What If We Could Search Human Experiences Instead of Opinions?

Problem

Every person who achieved something meaningful made a few decisions that quietly changed the direction of their life. Choosing an internship. Accepting a job offer. Pursuing higher studies. Joining a startup. Starting a company.
The interesting part is that almost nobody chooses to make a bad decision.
People make the best decision they can with the information they have at that moment. When things don't work out, it's usually because the guidance they found wasn't meant for someone in their situation.
Before making an important decision, most of us do exactly the same thing.
We search Reddit. Watch YouTube videos. Read LinkedIn posts. Ask seniors. Maybe even ask AI assistant.
Now imagine it's 11:30 PM.
Your internship offer expires tomorrow morning.
Your browser has twenty-three tabs open. Reddit is full of conflicting opinions. LinkedIn is full of success stories. YouTube has a roadmap for everything. AI gives you a perfectly balanced list of pros and cons.
Everyone has advice.
Yet you're still stuck.
Because the question you're trying to answer isn't "Which option is better?"
It's "What happened to someone who was exactly where I am today?"
Someone with a similar background. Similar skills. Similar goals. Someone who made this decision a few years ago, lived through the consequences, and can unknowingly help you make a better decision today.
The problem isn't that information doesn't exist.
It's that those experiences are scattered across Reddit threads, LinkedIn posts, blogs, podcasts, videos, and conversations that slowly disappear over time. Finding information is easy. Finding the right experience is incredibly hard.
Now imagine something different.
What if you could explore real human journeys instead of isolated opinions?
What if you could see the different paths people took from a situation just like yours, understand why they chose them, what happened afterwards, and the mistakes they wish they'd avoided?
What if, instead of generic advice, you had a system that could learn from real experiences to help you understand which path is most likely to work for someone like you?
That's the idea that eventually became PathFinder.

PathFinder - The Search Engine for Human Experience.

At its core, PathFinder is built on a simple belief:

The best guidance doesn't come from people telling you what to do. It comes from understanding what happened to people who were once in your place.

Instead of treating every experience as just another Reddit post, LinkedIn update, YouTube video, or blog article, PathFinder turns it into something much more valuable.
A journey.
Every person's background, skills, experiences, goals, decisions, challenges, and outcomes become connected pieces of a much larger story. Not just where they ended up, but how they got there. The difficult choices they faced. The risks they took. The mistakes they made. The moments where they changed direction. The lessons they learned along the way.
As more people share their journeys, PathFinder starts building something that's impossible to find on today's internet: a map of real human experiences.
Now, when someone comes to PathFinder with an important decision, the goal isn't to generate another opinion.
The goal is to answer a much more meaningful question.
"Who has already walked this path before me?"
So instead of searching through hundreds of disconnected posts, PathFinder searches through thousands of connected journeys to find people with similar backgrounds, similar skills, similar goals, and similar circumstances. Then it shows the different decisions they made, how their journeys unfolded, and what outcomes those decisions eventually led to.
Sometimes you'll discover that the path you were planning to take worked incredibly well for people like you.
Sometimes you'll discover that people with similar profiles succeeded by taking a completely different route.
And sometimes, the most valuable insight isn't what worked.
It's understanding why something didn't.
Because failures, pivots, and regrets often teach us as much as success stories do. That's why PathFinder isn't designed to showcase only achievements. It's designed to capture the complete journey, including the uncertainty, the setbacks, and the lessons people wish they'd known earlier.
We believe that's a fundamentally different way of making decisions.
Not by asking for another opinion.
But by learning from the collective experience of people who have already lived through the decision you're about to make.

The internet made information searchable.
PathFinder makes human experiences searchable.

Why Existing Platforms Still Leave Us Stuck ?

A fair question is:
"Don't Reddit, LinkedIn, YouTube, and even AI assistants already solve this problem?"
Not quite.
They each solve a different piece of it.
Reddit gives us opinions, but not credibility. LinkedIn shows us achievements, but rarely the failures, pivots, and uncertainty that came before them. AI can summarize information remarkably well, but it can only work with the information it has. It doesn't know which experiences belong together or how a person's journey unfolded over time.
PathFinder isn't trying to replace any of them.
It's trying to connect what they've never connected.
Instead of indexing posts, videos, or articles, PathFinder indexes journeys. Instead of matching keywords, it matches people. And instead of answering questions with generic advice, it lets you learn from the complete experiences of people who were once in a situation remarkably similar to yours.
We don't think the future of decision-making is about generating better opinions.
We think it's about making human experience searchable.

A Decision Starts with a Conversation

Most search engines start by asking you what you're looking for.
PathFinder starts by asking who you are.
When a user opens PathFinder, they don't immediately receive recommendations. Instead, the system spends a few moments understanding the context behind the decision. Someone deciding between a startup and a campus placement isn't defined only by that question. Their background, previous experiences, current goals, skills, and constraints all influence what the right answer might be.
Through a short AI-guided conversation, PathFinder gradually builds this context. Once it understands the user's situation, it searches for people whose journeys genuinely resemble theirs, not just people who mentioned similar keywords online.
The result isn't a list of links. It's a collection of real journeys. Users can explore the decisions others made, visualize how their careers evolved over time, compare different outcomes, and identify patterns that would be almost impossible to notice by reading hundreds of individual posts.
If someone wants an AI recommendation, it's available. If they prefer drawing their own conclusions, they can simply explore the underlying journeys, statistics, and decision graphs themselves.
Every journey shared also becomes part of PathFinder's growing knowledge graph, making the platform more useful with every contribution.

Under the Hood: How PathFinder Works

While using PathFinder feels like having a conversation, there's a lot happening behind the scenes to ensure every recommendation is grounded in real human experiences rather than generated opinions.
When a user asks a question, either by typing or speaking, the system first builds an understanding of their intent. Instead of simply matching keywords, it identifies the user's background, goals, current situation, and the decision they're trying to make. Voice queries are transcribed using Sarvam AI before entering the same processing pipeline.
Once the query is understood, PathFinder searches its Neo4j-powered knowledge graph, where every person's background, skills, experiences, decisions, transitions, and outcomes are connected into complete journeys. Rather than relying on a single search technique, the platform combines semantic vector search, full-text search, and graph traversal to retrieve journeys that genuinely resemble the user's situation, not just those containing similar words.
The retrieved journeys are then analyzed by the AI layer to identify recurring patterns, compare different decision paths, extract practical lessons, and generate evidence-backed insights. Instead of inventing answers, the AI reasons over real experiences and presents concise recommendations while allowing users to explore the underlying journeys and decision graphs themselves.
The same pipeline also works in reverse. When users choose to contribute their own story, PathFinder guides them through an AI-assisted conversation to capture their journey. Experiences, skills, goals, decisions, milestones, and outcomes are automatically structured into a graph, while supporting proofs such as GitHub repositories, certificates, resumes, or other documents help strengthen the credibility of each journey. Every verified contribution expands the knowledge graph, making future recommendations richer and more reliable.
The overall architecture powering this workflow is illustrated below.

At a high level, the system is organized into five layers. The Expo mobile application provides a conversational interface for discovering and sharing journeys. A Node.js backend orchestrates authentication, request handling, journey processing, and AI workflows. The intelligence layer combines Gemini, Groq, and Sarvam AI, each responsible for different tasks ranging from reasoning and intent understanding to multilingual voice interactions. At the core of the platform is Neo4j AuraDB, which stores interconnected journey graphs and enables graph traversal, semantic retrieval, and similarity search. Supporting services such as Redis, Cloudinary, and Clerk handle caching, media storage, and authentication, allowing the core pipeline to remain focused on transforming real experiences into meaningful decision intelligence.

Engineering the Experience

The real challenge was designing a system capable of representing human journeys—experiences that evolve over time, influence future decisions, build new skills, and connect to changing goals.
This realization shaped our first architectural decision. We built PathFinder on Neo4j AuraDB, where users, goals, experiences, skills, transitions, and proofs exist as interconnected nodes rather than isolated records. Representing journeys as a graph naturally models career pivots, multiple goals, and decision chains, while enabling efficient traversal without complex database joins. Neo4j isn't simply where PathFinder stores data—it serves as the platform's memory.

Once journeys became a graph, retrieval became a contextual reasoning problem rather than a keyword search problem. Users rarely describe situations using the same words as the people who experienced them. PathFinder therefore combines semantic vector search, graph traversal, and relationship expansion. Experiences are retrieved through embeddings, enriched with connected goals, skills, and transitions, and then ranked to surface journeys that match situations instead of keywords.
The reasoning pipeline is intentionally modular. Groq handles low-latency structured tasks such as conversational parsing, while Gemini performs deeper reasoning, onboarding extraction, journey summarization, transition inference, proof verification, and synthesis of evidence-backed insights. Separating these responsibilities keeps interactions fast without sacrificing reasoning quality.
To make contributions effortless, onboarding begins as a conversation rather than a form. AI converts natural language into structured goals and experiences that users can review before submission, combining conversational simplicity with reliable structured data.
Accessibility was equally important. Sarvam AI enables multilingual speech input, translation, and localized voice responses while preserving the same retrieval and reasoning pipeline regardless of the interaction language.
On the client side, Expo gave us the flexibility to build a single cross-platform application while rapidly iterating throughout the hackathon. Features could be tested, refined, and deployed quickly, letting us focus our effort on improving the product rather than managing separate mobile codebases.
Supporting services remain intentionally lightweight. Node.js and Express orchestrate workflows, Clerk manages authentication, Upstash Redis stores onboarding sessions and caches expensive operations, while Cloudinary manages uploaded proof documents.
Every architectural decision followed the same principle:

represent journeys the way people actually experience them, not the way databases traditionally store them.

The result is a system that reconstructs context, discovers decision patterns, and transforms individual experiences into collective decision intelligence.

Challenges We Faced

One major challenge was the cold start problem. A journey platform is only as valuable as the experiences it contains. To ensure meaningful recommendations from day one, we curated realistic seed journeys while designing the platform so that every new contribution continuously improves retrieval quality.

Another challenge was modeling nonlinear human journeys. Careers rarely follow a straight path; people pursue multiple goals, pivot, revisit interests, and make unexpected decisions. Capturing all of this in a way that remains searchable and meaningful required us to rethink how experiences should be modeled. This ultimately reinforced our decision to build the platform around a graph rather than traditional relational data structures.
Finding the right journeys proved to be much harder than finding matching journeys. Two users can describe the same situation using completely different words, while two identical queries can come from people with entirely different backgrounds and constraints. Instead of relying solely on keywords, we focused on understanding context and combining semantic retrieval with graph expansion to surface experiences that were genuinely relevant.
Trust was another challenge we couldn't ignore. If PathFinder is meant to influence important life decisions, the information it provides needs to be credible. We introduced proof uploads, AI-assisted validation, verification badges, community moderation, and reputation signals to encourage authentic contributions while reducing misleading or low-quality content.

Where PathFinder Goes Next

Hackathons are where ideas begin. The real challenge is building something that continues to grow long after the event ends.
HackHazards gave us the opportunity to turn an ambitious idea into a working prototype. More importantly, it challenged us to think beyond building features and focus on solving a problem that millions of people face while making life-changing decisions.
The most impactful platforms on the internet all share one characteristic. Their value compounds with every contribution. Every video uploaded to YouTube makes it a richer source of knowledge. Every repository on GitHub helps developers build faster. Every article on Wikipedia strengthens the world's collective understanding. Their greatest strength is not just the technology behind them, but the communities that continuously make them better.
We believe PathFinder has the potential to grow in the same way.
Every journey shared adds another layer to the knowledge graph. Every decision, challenge, success, failure, and lesson strengthens future recommendations. As more people contribute, the platform becomes increasingly capable of connecting users with experiences that truly match their own situation.
Looking ahead, we envision PathFinder expanding beyond career guidance into entrepreneurship, higher education, research, open source, and many other domains where learning from real experiences matters more than generic advice. As the graph grows, it will not only retrieve similar journeys but also uncover patterns, reveal unconventional paths, and help people understand not just what decisions worked, but why they worked.
The internet made information searchable.

Our vision is to make human experience searchable.

If YouTube transformed how we share knowledge, GitHub transformed how we build software, and Wikipedia transformed how we document information, PathFinder can transform how people learn from the experiences of others.
Because sometimes, the best guidance is not another opinion.
It is knowing that someone has already walked the path you are about to take.