“Let me check; I know I’ve seen that somewhere – maybe in my email, Confluence, Jira, somewhere.”
If you’ve worked in a fast-paced environment, you’ve probably heard or said this phrase a hundred times. It’s a symptom of a deeper problem: information scattered across systems, each siloed in its own way.
For me, working in a medical company where data is both abundant and fragmented, was a daily reality. Couple that with leading DevOps teams, where I often found myself answering questions about information that already existed, and it was clear: we needed a better way to connect the dots.
The Problem That Sparked an Idea
The issue wasn’t that we lacked documentation or tools – if anything, we had plenty. The challenge was knowing where to find the right piece of information at the right time.
One day, after yet another “where can I find this?” conversation, I started wondering:
What if I didn’t have to check? What if I could just ask?
That’s how the idea for Sage began: a tool that doesn’t just search but helps you access your scattered data in a conversational, human-friendly way.
The Road to Sage
The early days were about proving the concept. The first version of Sage was just a script tied to a single source, answering basic queries. It wasn’t fancy, but it worked well enough to hint at what could be possible.
The real shift happened when I started incorporating embedding engines and Large Language Models (LLMs). These technologies allowed Sage to go beyond simple keyword searches; it could now understand the context, summarize content, and prioritize what mattered most.
Here’s where it started to feel real:
[source.confluence]
username = "your_confluence_username"
server = "https://yourcompany.atlassian.net/wiki"
spaces = ["SPACE1", "SPACE2"]
With just a few configuration lines, Sage could connect to our Confluence spaces and make them searchable in a conversational way.
Sage enables you to communicate with your data in a natural and intuitive way. Whether you’re looking up information, summarizing content, or integrating with external tools, Sage is your personal data assistant, ready to help.
Key Features
Sage currently offers the following functionalities:
- Data Source Queries: Interact with multiple data sources directly through conversational prompts.
- Integrated Tools: Access tools like calculators, search engines, and Jira issue summarizers within the chat.
- Agent Mode: Activate Sage as an AI agent to handle complex queries and perform autonomous actions.
- Configuration Simplicity: Set up Sage quickly by specifying your desired tools and sources in a configuration file.
- Agent Capabilities: Utilize Sage in agent mode for advanced tasks.
- Continuous Source update: Continuously update the sage data sources via sage’s data loader process whenever your data gets updated
- Filter-out Data Source: Choose to interact with a specific source or all your configured data sources
Building for a Specific Need
What sets Sage apart from other LLM-based tools isn’t the technology – it’s the focus. While many AI applications aim to do everything, I wanted Sage to tackle a particular problem: simplifying access to scattered data.
It’s not trying to replace your tools; it’s trying to make them work better together. Whether it’s Confluence, Gitlab, or even plain-text files, Sage bridges the gap, letting you interact with your data without needing to know exactly where it lives.
What’s Next?
Sage is still evolving. There’s plenty on the horizon, from dynamic model switching to better UI designs. But for now, it’s functional, and that’s a win in my book.
If you’re curious to explore Sage, it’s open-source and available on GitHub. You’re welcome to dive in, experiment, or even contribute to its development.
But more than anything, Sage is a reflection of why I enjoy building things. It’s not just about solving problems – it’s about simplifying life for the people around us. And if it means I never have to hear, “Let me check, I know I’ve seen that somewhere,” ever again, then I’d say it’s been worth the effort.
