We’ve been peeling back the layers of AI one piece at a time.
I shared a simple four-step path to understand AI, and we dug into the first step. We looked at how you can use generative AI to have a dialog, conduct research, analyze information, and create content assets.
Now, let’s get into the second step of the progression: AI configurations.
What Are AI Configurations?
AI configurations are custom, reusable instructions that train AI and teach it how to create more specific output.
When you use out-of-the-box AI tools, you are working with a generic AI model. It’s the same AI model everyone else has. This is why your results feel plain, basic, and watered-down.
AI configurations are how you elevate your output to make it more unique, accurate, and high-quality.
How Can You Use AI Configurations?
AI configurations allow you to get higher-quality output without needing to provide lengthy instructions for every request. You upload your configuration once and can continue to use it.
Here are some ways you can use AI configurations to create better first drafts.
Brand Voice Writer: Give AI your brand guide, past blog posts, and a few of your best-performing emails so it can draft content using your brand voice.
Client Communications Assistant: Set up a configuration that knows your tone for client emails so it can draft replies, follow-ups, and check-ins that sound like you wrote them yourself.
Proposal & Pitch Builder: Give AI your service packages, pricing structure, and a few past proposals so it can create a proposal built on your actual offers instead of generic templates.
Social Media Repurposer: Train AI on social media platform best practices so it can turn one blog post or newsletter into a week’s worth of on-brand social content.
Research & Strategy Partner: Load in your target audience details, competitors, and business goals so AI can help you think through campaign ideas, content angles, or offer positioning based on your real, unique business details.
How Do You Set Up an AI Configuration?
AI configurations are available in each of the most popular generative AI tools. Each tool has its own directions for setting up configurations.
- Custom GPT is an AI configuration in ChatGPT. Directions to set up.
- Claude Skill is an AI configuration in Claude. Directions to set up.
- Gem is an AI configuration in Google’s Gemini. Directions to set up.
What Needs to Be In an AI Configuration?
A configuration only works as well as what you put into it. For it to produce consistent, high-quality results, include:
- Clear Role: Tell the AI who it’s acting as (a marketing expert, a customer service rep, an editor, etc.) and who it’s serving.
- Specific Instructions: Spell out exactly what the task involves and how you want it handled.
- Examples & References: Train AI on your style guides, past work, samples, and product details.
- Constraints and Guardrails: You have to tell the AI what you like and don’t like. Provide directions on what AI should avoid.
- Goals: Define outcomes so the AI knows what “done well” actually means.
Where to Get Started
You have endless options when it comes to training AI. To avoid overwhelm, I suggest starting with the basics.
Start with building a company knowledge base that includes information about your business, customers, offerings, unique sales propositions, and brand guidelines.
Once you teach AI about your brand, it will start producing results that reflect your unique voice and positioning. You can use that as a baseline for creating content, analyzing your industry, and generating valuable insights.
Let me know how it goes!
If you need help building your company’s knowledge base to train AI, let’s work together to build a custom AI assistant for your business.
