Risk Agent
Agent Mode and Guide Mode
Prevention Risk Playbook

Compass AI - Project Story

Inspiration

The inspiration for Compass AI came from a sobering statistic: in 2015, 50% of business failures were traced back to one flawed critical decision. As founders of inRoot.io, we've spent years helping organizations conduct Root Cause Analysis after problems occur. But we kept asking ourselves - what if we could prevent those problems in the first place?

We realized we're living through a fundamental shift from the Information Age to the Knowledge Age. During the Information Age, we had endless data at our fingertips, but we often ended up "shooting from the hip" on critical decisions, cherry-picking information that confirmed our gut instincts. AI changes everything - it doesn't just give us more information, it gives us the WHY behind the data.

The "aha moment" came when we recognized that the cost of knowledge just went to zero. What used to take weeks of expert analysis can now happen in minutes. We could democratize Root Cause Analysis for decision-making, not just problem-solving.

What it does

Compass AI is "The 2-Minute Pre-Mortem for Critical Decisions" - an AI-powered failure prevention tool that runs Root Cause Analysis on your decisions before you commit to them.

Here's how it works:

Input your critical decision (e.g., "Should I quit my job to start a consulting company?")
Answer 7 systematic questions designed to uncover failure modes, biases, and blind spots
Receive a complete pre-mortem report with failure scenarios, contingency plans, and risk assessment
Export your analysis

Instead of generic decision advice, Compass AI specifically hunts for what could go wrong and why, using proven Root Cause Analysis methodology. It transforms gut-instinct decision-making into systematic, knowledge-based choices.

How we built it

Technical Stack:

Frontend: React with TypeScript for robust component architecture
Styling: Tailwind CSS with inRoot.io's forest green brand colors (#234C3B)
AI Integration: GPT-4 with custom prompts embedding our RCA methodology
Backend: Node.js API handling LLM interactions and report generation
Export: Shareable failure prevention playbooks

Design Philosophy: We followed a Linear.app/Notion-inspired approach with clean, minimal interface design, progressive disclosure (one question at a time), and generous white space for approachability.

Development Process:

Rapid Prototyping: Started with core question flow and user journey
AI Prompt Engineering: Iterated extensively on Root Cause Analysis prompts
User Testing: Validated with real decision scenarios from our network
Polish Phase: Added animations, export functionality, and mobile responsiveness

Challenges we ran into

LLM Consistency: Getting the AI to consistently follow Root Cause Analysis methodology while maintaining conversational flow was harder than expected. We solved this with detailed system prompts and structured response formatting.

Response Quality: Balancing depth of analysis with speed of delivery. Early versions either gave shallow advice or took too long. We found the sweet spot with 7 carefully crafted questions that systematically uncover failure modes.

Mobile Experience: Fitting multi-step analysis flow on mobile screens without losing context. Implemented smart progress tracking and question persistence.

Time Constraints: With hackathon pressure, we had to make smart scope decisions - focusing on core functionality first, polish second, while staying disciplined about MVP scope.

Accomplishments that we're proud of

Systematic Innovation: We successfully translated complex Root Cause Analysis methodology into an intuitive, consumer-friendly interface that anyone can use in 2 minutes.

Real User Value: Our beta users report genuine "aha moments" - discovering blind spots and failure modes they never would have considered using traditional decision-making approaches.

Technical Execution: Built a production-ready application with clean code architecture, responsive design, and seamless AI integration under intense time pressure.

Brand Integration: Successfully extended the inRoot.io brand into a new market segment while maintaining our core expertise in Root Cause Analysis.

Knowledge Age Positioning: Created a compelling narrative around the transition from Information Age to Knowledge Age decision-making that resonates with both technical and business audiences.

What we learned

About Decision-Making Psychology:

People are remarkably bad at identifying their own cognitive biases
The same systematic thinking that works for post-incident analysis is incredibly powerful for pre-decision analysis
Users crave structure and methodology, not just generic advice

About AI Implementation:

LLM prompt engineering for Root Cause Analysis requires deep domain expertise
The quality of questions matters more than the sophistication of the AI model
Users need to feel like they're collaborating with the AI, not being interrogated by it

About User Experience:

2 minutes is the sweet spot - long enough for meaningful analysis, short enough to maintain engagement
Visual progress indicators are crucial for multi-step analysis flows
Export functionality is essential - people want to share and reference their analysis

What's next for Compass AI by inRoot.io

Immediate Integration: Compass AI will become the consumer entry point to inRoot.io's enterprise platform, creating a funnel from individual decision-makers to organizational Root Cause Analysis solutions.

Enhanced Features:

Decision Tracking: Follow-up workflows to track how decisions played out against predictions
Team Collaboration: Shared decision analysis for leadership teams and boards
Industry Specialization: Customized question sets for specific domains (healthcare, finance, tech)

Knowledge Age Platform: We see Compass AI as the foundation for a broader platform where AI doesn't just give you information, but provides systematic thinking frameworks to turn information into wisdom.

Market Expansion: This hackathon proved there's massive demand for decision intelligence - applying AI to improve decision-making, not just automate tasks. We're positioned to lead this emerging category.

The future isn't about having more data - it's about asking better questions. And that's exactly what Root Cause Analysis provides, now democratized through AI.

Built with ❤️ by the inRoot.io team - because the best time to solve problems is before they happen.