At Hg's 2024 Digital Summit in Amsterdam, brought together over 120 technical leaders to explore emerging use cases and the latest advances in AI. During the summit, Hg's Key Vaidya caught up with Dave Mangot, Founder of DevOps consultancy Mangoteque and author of 'DevOps Patterns for Private Equity'.

Generative AI tools have the potential to unlock value within software companies, enabling engineering teams to do more with the same, drive automation and increase the frequency of software deployments and releases.

Yet in order to stand out from the crowd, much will depend on understanding how best to take advantage of, and leverage Gen AI to maximum effect. Especially with so much hype surrounding it.

Be aligned

Like every other technology shift, there’s no magic wand one can wave to effect immediate change with Gen AI. What is important is to have the foundational elements in place. And while there are myriad opportunities to leverage various tools, there is ultimately one single goal: solve customer problems.

Key Vaidya, Portfolio CTO at Hg, advises that nobody is going to pay you because you’ve achieved a better developer product experience or a better CI/CD pipeline. That is not where customer value is. The more software companies can focus on their core competencies, it will help open up innovation cycles, help driver up customer NPS scores…and enhance their point of differentiation.

Doing so requires strong team alignment. If there’s a Dev leader and an Ops leader and those two leaders aren’t aligned, you end up with a classic case where the Ops team wants stability, and the Dev team wants speed and features. Research shows that developers using GitHub CoPilot code up to 55% faster “but in practice we’re not seeing anywhere near that much at Hg,” Vaidya remarked in conversation with Dave Mangot, Founder & Author, DevOps Patterns for Private Equity, at the Hg Digital Summit 2024.

Writing code is not the hardest part of software development. It is managing teams, getting them to work together. This is more of an architectural problem, according to Mangot.

By overlooking this point, companies who introduce Gen AI tools might ask, ‘Why am I not getting all these incredible results?’ Well, maybe they are, but in specific areas as opposed to across the value chain.

As such, companies should focus on alignment and for DevOps teams to realize that they are part of the same uniform system. Operating in siloes won’t work. It is not a Dev product or an Ops product that the customer is buying. All they want is “the product”.

Be intentional

When looking to make effective use of Gen AI, the aim should always be to gain an advantage on one’s competitors rather than seek to copy them. Companies need to consider why they want to introduce new tools; what is the intention?

Intentionality, in this instance, could apply to delivery pipelines, product development, testing, in order to shorten and amplify feedback loops. Look for bottlenecks (i.e. wait times, hand-offs) and areas of high leverage.

If companies go from releasing software four times a year to five times a year, there is no point, for example, using DORA metrics to measure this. Of far greater value would be to use DORA metrics in response to the question, ‘Why is it important to shorten and amplify feedback loops, and where are the bottlenecks that need to be alleviated?’

As such, this will require some companies to shift more from a project-based mentality to a product mentality.

Just in time…to scale growth with Gen AI

Having a clear roadmap on where best to leverage Gen AI to speed up the deployment and release cycle offers the potential to drive growth at scale. Getting software in the hands of customers…this is where the value lies.

Toyota solved its inventory problem with the development of ‘just in time’ manufacturing. This allowed the firm to avoid spending considerable capital on inventory that would sit in warehouses for days and weeks at a time.

The software industry could face a similar predicament, with faster coding creating the prospect of increased inventory in Git repositories. Thankfully, Gen AI has the potential to overcome this issue.

As Mangot explained: “Generative AI can help grease some of those skids to get software out and manage bottlenecks on things in release, in testing.”

“Gen AI is an engine that will force people to get much better at releasing software.”

Those who can use technology to tackle some of these problems are going to see the biggest benefits, on a go forward basis.

Fast release cycles tend to be perceived as the hallmark of modern, stable companies; i.e. the Netflix’s and Google’s of the world.

As Gen AI adoption grows, testing is one area where it could be used to speed up the process of software deployment and release, although Vaidya notes there is evidence of misalignment. Some companies, for example, are choosing to use Gen AI tools to perform and run the tests. A better option would be to harness the technology to actually write the tests themselves.

Dark launching to drive optimization

Another consideration to help drive growth is to use Gen AI tools for dark launching. This involves releasing software features to a subset of users, through the use of feature flags, and monitoring user response. Doing so creates a safe environment for developers in which to make improvements. Facebook, for example, dark launched its Messenger app to avoid risking a potential global blow-up.

Gen AI offers the potential for DevOps teams to run many more of these behind the scene experiments, to understand how new features could best be optimized.

This could be a big differentiator for high performing software companies as they seek to further increase their release cycle; particularly those at level 4 SaaS maturity, where they are able to decouple the processes of deployment and release (and help sales & marketing teams excel in their respective roles).

Indeed, many companies end up linking the two processes, which forces them to move at the slowest component speed.

Less stress. Greater productivity

As software companies learn to embrace Gen AI, taking small steps could lead to efficiency gains. Helping with documentation and coding, or developer onboarding, is likely to reduce stress and make DevOps staff feel happier in their roles.

Kaizen, the Japanese business philosophy of continuous improvement, serves as a good example of what could be possible with the future application of Gen AI. The purpose of automation, after all, is to make people even better at their jobs.

Documentation is something every software company struggles with this. Thankfully, that is something large language models are really good at. With respect to developer onboarding, if firms can achieve this within two months instead of nine months, it could free up a lot of productivity time to spend with the developer; time that could contribute real value to the business.

Furthermore, on the coding side, Gen AI doesn’t have any fears about going in to an old code base to understand what it is doing (and where improvements could be made).

The Gen AI transition is now well underway, as the adoption rate continues to gather pace.

Companies born on the cloud, who are already good at deploying and releasing software, are likely going to have a big advantage. The fear is that they just keep on getting better and accelerate down the racetrack, leaving their competitors behind. All of which makes Gen AI adoption a fascinating trend to keep a close eye on.

This is not about the big versus the small anymore. It is about the fast versus the slow.