Authors:
Susan Brosnan, Jessica Snead, Patricia Hogan, Daniel Goins
Abstract:
The present idea is to automatically create and maintain a matrix of good contacts based on analyzing various systems to see who does what. This is to assist newcomers and people on new projects that do not know the proper contacts.
Background:
When new people join a company or people move to new projects, how can we make sure they are connected to the subject matter experts or people with experience, so they have the best contacts to assist the new person in problem/solution/documentation. Often new people do not know who to ask or they don't know where to find an existing solution to a problem they are encountering.
There is also the problem that the Resource Management Office (RMO) that is supposed to know who has the skills to work on which type of projects and who is assigned to what projects also does not know how to find people.
Various teams have tried to solve the problem of who a new person should ask for help. In the past those teams have tried collabServer, SharePoint, OneNote, Teams, Confluence, etc. The project manager creates a page in one of these tools with links to various tools or documentation. The page might
- list the names and roles of various team members,
- have a link to the process to create a development environment on a developer's laptop
- have a link to how to open a defect
- have a link to test environments
- link to product documentation
- link to TestRail test cases
- link to test results
- link to Agile Functional Specification
- link to Requirements document
- link to test plans and standard test plan document templates
- link to IT help or service now
- link to how to request access for gitlab, JIRA, Version One, Accurev, Confluence, email, teamroom, ....
But as time goes on the page does not get updated and the information is stale, or the company moves to a new collaboration tool collabServ to Teams to SharePoint to Confluence and the information is lost or not moved to the new tool. Often it does not tell who is the expert that you need to talk to on another project that you need to integrate with.
Description:
Our idea is to automatically create and maintain a matrix of good contacts based on analyzing various systems to see who does what. For example, the tool could analyze:
- who does gitlab check-ins for various projects
- who does gitlab code reviews
- who does gitlab code merges
- JIRA assigned developer, tester, reporter, people who comment on a story or defect, or follow the story or defect or add links to the story or defect or even just read the story or defect
- who creates test cases/matrices
- who creates meetings for various topics and who they invite as the interested parties for those topics
- maybe parse the transcripts of Teams meetings to find who knows about various topics
- who are the authors and reviewers of Design Docs
- who creates requirements documents
- who creates documentation for customer installations
- information from Services Teams statements of work for customer projects and the resources that get assigned to those projects
For the resource management office, they could get a matrix of who has ever done C basic code, C++ code, Java, Java script, Python, helm charts, shell scripts, Jenkins jobs, SQL scripts at any stage, who is a tester, who is a coder, who is a Business Analyst who writes documentation, who does lots of work in a particular area. For example, a software solutions company that writes point of sale software might want to use the tool to identify which people do promotions and loyalty work, who does payment and PIN pads, who has worked with a retailer before…
Our idea is that from this AI analysis the tool would automatically create and maintain a WordTile of people, matrix diagram, list, etc.
The tool would also have a natural language interface and be able to answer "who can help me with" questions, with a ranked confidence level on the recommendation.
The tool could also be integrated to various tools so that when someone checks code or a test case in or does other tasks the tool can prompt with an Automated Question: Did you ask ABC to review your dev, test plan, etc. This could also provide a prompt checklist for the next tasks or completion tasks to adhere to the defined process.
Supporting Art:
Much of the existing art is not using AI to look through code repositories, requirements documents, etc., to find the patterns of who worked on this area, how recently and how much did they contribute on a topic to rate how relevant their knowledge is.
The prior art does not encompass all the ideas discussed here. Many of the existing products gather the data manually or require people to tag information.
https://www.getguru.com/features/ai-suggest-expert
https://www.zendesk.com/blog/ai-for-employee-experience/
https://patents.google.com/patent/US20210264372A1/en - Automated human resources management and engagement system and method
TGCS Reference 4017