IT Brew survey shows heavy GenAI use for code, documentation

Docs and development are why IT pros are turning to AI and automation, according to an IT Brew survey.

Our State of Automation and AI report, released in April, found that 63% of surveyed IT practitioners said they’re actively piloting or adopting generative AI techniques for code generation and documentation.

We pulled together our notes and spoke with IT pros about what that looks like in practice.

Code generation

While Mike Toole, director of security and IT at Blumira, won’t be using AI to code sophisticated projects like an EDR or a firewall anytime soon, he’s found ways to use workflow automation platforms to construct slightly simpler assets and internal tools. For example, he built an employee-meetup tool, dubbed “Coffeebot,” by using plain-language prompts, necessary API calls, and data sources.

He’s not alone in experimenting with the technology: Gallup, in a nationally representative survey of just over 23,000 employed US adults, found that 14% of AI users have taken coding assistants for a spin.

AI coding assistants help users with autofilling standard functions and offering code suggestions. Popular use cases, as shown in StackOverflow’s annual survey of developers, included searching for answers, generating content, learning new concepts, and documenting code. (IT Brew has written about AI supporting tasks of senior developers, software leads, and veteran coders eager to keep their skills up to date.)

“If you’re a company that has any type of software development or code generation, you should be using AI to do that. It is very good at it. It will make the engineers doing it much faster,” Shanti Greene, head of data science and AI innovation for enterprise AI solutions company AnswerRocket, told us.

But there are potential downsides, too. “The AI isn’t writing the best-quality code. It’s not writing the things that are easiest to maintain,” Darren Meyer, security research advocate at appsec company Checkmarx, told us in February. “I think the gamble that people are taking is that AI is going to continue to be affordable and good enough at its job—and get better at its job—that its writing and making these poor technical decisions won’t matter, because it’ll fix itself in the future.”

Documentation

Project scope is a structured description of the services that will be provided for a given customer. In other words: documentation, hopefully as detailed as possible. 

Travis Rehl, CTO and head of product at cloud consultancy and AWS premier-tier services partner Innovative Solutions, created an AI-driven sales platform called DarcyIQ, which provides templates for scope documents based on previous deals—a user can then modify the template as needed.

New tools like Slackbot also support the pulling together of help-desk answers from a variety of disparate sources, including Google Calendar, Outlook, Salesforce, and knowledge-based articles or help-desk tickets. Those assets can quickly be transformed into documentation.

“IT leaders can ask it to review support conversations from the past few weeks to identify recurring issues, escalation triggers, or documentation gaps, and then help turn those insights into runbooks, FAQs, or action plans,” Sara Bee, senior staff software engineer at Slack, wrote to us in February.

With LLMs effectively connected to data sources, you can ask a question like, what is our acceptable use policy for Azure?, “without having to kind of go to all these different repositories that might exist,” Cameron Haight, VP analyst, business and technology insights, at market intelligence firm Gartner told us.

However, data might have inconsistent formats and need to be properly structured and organized, Haight also warned. Brian Luckey, CIO at managed services firm Integris, saw early-adopter clients connecting their Copilot assistants to SharePoint drives, which are often disorganized and therefore not the most AI-ready.

“They quickly realized that while it could bring back some data and information they found it was actually not the value they were looking for,” Luckey said.

According to our survey, IT teams are also actively piloting or adopting AI and automation for help-desk automation (39%) and cybersecurity (24%), as well as infrastructure tasks like automated patching (21%).