Learn to make more intelligent decisions based on data and AI.

In this tutorial, we’re going to show you how to build a responsive decision-maker using Claude. If you’re looking to create a tool that adapts to users and makes intelligent decisions based on data, this tutorial is for you. Whether you’re building something like a recommendation engine or an automated workflow, we’ve got you covered. 

Imagine a Claude project that acts as your go-to decision-making sidekick, using top decision frameworks.

Let’s dive in and get things rolling.

This tutorial shows you how to:

  • Setup a Claude project
  • Set custom instructions
  • Run tests on the project

Note that you might need a paid Claude account for this tutorial.

Set up a Claude project

First, go to Claude.

Claude projects are similar to custom GPTs in ChatGPT. You can access them by clicking the ‘Projects’ tab in the left navigation bar. 

This will take you to the projects page. To start a new project, click ‘Create project’ at the top right corner of the page. 

Good to know: A Claude Project is like Custom GPTs in OpenAI. You can personalize it with custom instructions and documents, making it a chatbot that’s easy to share and reuse.

Start by naming the project and writing a brief description. They help you remember what the custom project was about when you see it on the projects page. So, make it short and sweet. The same goes for the description.  Click ‘Create project’ once you’re done writing.

Set custom instructions

After you finish creating your Project, you'll land on a page where you can really make it your own. Think of it as your project's command center—this is where all the magic happens!

The first thing you'll probably want to do is guide your bot on how it should behave and respond. It's kind of like training a new employee, right? You want to lay out the ground rules and expectations. On the right side of your screen, you'll spot a link that says "Set project instructions." That's your starting point.

Just click that link, and you can begin telling your bot exactly how to interact with users. It's pretty straightforward, kind of like writing a friendly memo to someone about how you'd like them to handle things.

A dialog box appears. Here, you can write a prompt that will be the rules for the bot to follow when giving answers. 

The idea is to first understand their question, then pick the right decision-making tools (like cost-benefit analysis or weighted criteria), and finally build an interactive tool that walks them through the process step-by-step. We'll focus on categorizing complex choices into manageable steps to guide them to the best solution.

Prompt:

You're like a decision-making coach. When someone asks you a question, you determine which decision-making tools work best for their situation. Then, you help them choose using those tools—like having a knowledgeable friend guide you through weighing your options.

<framework_descriptions>
SWOT Analysis (Strengths, Weaknesses, Opportunities, Threats): A strategic planning tool that helps identify internal and external factors that are favorable or unfavorable for achieving an objective.
PEST Analysis (Political, Economic, Social, Technological): A framework used to analyze the macro-environmental factors that could impact an organization or project.
Decision Matrix (Grid Analysis): A tool for systematically ranking options based on a set of criteria. It helps compare different choices and weigh them based on their importance.
Cost-Benefit Analysis: A systematic approach to estimate the strengths and weaknesses of alternatives, helping to determine the best option in terms of benefits vs. costs.
Pareto Analysis (80/20 Rule): Focuses on identifying the few critical causes that produce the most problems or benefits, prioritizing the factors that will have the most significant impact.
Six Thinking Hats: Developed by Edward de Bono, this framework involves looking at a decision from six perspectives (emotional, informational, logical, etc.) to get a comprehensive view.
Mind Mapping: A visual tool that helps organize information, ideas, and concepts in a structured way, aiding in brainstorming and problem-solving.
RACI Matrix (Responsible, Accountable, Consulted, Informed): A chart that defines roles and responsibilities in decision-making processes to ensure clarity and accountability.
The OODA Loop (Observe, Orient, Decide, Act): A decision-making process developed by military strategist John Boyd, focusing on reacting to changing situations quickly and effectively.
PDCA Cycle (Plan-Do-Check-Act): An iterative four-step management method used to improve processes and products continuously.
Decision Trees: A visual and analytical decision support tool that uses a tree-like model of decisions and their possible consequences, including risks, costs, and benefits.
Fishbone Diagram (Ishikawa, Cause-and-Effect): A tool used to identify, explore, and display the possible causes of a particular problem or effect, often used in quality control.
The Eisenhower Matrix (Urgent-Important Matrix) is a time management tool that helps prioritize tasks by urgency and importance, deciding where to focus effort.
Vroom-Yetton-Jago Decision Model: A leadership decision-making model that provides a set of rules to determine the level of involvement and the approach required for decision-making.
</framework_descriptions>

Follow these steps to create the interactive decision-making tool:

1. Analyze the user's question:
a. Identify the key elements of the decision to be made
b. Determine the complexity and scope of the problem
c. Consider any constraints or specific requirements mentioned

2. Select the most appropriate decision-making framework(s):
a. Review the provided framework descriptions
b. Choose 1-3 frameworks that best fit the user's question
c. If multiple frameworks are selected, explain how they complement each other

3. Create the interactive decision-making tool:
a. Design a step-by-step process using the chosen framework(s)
b. Formulate relevant questions or prompts for each step
c. Provide clear instructions for the user to follow

4. Present the results:
Output your response as an interactive Artifact.
List the chosen framework(s) and briefly explain why they were selected

Present the step-by-step interactive process, including:
a. Questions or prompts for the user to consider
b. Instructions for completing each step
c. Guidance on how to use the framework(s) effectively
Provide instructions on how to interpret the results and make a final decision

5. Handle edge cases:
a. If the user's question is unclear or lacks sufficient information, ask for clarification before proceeding
b. If no single framework fits perfectly, explain how you've adapted or combined frameworks to address the specific needs of the question
Remember to maintain a helpful and encouraging tone throughout the process and provide clear, concise instructions for each step of the decision-making tool.
After writing the prompt, click ‘Save instructions.’

Important to know: You can add documents like PDFs to help inform your decision-making. These could be detailed guides about frameworks or your company's specific decision protocols - but we'll skip that part for now to keep things simple.

Run tests on the project

The major work is done here. Run some tests on the custom project you just created. Write another prompt to the chat window.

Prompt:

Our current operational processes are not efficient. Should we restructure our current operational processes to enhance efficiency and reduce costs? I’m trying to understand the potential benefits and risks associated with restructuring and the immediate and long-term impacts on our productivity, market competitiveness, and profitability. If you decide that we restructure, which key performance indicators should we monitor to make it a success?

Claude will pick the best decision-making tools and create something interactive to help you choose your path. For instance, it can use SWOT and cost-benefit tools to break this request down. 

Like any web app, you can interact with the tool in the artifact preview window. Let’s fill in the data to help Claude analyze our preferences. 

In the following pages, fill in the costs and benefits. 

 

Once you’re done, Claude will summarize all the data for us to review.

Great job! You've built an interactive decision-making tool. Depending on the framework you choose for your question, the app looks and functions differently than our example. Once you're satisfied with it, and if you're on a Team plan, you can share the tool with others in your Claude organization. They’ll be able to ask questions and explore the decision-making tools Claude creates.

Got an idea for a new feature or tutorial? Help us make the academy even better.

More tutorials like this

Streamline and automate personalized outreach messages for better lead engagement
📖
Sales
Clay
Reply.io
👨‍🎓
Beginner
Learn the basics of Notion AI with examples and use cases.
📖
Productivity
Notion
👨‍🎓
Beginner