Five Key Takeaways From GLOWLive 2025: The AI Coding Revolution

That’s a wrap

Jellyfish GLOWLive 2025 was one to remember. More than 500 participants tuned in to our 90-minute virtual event exploring how AI is transforming software engineering and what it means for the future of our industry.

Kicking off the event, Jellyfish CEO Andrew Lau reminded us that, despite AI’s meteoric growth, it’s still early innings. Later, in his fireside chat with FanDuel CTO, Andrew Sheh, the two discussed how to actually adopt AI tools and best measure their impact. They also covered how Sheh supports a team of more than 2,000 engineers distributed around the world.

You can watch a full replay of Andrew’s keynote and his discussion with Andrew Sheh here.

Next, Adam Ferrari, Jellyfish advisor, and Ryan Kuchova, SVP & Field CTO, dove into the Software Engineering Intelligence Maturity Model. Ferrari and Kuchova discussed how to better align engineering efforts to business goals and how to benchmark those efforts to ensure peak performance. Check out the recording to learn more.

Lastly, in a panel discussion, Jellyfish Head of Product Krishna Kannan spoke with three engineering executives from leading technology companies about measuring AI coding tool adoption and impact.

You can watch Krishna’s panel here or keep scrolling for five key takeaways from panelists Kevin Isacks, SVP Engineering at Avid; Andrew Weiss, Field CTO, Public Sector at GitHub; and Daniel Koch, GitHub Practice Lead at IBM.

1. AI helps engineers tackle new challenges

AI allows engineers and developers to step outside their comfort zone. With AI coding tools, engineers can explore new ways to build creative solutions and bring their ideas to life.

“This is one of those once-in-a-decade shifts in technology,” said Andrew Weiss, field CTO at GitHub. “Generative AI and its use in coding is changing the way we approach software development by empowering people to learn more and tackle new challenges. We’re seeing that with the advent of tools like GitHub Copilot and the use of AI throughout the software development lifecycle.”

In his role as GitHub practice lead at IBM, Daniel Koch has seen teams that were previously siloed suddenly have to think about the entire infrastructure of the organization. GitHub Copilot helps learners get the guidance they need on infrastructure-as-code, without having to pull other colleagues away from their work.

“Copilot helps people stay within the bounds of sanity when they’re learning Terraform or any other infrastructure as code language,” explained Koch. “It’s awesome that AI can help teams adapt to new roles – I love seeing that.”

2. Engineers are happier and more productive with AI

Avid, a technology company serving media creators, initially made GitHub Copilot available to around 20 senior and junior coders. They wanted to make sure the tool delivered real value before rolling it out to the wider organization. The small pilot group quickly evangelized usage to their colleagues – now, over half the organization is using Copilot each day.

“Everybody you talk to says how much they love it,” said Kevin Isacks, SVP Engineering at Avid. “Truly, people are finding it useful, and once they’ve used it, they’re hooked.”

DevEx surveys suggest developers who use AI coding tools are happier than those who do not. They have time to focus on the work they enjoy, reach a flow state more often, and, in turn, are more productive. A poll launched at GLOWLive showed that 82% of engineering leaders see AI coding tools impacting their teams positively in the long term.

Isacks combines this subjective evidence with objective data for a complete view of engineering productivity. “That’s where Jellyfish metrics have been very valuable,” shared Isacks. “It compares the population that’s using Copilot against the population that’s not, and I can see great productivity gains.”

3. There are still barriers to adoption

At Avid, not having the right tech stack is the main barrier to AI coding tool adoption. However, as Isacks points out, that’s an easy fix. Trust issues are harder to break down – tools that change how people work are generally met with skepticism. For Isacks, starting small was key to building trust. “That little tiger team evangelized usage and cheered on the effort,” he explained. “It made a huge difference to have people saying, ‘this is something that we want to use.'”

GitHub Copilot prioritizes model choice – you can select whichever model you want based on the desired outcome. Weiss recognized that this means engineers might not know which model is best for their situation, and, because models are non-deterministic, they could give different answers to the same question.

Weiss added that some engineers are yet to adopt Copilot because it doesn’t meet all the requirements of their use case. “Maybe it doesn’t have information about the runtime state of their application stack – although that’s now changing with the model context protocol,” said Weiss. “But, as more and more teams adopt it and the models get even better, we’re going to see those barriers fall.”

4. AI can improve overall code quality

Everyone agrees on the importance of test-driven development and code quality. But when engineers are working on a tight timeline, testing can often fall by the wayside. AI coding tools take care of these repetitive yet essential tasks, resulting in better test coverage and improved code quality.

AI coding tools offer an opportunity for companies to emphasize foundational practices. Many organizations fail to enforce code review, despite it being a critical part of the software development process. Now, with the introduction of Copilot Code Review, it’s getting the attention it deserves.

“Copilot Code Review is kind of your first gate,” explained Isacks. “Then, you still do need to go through a human review.” By making it impossible to skip this vital step, AI coding tools help prevent bugs slipping into production.

5. Leaders need to keep up with changing technology

Another GLOWLive poll showed that 60% of engineering leaders are using AI coding tools, with 33% saying they use multiple. AI in coding is evolving at breakneck speed and shows no signs of slowing down. “Every month there’s something new that is just truly, truly changing the game,” said Isacks. “It doesn’t matter what it is, you can throw it at generative AI and it’s able to help solve the problem and make everybody’s life better, which has been fantastic.”

While the panel agreed that progress is a good thing, it can be hard to keep up with the pace of change. Rounding off the discussion, Isacks, Koch, and Weiss offered their advice to R&D leaders.

Weiss suggests leaders and decision makers experiment with AI tools, especially if they’re from non-coding backgrounds. “It’s not just for technology enthusiasts, it’s for everyone,” said Weiss. “Understanding how these things work at a high level and using them day to day is probably the best way to get acclimated. You can then have valuable conversations with folks within your organizations, understand how they’re using it and where they see value.”

For more insights on how leading enterprises are using AI coding tools, watch the full replay of Krishna’s panel discussion here.