Aristotle - AI Decision Coach

Stop guessing. See what your big decision really comes down to.

Comment

Inspiration

Big decisions are rarely hard because we lack information. They are hard because we do not know what actually matters. We watched friends and family struggle with choices like taking a job versus pursuing a degree, moving across the country, or choosing between two great opportunities. Most tools either oversimplify the decision or overwhelm users with information. We built Aristotle to help people think clearly when the stakes are high. Our belief is simple: AI should not make decisions for people. It should help people understand their decisions better.

What it does

Aristotle is an AI Decision Coach for important, hard-to-reverse decisions. Users describe their situation in plain language, and Aristotle helps them structure the decision, evaluate options, test assumptions, and understand tradeoffs. Its signature innovation is The Hinge: the single factor a decision turns on. Instead of simply recommending an option, Aristotle identifies the one thing a user should verify before committing. Aristotle also provides a Fragility Score to show how stable a recommendation really is, a Future-Self Lens that compares decisions through both 12-month and 10-year perspectives, stress tests and scenario checks to pressure-test assumptions, and AI second opinions that surface blind spots, explain tradeoffs, and explore failure risks. Most importantly, Aristotle never makes the decision. The final call always remains with the user.

How we built it

Aristotle combines a transparent decision engine with agentic AI. The decision engine runs on the user's device and powers scoring, sensitivity analysis, uncertainty ranges, stress tests, and The Hinge. On top of this foundation sits a CrewAI-powered reasoning layer consisting of a Blind-Spot Finder, Tradeoff Explainer, and Premortem Analyst. These agents challenge assumptions and provide perspective, while the deterministic engine owns the recommendation. The system is deployed through AWS Lambda with secure cloud execution and includes an on-device fallback to ensure reliability even when cloud services are unavailable. To ensure trust and accuracy, we built four automated test suites with 64 checks covering scoring logic, recommendation flips, edge cases, and system reliability.

Challenges we ran into

Our biggest challenge was making AI essential rather than cosmetic. Early versions felt like a calculator with an AI summary attached, so we redesigned the experience to make AI the reasoning layer while keeping the deterministic engine responsible for the final ranking. We also overcame cloud integration, reliability, and security challenges, ensuring that the application remains responsive, protects credentials, and never silently alters a user's decision inputs. Throughout development, we focused on preserving trust and keeping the user in control.

Accomplishments that we're proud of

We are proud to have built a deployed CrewAI-powered agentic system that goes beyond traditional decision matrices and chatbots. Our proudest innovation is The Hinge, which identifies exactly what a decision turns on and directs the user's attention to the information that matters most. We are also proud of our commitment to honest uncertainty through Fragility Scores, scenario testing, and sensitivity analysis rather than presenting false confidence. Responsible AI is embedded into the architecture: the AI advises but never decides, transparent math owns the recommendation, and the user always makes the final call. The application is fully tested, reliable, and publicly accessible.

What we learned

We learned that the hardest part of decision-making is not arithmetic. It is identifying assumptions, understanding uncertainty, and seeing what we may have overlooked. Combining deterministic decision science with generative AI created a system that is both trustworthy and insightful. We also gained valuable experience building multi-agent systems with CrewAI, deploying secure cloud architectures with AWS Lambda, implementing sensitivity analysis from first principles, and rigorously testing an AI-powered application under real-world constraints.

What's next for Aristotle - AI Decision Coach

Our vision is for Aristotle to become a lifelong decision companion. Next, we plan to introduce cross-device accounts and journals using AWS Cognito and DynamoDB, enabling users to preserve and revisit their reasoning over time. We also want to build a coaching loop that checks back after major decisions, helping people learn from outcomes and improve future decision-making. Future enhancements include additional decision archetypes, a custom decision builder, mobile support, and Amazon Bedrock integration. Our mission remains unchanged: helping people discover what their decision truly turns on, how fragile that conclusion is, and what they should learn before making the final call.

Built With

Share this project:

Updates