Inspiration:

The initial spark for Issue DetectiveAI came from observing and experiencing the myriad of challenges we face when sifting through vast amounts of attachments with important and value information. There's an inherent complexity in understanding, categorizing, and finding resolutions for issues, especially when they're presented in diverse formats or embedded within images. We propose a tool that could seamlessly read, understand, and even suggest solutions – combining the potential of Optical Character Recognition (OCR) with the intelligence of AI. This application not only streamlines processes but also significantly speeds up the work of employees who no longer need to manually rewrite error messages and search for issue solution, thus increasing efficiency and productivity.

What We Learned:

In creating Issue DetectiveAI, our technological odyssey imparted invaluable insights:

Languages: We capitalized on JavaScript for its dynamic frontend capabilities and Python for its potent backend and AI libraries.
OCR Insights: Our exploration with TesseractOCR and Google Cloud OCR highlighted the nuances of OCR tools and their text recognition efficacy across diverse backgrounds.
AI Dive: Leveraging TensorFlow and GPT-2, we delved into machine learning and NLP, appreciating their prowess in simulating human-like textual interpretations.
Integration: The Forge API enabled smooth integration into Jira, while Docker and Google Cloud Run enhanced our deployment, emphasizing scalability.

How We Built It:

Jumping into the hackathon midway, we quickly spotted a compelling opportunity and were eager to contribute. Motivated by a personal challenge we encountered with Jira, we sought a streamlined solution. With the Forge API as our foundation, we seamlessly blended our tool into Jira's framework. To ensure precise text capture, we zeroed in on a top-tier OCR engine. In terms of AI, we leaned into sophisticated NLP algorithms. We dedicated substantial effort to train these algorithms with datasets, aiming for them to present relevant solutions across diverse issues. The result? Issue DetectiveAI – a real-time diagnostic tool that can scan, decipher, and recommend solutions instantaneously.

Challenges Faced:

Balancing this project with our full-time jobs was a challenge in itself. Late nights and dedication were constants. We grappled with blending OCR, AI, and the Forge API efficiently. The varied quality of text in images posed hurdles for OCR. In AI, tuning our model to grasp the diverse Jira issues and provide the right solutions took iterative refinement. Meanwhile, crafting a user-friendly experience without compromising on functionality was key. Through it all, our aim was clear: stay driven and avoid needless sidetracks.