How to find the business process that’s right for AI

CIOs embracing AI might have the same question as somebody trying to clean up after a party: Where do we start?

Eyal Bukchin, CTO and co-founder of software developer tool builder MetalBear, has gone from “wait and see” to “let’s get going” when it comes to AI. He recently encouraged his developers to experiment with tech like Claude Code, ChatGPT, and Gemini.

In a recent blog post, Bukchin shared how the tech helped his team, including:

• Creating scripts for software chores, like setting up a test environment
• Getting familiar with unfamiliar code
• Exploring ways to implement a feature

Now Bukchin is making more budget available and prompting his team to share their use cases. “In the last two or three months, I’ve been really feeling the pressure to maybe be more deliberate about [AI usage],” he told us.

With AI, he added, “you get velocity, you get quality at speed, and you deliver more features and value to your customers.”

Bukchin and other IT pros spoke with IT Brew about where to start with AI, and why choosing the right process is especially important for organizations that want to move fast.

Start your AIngines. Bubble or not, companies are eager to invest in AI, according to a recently published survey from Accenture. Dual global surveys of C-suite executives and other workers found that:

• Almost half (46%) of leaders would continue to invest in AI “even in the event of a market correction.”
• AI and digital-tool investments grew as a strategic priority for 71% of C-suite respondents—up from 53% in the summer of 2025.

A global survey from McKinsey found that 30% of survey organizations are “piloting” or implementing their first business use of AI, and almost a third (32%) are “experimenting.”

Getting down to business. Boris Kolev, global head of technology at JA Worldwide, a nonprofit focused on financial literacy and entrepreneurship for young people, felt the pressure to move fast with AI because he assumed the group’s students expected their education to feature the latest technologies. The company’s employees were already using it, and Kolev wanted to make sure they weren’t sharing sensitive data with free AI tools, which aren’t locked down in the same ways as enterprise versions.

Kolev and his IT team looked at each business unit and identified “bottlenecks” where automation could save the most time.

JA Worldwide now has an AI agent proof of concept to automatically summarize and consolidate global annual reports for a group of about 10 employees on the finance team; these reports are often in different formats and languages, Kolev said. Another agent scans for open grants online and drafts tasks for the fundraising team.

And there are rumblings for even more AI features to address future needs. “As a nonprofit, we have limited resources, and our IT team is quite small, so there is pressure about, ‘Okay, we want to be the next one. Let’s automate marketing tasks. Let automate brand tasks,’” Kolev said.

AI x acceleration. Constellation Research recently explored the importance of decision velocity, defined in a November 2025 report as “how fast and effectively an organization can sense, decide, act, and learn to lift measurable outcomes quickly and accurately.” The study concluded that “boards and CFOs expect more than POCs [proofs of concept]. They expect proofs of value (POVs) with measurable improvements in the speed, accuracy, and effectiveness of the decisions that run the business.”

Michael Ni, VP and principal analyst at Constellation Research and the report’s author, provided recommendations for companies looking to move beyond the POC to POV:

• Start with a small process with measurable impact
• Look for high-volume, repetitive, and time-sensitive processes
• Add automation to existing workflows
• Deploy pilots that build internal knowledge and buy-in

If you have a lot of proofs of concept these days, it might be time to get “surgical,” according to Muqsit Ashraf, group chief executive for strategy at Accenture, and find projects with the most meaningful transformation possibilities.

“What are some of the most critical decisions that your company makes, and how do you increase the effectiveness and/or efficiency of those critical decisions?” Ashraf said.

Standards and practices. After the past few years of AI excitement, Bukchin realized his group needed to “standardize” its use of the technology. For now, he wants to give his team the mandate and budget to choose their tools and demonstrate improvement: “Almost everything that an engineer does can benefit to some extent from AI. You just want to sort of start where it’s easiest and then move from there.”