I’ve been having lots of calls with my students recently, helping them chart a course and giving me more insight into how I can better help them.

A lot of my students are very impressive. Intimidatingly so! We’re talking founders, owners, C-suite and VP level, PhD. and experts in their fields. Amazingly high level people.

They have all identified high leverage uses of AI in their various fields. These are experts who have deep domain knowledge and know the sort of problems that AI could be applied to.

When we talk about what they need to push forward the answers fall into two categories:

1. Building. How to prepare a solution for their industry.

2. Marketing. How to get the industry on board and purchasing.

These are the two pillars of AI entrepreneurship we’ve been discussing throughout this Playbook. It was discussions with students that helped me to come up with this dual pillar focus so that makes sense!

In this Part we shift gears again to a new stage.

Let's get started:

Why Custom AI Apps?

Previously we talked about AI automation. Sometimes this is insufficient and we need additional sophistication.

There are several reasons why you might need to build a custom AI app:

1. Integration Limitations: While workflows can connect existing tools, they might not offer the deep integration needed for complex processes. A custom app can provide seamless data flow and user experience.

2. Performance Requirements: For applications requiring real-time processing or handling massive datasets, a custom solution might be necessary to achieve the required speed and efficiency.

3. Intellectual Property: Building a custom AI app allows you to create proprietary technology, potentially giving your business a significant competitive advantage. It’s a protectable IP asset.

4. Security: Some clients just will not want their data being sent through automation flows and want it instead centralised and protected.

5. Funding: More of an edge case but some countries provide grants/loans for new AI IP being created. A custom deployment is much more likely to be approved for funding than a set of AI automation connected in Zapier!

Building custom AI solutions is the next step if simpler automations are not viable. They are also much more expensive when you charge a client - if automations were in the tens of thousands we are now moving into the hundreds of thousand range.

Now, let's dive into the two pillars when building custom AI apps: the Business Pillar and the AI Pillar.

Business Pillar: Strategising Your Market Approach

Before we start building, we need to ensure there's a market for our custom AI app. Again we need to avoid this situation:

Notice his problem here? “Built the product, talked to customers”.

We have to talk to customers first! Then build what they are telling us they want to buy. I sound like a scratched record on this point - and I won’t apologise!

Ideally you already have a customer lined up. Someone who has (after your workshops and automation work) asked you to help them with something bigger. That’s the ideal.

Still need some ideas? Let’s run this prompt to kickstart:

You are an AI innovation consultant specialising in custom AI applications. Based on the following industry and customer problems, generate ideas for custom AI apps that address these issues.

Industry: [Insert industry here]
Customer Problems: [List 2-3 major problems customers in this industry face]

For each problem, suggest a custom AI app idea. Provide 5 ideas in total. For each idea, include:

1. App Name: A catchy, descriptive name for the AI app.

2. Brief Description: Write a tagline and basic marketing materials I could use to pitch this to people in the industry. A 2-3 sentence explanation of what the app does and how it solves the problem.

3. Key Features: List 3-5 main features of the app.

4. Custom AI Justification: Explain why this needs to be a custom AI solution rather than using existing tools. Choose from and elaborate on 1-2 of the following reasons:
   - Unique problem set
   - Integration limitations
   - Performance requirements
   - Intellectual property creation
   - Scalability needs
   - Regulatory compliance
   - User experience control
   - Data security and privacy
   - Competitive differentiation
   - Complex decision making

5. Potential Impact: Describe the potential impact this app could have on the industry or customer operations.

6. Data Requirements: Briefly outline the type of data this AI would need to function effectively.

7. Possible Challenges: Identify 1-2 potential challenges in developing or implementing this solution.

Present your ideas in a clear, structured format. For each idea, provide a compelling case for why it would be valuable as a custom AI solution in the specified industry.

Use this prompt to get a starting point for your marketing and sales approach.

You need to validate with actual people. Actual customers. AI can’t help you here. Go and talk to people!

AI Pillar: Building Your Custom AI App

Now that we've identified a promising custom AI app idea, it's time to figure out how to build it. The approach you take will depend on the complexity of your app, your current technical skills, and the specific requirements of your solution.

There are several levels of AI app development, each requiring different skills and tools:

1. No-Code Platforms: Tools like Bubble or LaunchLemonade allow you to build simple AI apps without coding.

2. Low-Code Platforms: Platforms such as Mendix offer more flexibility with minimal coding required.

3. AI Development Frameworks: For more complex apps, frameworks like TensorFlow or PyTorch provide powerful tools for building sophisticated AI models.

4. Custom Development: For highly specialised needs, you might need to build from scratch using programming languages like Python or Java. You can use tools like Cursor to aid you here but will still need to know coding.

To help you determine the most appropriate approach for your app and understand the potential for future development, use this prompt:

You are an AI development strategist specialising in custom AI applications. Based on the following AI app idea, recommend the most suitable development approach and provide a roadmap for potential future enhancements.

AI App Idea: [Insert your app idea here]

Please provide:

1. Recommended Initial Development Approach:
   - Suggest the MOST appropriate method (No-Code, Low-Code, AI Framework, or Custom Development)
   - Explain why this approach is suitable for the current idea
   - List 2-3 specific tools or platforms to consider
   - Outline the basic steps to start development using this approach

2. Skill Development Roadmap:
   - Identify the next level of development sophistication
   - Explain the benefits of this higher level for the app idea IF it's required.
   - List the key skills needed to reach this level
   - Suggest 2-3 resources (courses, books, etc.) for acquiring these skills

Present your response in a clear, structured format that guides the entrepreneur through the current development needs and future possibilities.

This prompt has been designed to identify the level of skill required to tackle your project idea and then also push your development by suggesting the next level.

Remember, the journey from concept to custom AI app is not just about coding or using the right tools. It's about understanding the unique problems in your industry, identifying opportunities where custom AI can make a real difference, and then strategically approaching both the market and the development process.

Got to do both! Do not just build.