Using AI to Provide Expert Help to IT and Developer Teams
AI ML have shown early promise for automating high volume, simple questions typical in call centers and front-line IT and developer support.
AI chat-bots for customer support are getting a great deal of attention and VC investment in industries ranging from retail to financial services to internal employee help desks. These technologies have shown early promise for automating high volume, simple questions typical in call centers and front-line IT and developer support.
An important area that has been neglected is support for engineers and developers who are driving IT innovations in a high-paced world of technology change. AI and machine learning has high potential to transform today’s community forums into highly effective expert systems ideally suited to the needs of these tech workers.
Engineers have little patience for pre-wired chat-bots and view opening a support ticket as a last resort. Instead, after a Google keyword search, they typically look for insights from trusted experts on community forums such as Stack Overflow or vendor-sponsored communities. When done well, these open Q and A platforms can create loyalty to a product, a vibrant expert community, and deflect costly support cases.
Unfortunately, answers on forums are typically provided by community members on a best-effort basis leading to unpredictable results. These experts are motivated to provide answers by both personal pride and public reputation as their answers are accepted and upvoted. But their time is also in high demand with few tools to help them. The result is significant untapped potential for community forums.
In a recent benchmark study on IT and Developer teams of over 50 industry and vendor forums, it was found that:
Only 31% of reported posts on forums are resolved;
Just 19% of posts are resolved in less than 24 hours;
A mere 14% of posts are resolved on the first response.
The same benchmark study shows this potential by identifying today’s best-performing forums which have 2x better results:
60% of reported posts on forums are resolved;
40% of posts are resolved in less than 24 hours;
33% of posts are resolved on the first response.
A key characteristic separating the best forums from the average is a highly engaged expert community. For example, in an average forum 20% of the community generates over 80% of the answers. By contrast, this rises to over 70% for the best signaling an enthusiastic and engaged expert community.
Machine learning approaches can move the average forum closer to the industry’s best and beyond, but require three things:
First, industry and product context must be determined so question intent can be well understood. Second, concisely summarized and ranked recommendations must be sourced from all available forum conversations and published content. This extracts the most relevant bits of technical documentation and conversations into readable, relevant snippets matched to the problem context. Third, appropriate tools must be provided for community experts that allow them to browse recommended answers in order to compose the best response. This reinforces the community forum as a place to hear from trusted experts, not bots.
As this pace and complexity of IT and developer innovation increases, community forums must also evolve. By combining machine learning techniques with judgment of community experts, every forum has the potential to become the best place for tech workers to get timely and trusted answers.