design & DEV JOURNAL · 2026
Obscura
Obscura
Obscura
What is this about
What is this about
Building Obscura, an investigation workspace from the ground up.
Building Obscura, an investigation workspace from the ground up.
Services
Product Design
Product Design
Client
Personal Project
Personal Project
Platform
Web
Web
Live Website
Overview
Overview
Obscura is a personal investigation tool for collecting clues, organizing information, and connecting the dots. From case files and reference links to people and locations, it brings everything into one place, making it easier to explore ideas and keep track of the details that matter.
Obscura is a personal investigation tool for collecting clues, organizing information, and connecting the dots. From case files and reference links to people and locations, it brings everything into one place, making it easier to explore ideas and keep track of the details that matter.
THE Process
Where it started
Where it started
It started while watching a crime drama. I was drawn to how detectives piece together an investigation, pinning photographs to boards, collecting documents, connecting people, and following leads until things start to make sense. It got me thinking: what if I could do something similar with my own research? A place to collect interesting things, keep track of details, and connect the dots instead of leaving everything scattered across different apps. I also started exploring the idea of having a trained AI agent alongside me, helping me make sense of the information and uncover connections I might have missed. That thought became Obscura.
It started while watching a crime drama. I was drawn to how detectives piece together an investigation, pinning photographs to boards, collecting documents, connecting people, and following leads until things start to make sense. It got me thinking: what if I could do something similar with my own research? A place to collect interesting things, keep track of details, and connect the dots instead of leaving everything scattered across different apps. I also started exploring the idea of having a trained AI agent alongside me, helping me make sense of the information and uncover connections I might have missed. That thought became Obscura.
A bit of everything
Before settling on a direction, I started throwing ideas around. I looked at detective boards, investigation walls, case files, research tools, and anything that could help me figure out what Obscura should feel like. I wasn't trying to lock myself into a specific layout or style just yet. I wanted to explore different possibilities, mix things up, and see what made sense for the kind of experience I had in mind. Some ideas were more about recreating the classic detective setup, while others leaned towards a more structured digital workspace. I collected references, sketched out rough concepts, and experimented with different ways to bring the two together.
It was a bit of a mess at this point, but that was the point. I needed to get the ideas out before figuring out which ones were worth pursuing.
Before settling on a direction, I started throwing ideas around. I looked at detective boards, investigation walls, case files, research tools, and anything that could help me figure out what Obscura should feel like. I wasn't trying to lock myself into a specific layout or style just yet. I wanted to explore different possibilities, mix things up, and see what made sense for the kind of experience I had in mind. Some ideas were more about recreating the classic detective setup, while others leaned towards a more structured digital workspace. I collected references, sketched out rough concepts, and experimented with different ways to bring the two together.
It was a bit of a mess at this point, but that was the point. I needed to get the ideas out before figuring out which ones were worth pursuing.
Shaping the experience
Shaping the experience
CaseFleet
Even though its more used for legal purposes it focuses on organising evidence, building timelines, and connecting facts to legal claims. I wanted to explore a more visual, flexible way to work with clues.
Even though its more used for legal purposes it focuses on organising evidence, building timelines, and connecting facts to legal claims. I wanted to explore a more visual, flexible way to work with clues.
CORE
Helps investigators connect people, places, and events. I was interested in making that process feel more approachable and easier to navigate.
Helps investigators connect people, places, and events. I was interested in making that process feel more approachable and easier to navigate.
Redline Discovery
Focuses on managing and reviewing large volumes of documents. I wanted to explore bringing different types of evidence together in one interactive workspace.
Focuses on managing and reviewing large volumes of documents. I wanted to explore bringing different types of evidence together in one interactive workspace.
By analysing existing investigation tools, I started understanding how they handle evidence, people, and relationships, and where I could take a different approach. I began thinking about what I'd want in my own app: a way to collect evidence, track people and locations, save useful links, and connect everything as an investigation evolves. I also wanted an AI assistant that could help make sense of the collected information, all within one workspace that feels intuitive rather than complicated.
By analysing existing investigation tools, I started understanding how they handle evidence, people, and relationships, and where I could take a different approach. I began thinking about what I'd want in my own app: a way to collect evidence, track people and locations, save useful links, and connect everything as an investigation evolves. I also wanted an AI assistant that could help make sense of the collected information, all within one workspace that feels intuitive rather than complicated.
Figuring out the layout
Figuring out the layout
With the core pieces in place, I started figuring out how they could live together on screen. I needed the investigation board to be the main focus, while keeping case details, evidence, and other tools easy to reach. I also wanted the AI assistant to feel like part of the investigation, not something that pulls me away from the work. Giving it a dedicated space alongside the board felt like a natural fit. After trying to work out how everything should fit together, I landed on a layout with navigation on the left, the investigation workspace in the center, and the assistant on the right. This gave the board room to breathe while keeping the rest of the tools within reach.
With the core pieces in place, I started figuring out how they could live together on screen. I needed the investigation board to be the main focus, while keeping case details, evidence, and other tools easy to reach. I also wanted the AI assistant to feel like part of the investigation, not something that pulls me away from the work. Giving it a dedicated space alongside the board felt like a natural fit. After trying to work out how everything should fit together, I landed on a layout with navigation on the left, the investigation workspace in the center, and the assistant on the right. This gave the board room to breathe while keeping the rest of the tools within reach.
Visual language
I experimented with different visual directions and tested them in the actual interface to see what worked best. I wanted Obscura to have its own character without making the workspace feel overwhelming. The board was a big part of this exploration. I wanted to capture the feeling of a detective's evidence board without making it look cluttered or visually heavy. I experimented with ways to present clues, cards, and connections while keeping the layout clean and easy to follow. I kept refining the colors, icons, and details, stripping away anything that competed for attention. The goal was to retain the investigative feel while making the experience comfortable to work with.
I experimented with different visual directions and tested them in the actual interface to see what worked best. I wanted Obscura to have its own character without making the workspace feel overwhelming. The board was a big part of this exploration. I wanted to capture the feeling of a detective's evidence board without making it look cluttered or visually heavy. I experimented with ways to present clues, cards, and connections while keeping the layout clean and easy to follow. I kept refining the colors, icons, and details, stripping away anything that competed for attention. The goal was to retain the investigative feel while making the experience comfortable to work with.
Adding a case
Adding a case
I wanted creating a case to be as simple as possible. Users can enter their case details, and the AI analyses the information to identify key people, places, and other details, adding them as nodes to the board. This gives users a starting point for their investigation without having to organise everything manually.
I wanted creating a case to be as simple as possible. Users can enter their case details, and the AI analyses the information to identify key people, places, and other details, adding them as nodes to the board. This gives users a starting point for their investigation without having to organise everything manually.
I started bringing everything together on the case screen, with the investigation board at the center, case tools within reach, and an AI assistant alongside it. I wanted to make it easy to explore evidence, connect clues, and ask questions without constantly switching between screens.
I started bringing everything together on the case screen, with the investigation board at the center, case tools within reach, and an AI assistant alongside it. I wanted to make it easy to explore evidence, connect clues, and ask questions without constantly switching between screens.
Evidence inbox
Evidence inbox
The Evidence Inbox makes collecting information easier. Users can upload files directly or use a unique email address to forward relevant emails into a case, keeping important information in one place without having to switch between mail and the investigation.
The Evidence Inbox makes collecting information easier. Users can upload files directly or use a unique email address to forward relevant emails into a case, keeping important information in one place without having to switch between mail and the investigation.
Figuring out the tech
Figuring out the tech
Before committing to the design, I wanted to figure out how to bring Obscura to life. The investigation board was the biggest challenge. I needed a way to arrange evidence cards, connect related information, and keep everything interactive without making the experience feel heavy. I settled on Next.js as the foundation and explored the tools available around it to figure out how to bring the different pieces together. The goal was to find a setup that gave me enough flexibility to experiment while keeping the app practical to build and maintain.
Before committing to the design, I wanted to figure out how to bring Obscura to life. The investigation board was the biggest challenge. I needed a way to arrange evidence cards, connect related information, and keep everything interactive without making the experience feel heavy. I settled on Next.js as the foundation and explored the tools available around it to figure out how to bring the different pieces together. The goal was to find a setup that gave me enough flexibility to experiment while keeping the app practical to build and maintain.
AI & Case Sandbox
AI & Case Sandbox
At the core of the Case Sandbox is Watson, an AI investigator powered by Groq models and guided by the context of a seasoned investigator. It analyses images, documents, and PDFs to extract details that can be searched and explored simply by asking questions. It can also turn key findings into nodes on the investigation board, making connections easier to explore.
Watson keeps each case focused on its own evidence, highlights confidence levels, and challenges assumptions instead of agreeing without facts. It questions our findings, separates facts from suggestions, and pushes for evidence before drawing conclusions. Watson isn't fine-tuned yet; that's something I'm planning to explore as the project evolves.
At the core of the Case Sandbox is Watson, an AI investigator powered by Groq models and guided by the context of a seasoned investigator. It analyses images, documents, and PDFs to extract details that can be searched and explored simply by asking questions. It can also turn key findings into nodes on the investigation board, making connections easier to explore.
Watson keeps each case focused on its own evidence, highlights confidence levels, and challenges assumptions instead of agreeing without facts. It questions our findings, separates facts from suggestions, and pushes for evidence before drawing conclusions. Watson isn't fine-tuned yet; that's something I'm planning to explore as the project evolves.
Evidence Board
Evidence Board
The evidence board is the heart of the investigation. When a case is created, Watson automatically generates nodes from the available information and connects related details. Each piece of evidence can also have suggested nodes, helping uncover possible connections and new leads. I used React Flow to build the interactive board, giving users the freedom to explore evidence, move things around, and connect the dots as the investigation unfolds.
The evidence board is the heart of the investigation. When a case is created, Watson automatically generates nodes from the available information and connects related details. Each piece of evidence can also have suggested nodes, helping uncover possible connections and new leads. I used React Flow to build the interactive board, giving users the freedom to explore evidence, move things around, and connect the dots as the investigation unfolds.
What's Next
What's Next
There's still a lot left to build. The Evidence Inbox needs more work, and several other modules are still on my list. Watson also has room to grow, especially in how it analyses evidence, questions assumptions, and uncovers connections. For now, seeing the core idea come to life feels satisfying in itself. I'm planning to expand the project, experiment with new features, and learn from people who try it. Rather than building only around my own assumptions, I want feedback to help shape what comes next. There's a long way to go, but I'm excited to see where this takes me. Hopefully, it becomes a genuinely useful tool for anyone who enjoys investigating, connecting clues, and making sense of scattered information.
There's still a lot left to build. The Evidence Inbox needs more work, and several other modules are still on my list. Watson also has room to grow, especially in how it analyses evidence, questions assumptions, and uncovers connections. For now, seeing the core idea come to life feels satisfying in itself. I'm planning to expand the project, experiment with new features, and learn from people who try it. Rather than building only around my own assumptions, I want feedback to help shape what comes next. There's a long way to go, but I'm excited to see where this takes me. Hopefully, it becomes a genuinely useful tool for anyone who enjoys investigating, connecting clues, and making sense of scattered information.