Inspiration

At Izymes we have been using AI to assist with valuable content creation for our blogs, requirement specs and product ideas which we track in Confluence. The constant back-and-forth between the AI web interface and Confluence was so annoying! Why not create AlBert, a tool that integrates AI assisted content creation with Confluence seamlessly! No back-and-forth, copy and paste and context switching.

What it does

AlBert takes page title and page content (only the content you select, not the entire page!) and engineers sophisticated prompts for ChatGPT based on what you want to do including

  • Create content from scratch, just provide a page title
  • Elaborate on selected content - provide more details and insights
  • Create actionable task lists based on selected content
  • Rewrite and improve content accuracy and readability
  • Summarize selected content to create abstracts and excerpts.

How we built it

Based on a forge ai template with custom-ui and resolvers plus the ability to call out from forge runtime to 3rd party API such as api.openai.com it was quite straighforward to implement AlBert's business logic.

Challenges we ran into

  1. Not being able to plug into the Confluence cloud Editor was a massive shock! Atlassian, there is massive potential to enhance the edit experience if such integrations were made available
  2. The 25s forge completion timeout and slowness of OpenAI chat completion posed a real challenge, but we worked around it.

Accomplishments that we're proud of

AlBert is a tool that we are already using internally (instant adoption) because our folks just love it. Getting good at programmatic, context sensitive prompt engineering.

What we learned

The possibilities of AI assisted content creation and manipulation are absolutely endless. Forge allows developers to create re-usable patterns that makes it easy to add new functionality and features in no time.

What's next for AlBert: Effortless content creation in Confluence

Connecting AlBert to any domain, knowledge base, documentation resources, including customer's own Confluence instance via LangChain to tap into domain specific knowledge in order to make content creation even more valuable and tailored to customers.

Share this project:

Updates