Building Insight

Insight was an agile dashboarding tool I co-founded to bring transparency to the software development field. It was born out of a classic “scratch your own itch” moment and evolved into a SaaS product used by hundreds of teams worldwide.

The concept was simple: Development teams generate massive amounts of data in tools like Pivotal Tracker and GitHub, but that data is often hidden in lists and tickets. Insight transformed that raw data into beautiful, real-time dashboards that managers, developers, and clients could understand at a glance.

The Origin Story

The idea started when my co-founder, Jeremy Fabre, and I were working as developers on a small team. We wanted to prove to our boss that our agile practices were actually working—that we were shipping faster and improving code quality.

We looked for tools to visualize our velocity and burn-down charts, but everything on the market was overly complicated. The alternative was manually copying data into Excel sheets to generate daily reports. As developers, the idea of manual data entry was painful. We wanted automation.

We built a prototype over a weekend. It connected to the Pivotal Tracker API, pulled our data, and displayed it on a TV screen in the office. Immediately, the dynamic changed. The “black box” of development was opened, and everyone—from management to the dev team—had a shared understanding of progress.

The “Lean” Approach

Insight was built strictly following the principles of The Lean Startup and The Four Steps to the Epiphany. We didn’t just guess what features people wanted; we got out of the building.

Before writing the bulk of the code, we conducted over 1,500 interviews with managers, developers, CEOs, and CTOs via Skype and email. We learned that while every team used a slightly different process (Scrum, Kanban, Hybrid), they all shared the same pain point: a lack of visibility.

This feedback loop drove our product decisions. We realized we needed to support custom workflows and flexible widgets because no two teams worked exactly alike. We focused on a narrow vertical first—Pivotal Tracker users—to ensure we solved one problem perfectly before expanding to Trello and GitHub.

Under the Hood

The Tech Stack

Insight was built during a transitional era in web development, and our stack reflected the cutting-edge (and sometimes experimental) tools of the time.

• Backend: Ruby on Rails served as the API layer. It handled the heavy lifting of synchronizing with external APIs (Pivotal, GitHub), normalizing the data, and storing it in Postgres.
• Frontend: We used CoffeeScript and Batman.js. Batman.js was a binding framework similar to the early days of Ember or Angular, chosen for its “Rails-style” approach to frontend code.
• Visualization: The core value of Insight was the charts. We used D3.js to build custom, interactive visualizations that could render complex metrics like velocity, cycle time, and rejection rates.
• Layout: We implemented a grid-based drag-and-drop system (using libraries like Gridster) that allowed users to resize and arrange widgets to fit their specific needs, whether for a desktop monitor or a wall-mounted TV.

Technical Challenges

The biggest technical hurdle was data synchronization. We were effectively a real-time analytics layer on top of other SaaS products.

1. Rate Limits: We had to be extremely careful with API rate limits from providers like Pivotal Tracker. We implemented aggressive caching strategies and background workers to keep data fresh without getting blocked.
2. Data Normalization: A “ticket” in Trello looks very different from a “story” in Pivotal. We had to build an abstraction layer that normalized these different data structures into a common format that our charting engine could understand.
3. Real-time Updates: Users expected the dashboard on the TV to update instantly when they closed a ticket. We used polling and eventually webhooks to ensure the “TV mode” was always accurate.

The Outcome

Insight grew steadily and acquired hundreds of paying customers. It became the source of truth for over 200 teams, helping them save time on reporting and focus on shipping code.

Although the product is now shut down, it remains a project I am incredibly proud of. It taught me the value of talking to customers before coding, the power of data visualization, and the complexities of building a business on top of other platforms’ APIs.