Nuance is a high volume human-intelligence task queue. It has two major components:
- Importers: Infrastructure for actively pulling data from external sources.
- Queues: Routing rules and UI for processing human intelligence tasks quickly.
Broadly:
- Sources define where items come from (e.g., email, twitter, GitHub, a web form, an RSS feed).
- Items are created when new stuff shows up on a source (a new email, a new tweet, a new pull request, a new form submission, etc).
- Sources define routing rules for items which are Herald-like logic to route items into Queues (or possibly do other things with them automatically).
- In a Queue, a human works through items rapidly, making decisions about what to do with them.
Use cases might include:
- Support queue: pipe email into a queue. Humans can respond directly or escalate them into tasks.
- Monitor Twitter: pipe tweets into a queue. Humans can keep tabs on what people are saying about you and reply.
- Content reports: If your application has a "report" feature, support representatives can work through user reports and take action on them.
- Special "Security" or "Feature Request" or "Report Bug" form: special web form that routes into Nuance, putting a layer of indirection between Maniphest and external users.
- Application-to-Queue: pipe query results from another application into a queue, then work through them quickly (e.g., triage a lot of tasks or handle a bunch of audits).
- Import-to-Application: Pipe items directly into an application (automatically create tasks or feed stories from external sources).
The primary driver for this application is currently upstream interest in better support queues for the Phacility cluster.