Artificial Intelligence systems like ChatGPT, The New Bing, Jasper, and YouChat are gaining popularity. No longer just used for a quick laugh or to impress friends, AI is rapidly maturing into functionality that businesses must seriously consider.
One of the more relevant use cases to the SaaS ecosystem is leveraging AI to support programming needs. While we caution against using AI-generated code directly in production without a peer review, there are benefits to tactically using an AI engine to support programmers. This includes increased velocity and quality of development.
At QualityClouds, we have developed a Recommendation Engine to support best practice alignment across SaaS platforms. We chose GPT-4 as it’s the most advanced programming AI available and leverages Azure architecture. Azure provides a private environment supporting GPT-4 models: data is destroyed after use. Internally, we have been working on adjusting our prompts and testing the Recordation engine to promote a few targeted results for developers:
Individual Best Practice Support
Our customers today leverage a combination of platform scans, Live Scans during development, and automation enforcement of Best Practices through defining Quality Gates. Our AI Based Recommendation Engine is for those wanting additional support for developers.
Developers who use the Recommendation Engine will have an additional button in their IDE that will send an individual instance of a best practice violation for analysis. Typically, within less than 5 seconds, our Recommendation Engine will output code with a potential pathway for remediation.
a. Triggers full-scan and LiveCheck We have designed the Recommendation Engine to target individual best practice violations found through LiveCheck and find bugs. Ability to include additional rules or train models. Propose issue-fix recommendations, copy-paste the solution, and run LiveCheck again.
Narrative Description of Code Changes and Impact
The goal is to increase the skillset of SaaS developers. It’s not enough for the Recommendation Engine to output code and best practice fixes without explanation and context for our developers. To reach that goal, we have configured our AI support to provide a narrative description of what code changes were made and the impact of those code changes on the platform. Meaning that every time a developer leverages the Recommendation Engine, they can learn and increase their skill set, ultimately upping the quality of their development moving forward.
a. Triggers: Peer Review. Propose solutions to address a write-off request or a code Peer Review.
b. Triggers: Livecheck issue. Propose missing doc issues and fix recommendations.
c. Triggers: New rule creation. Generate examples of compliant and non-compliant code. (Plus – run the batch script to generate all existing rules)
No AI Code is Auto Applied
Platform integrity and configuration drift are of the utmost importance to the SaaS ecosystem and our customers. To prevent side effects, we have configured the Recommendation Engine to never change or apply code without direct intervention from a developer. While AI may one day be capable of programming with nuance and deft, its use for businesses in the SaaS space should be tactful and cautious.
a. Triggers: Livecheck issue. Propose missing unit testing issues and fix recommendations (Salesforce today).
The Quality Clouds Recommendation Engine prototype enhances our portfolio of products and further enhances our customer’s ability to increase performance in addition to reliability, security, and scalability. Recommendation Engine will save time, empowering them to create, innovate, and scale like never before.
Visit our docs site for more information on LiveCheck for ServiceNow.