Why AI isn’t speeding up your product team

Introduction0:00

Hello everyone, and welcome to the inaugural episode of the Ten product podcast. I’m your host, Nikhil Kakarla, co-founder of Ten. I studied computer science at MIT before working in McKinsey’s digital and AI practice, setting up product teams, improving ways of working, and building digital solutions across Fortune 500 companies.

Today’s episode is focused on a topic that’s been on the minds of almost every engineering manager, PM and CTO out there: how can I accelerate my development cycle and ship at the speed of AI? A recent McKinsey study of engineering leaders found that despite AI tool usage, only 25% of the respondents reported meaningful AI acceleration in their organizations. Concerningly, 30% actually reported that their team’s productivity had fallen. How is that possible?

Today, we’re going to explore why AI productivity gains often don’t materialize inside organizations, and what we can do to solve it. We’re going to take you on a journey, starting with the McKinsey standard product team setup, then to the new ways of working in the AI era, and finally our new Hybrid product system. Each step along the way represents a five to ten times jump in efficiency, throughput and, at the end of the day, revenue.

The old way: the scrum council1:12

The McKinsey method that I learned is the canonical setup, and it represents the antiquated ways of working in the AI era. When I set up product orgs, each data product, imagine a customer data platform, had a scrum team assigned to it. The scrum team was made up of a council featuring business stakeholders, product managers and engineers that would meet on a biweekly sprint cadence.

At each meeting, they would intake requests from the business, prioritize the backlog, and then hand off tickets to the engineering managers with PRDs, user stories and epics. Everything was tracked in central software like Aha! and Jira, and it took months to deliver basically anything. Once engineering was done, they would bring the feature back to the council, iterate as needed, and then finally release it into production. It could take months for even the simplest features to work their way through the process, and by then the business requirements had often changed.

The benefits of this process are in the communication: constant contact between engineering, PMs and the business, and overall alignment. But the meetings and the backlogs are slow and quickly spiral out of control. Today, most enterprise product organizations are in this state.

What coding agents changed2:18

The introduction of coding agents has changed two things. The first is that non-technical teammates can now contribute to features via code, rather than English documents like PRDs, epics and stories. The second is that the actual coding and development time has reduced drastically. The new bottlenecks preventing deployment acceleration for so many are now code review, scrum team cadences and release cycles.

The new way: collaborative tools and AI review2:43

As we move forward in time, several new visions of product organizations have emerged in the AI era. There are two trends compounding to help teams accelerate their development. The first is the rise of real-time collaborative prototyping and design tools. The second is AI-based code review to flag high- and low-risk PRs and accelerate review times.

The first is pioneered by Builder.io, Figma Make and Lightsprint. These tools let designers, marketers, PMs and engineers collaborate, like a Google Doc, on new prototypes and features, and so eliminate some of the overhead of meetings, PRDs and review cycles. The second, like I talked about, is code review tools like Greptile and CodeRabbit that have accelerated review cycles by flagging errors early, and only needing human review for completed work and the highest-risk PRs.

While these technologies are incredible and have certainly helped accelerate development teams, each requires fundamentally changing your ways of working, restructuring your team, and re-educating everyone in your organization on how to use these new tools. Having done it multiple times, it’s very, very difficult.

The Hybrid way: skip the process when it’s safe3:46

At Ten, we believe there’s a better way. The entire world is focusing on improving the product team itself. Fundamentally, when a business request comes in, how do we process, design and ship it as fast as possible? The question that no one is asking is: how can we circumvent the product process entirely? Not for every feature, of course. But what if we could route 40 to 50% of tickets directly from the business request to production, without needing a meeting, and without engineering needing to review and implement them?

That is the fundamental question we are answering at Ten. Using our Hybrid platform and Rulebook technology, we allow any non-technical user, think a PM, business user or account manager, to not only create a PR with the changes they need, but automatically QA it and push it into production in the next release.

Of course, engineers can review anything that’s required, but our Rulebook technology allows Hybrid to deterministically understand which changes are safe. If it’s copy, aesthetics or a new UX flow, it can be automatically approved and routed to a designer. Changes that require engineering approval, like AI integrations, payments or auth, get properly routed to an engineer for review and added to the backlog.

Our photo and video approval processes let non-technical teammates see all the potential states they may not be considering, and improve behaviors and aesthetics on their own. Every request no longer needs to route through engineering, and that’s the fundamental acceleration that everybody is going to experience.

The future of product organizations5:20

The future of product organizations is collaborative, with AI coding tools and AI code reviews. But the future of businesses is empowering every person to make the change they need, when they need it: meet their customers where they are, and deliver solutions on time and quickly.

Imagine if your engineering team could focus their time, energy and effort on the core improvements and building the product, while letting business users and customer support managers do what they do best and make sure the product meets users’ requirements. That’s the Hybrid vision. Not a vision of collaboration that feeds into engineering review, but a vision of shared accountability across the entire product and platform team.