Coming soon: An open source approach for using natural language for analyzing source system readouts.
Today in Mattermost you can flow readouts from source systems like GitHub into channels, and have natural language questions answered by analyzing “headline summaries”, which includes data and metadata pushed into the channel using a webhook or integration (e.g. Issue Title, Date-Time, Author, etc.).
We’re preparing to release the ability to access “source context” from the underlying GitHub issues and pull requests, which pulls in information from links to GitHub issues using GitHub accounts that have been securely connected to the Mattermost user account–letting you access private data in your backends.
This lets you ask interesting and powerful questions across your source systems. In the GitHub example:
- Scan the last 30 GitHub issues in this channel, and find the one that looks easiest to complete.
- What issue looks like the hardest to complete?
- Which issue seems like it might have security implications?
The framework for this connectivity is open source, so you can take the source code from our GitHub example and adapt it to your custom systems for real-time ingestion and analysis.
All this and more coming soon to https://openops.mattermost.com
