous: Security & GDPR. Back to: Course Overview.