Faster than the Team, Faster than the Customer: Tool Integration, Collaboration, and Organisational Lag in AI-assisted RE
This paper investigates AI-assisted requirements engineering at XITASO, revealing that while practitioners have rapidly advanced tool integration and use cases, organizational lag and poor tool integration often limit benefits to individual productivity rather than improving team collaboration or customer readiness.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine a software company, XITASO, as a busy construction site. For years, the Product Owners (POs) have been the architects and foremen. Their job is to talk to the developers (the builders) and the customers to figure out exactly what needs to be built, then write those instructions down in a "backlog" (a to-do list).
Recently, the company started using AI tools (like super-smart, fast-talking assistants) to help the architects write these instructions faster. This paper is a report on what happened when they tried to use these new assistants in the real world, rather than just in a test lab.
Here is the story of what they found, explained simply:
1. The "Super-Speedy Architect" vs. The Rest of the Crew
The main finding is that the AI made the Product Owners incredibly fast. They could write instructions, organize lists, and even draft contracts in minutes instead of hours.
However, the team and the customers didn't speed up at the same rate.
- The Analogy: Imagine a Formula 1 driver (the PO) who suddenly gets a car that can go 300 mph. But the rest of the pit crew (the developers) and the race track (the customer's approval process) are still moving at 60 mph.
- The Result: The driver is now faster than the team and the track. The paper calls this "Faster than the Team, Faster than the Customer." The AI helps the individual PO do more work, but it doesn't automatically make the whole project finish faster because the other parts of the process can't keep up.
2. The "Magic Pen" vs. The "Human Conversation"
One of the biggest surprises was how the developers reacted to the AI-written instructions.
- The Problem: The AI wrote very clean, perfect-looking instructions. But the developers often didn't care about the perfection; they felt the instructions lacked the "soul" of a real conversation.
- The Metaphor: It's like the architect handing the builders a perfectly typed blueprint that was generated by a robot, without ever having sat down with the builders to discuss why a wall is there. The builders felt disconnected. Sometimes, they even got annoyed or suspicious of the AI-written notes, feeling like the "human touch" was missing.
- The Risk: The AI started doing the job of "talking it through." Instead of the architect and builder chatting to agree on a plan, the architect just talked to the AI, and then handed the result to the builder. This broke the team's shared understanding.
3. The "Missing Plug" Problem
The paper found that the power of the AI wasn't the main issue; the plugs were.
- The Analogy: Imagine you have a super-powerful electric drill (the AI), but your power outlet (the software tools the company uses, like Jira or Azure) is in a different room, and you don't have an extension cord.
- The Reality: When the AI was plugged directly into the project's tools, it saved massive amounts of time. But when it wasn't plugged in, the architects had to copy and paste text manually, print things out, or re-type data.
- The Lesson: The AI is only as good as its connection to the rest of the office. If the "plugs" don't fit, the AI is just a fancy toy that creates more work, not less.
4. The "Silent Renegotiation"
Because the POs were getting faster with the AI, they started changing how they worked without telling anyone.
- The Situation: The POs started doing things they used to do with the developers (like checking if a feature was technically possible) by asking the AI instead.
- The Metaphor: It's like a chef who used to ask the dishwasher if the pots were clean before cooking. Now, the chef has a robot that tells them the pots are clean, so they stop asking the dishwasher. The dishwasher (the developer) feels left out of the process, and the team dynamic shifts silently.
- The Issue: The paper warns that while the PO is moving fast, the "rules of the game" (who does what, and how customers approve things) haven't caught up yet. The customers and the team are still waiting for the old, slower way of working, creating a gap between the fast AI and the slow human systems.
Summary: What Should You Take Away?
If you are a manager or a worker thinking about using AI to write requirements or plans:
- Don't just buy the tool; check the plugs. If the AI can't talk directly to your project software, you will spend more time copying and pasting than you save.
- Don't let the AI replace the chat. If the AI writes the instructions without the team discussing them, the team might reject the work. The "conversation" is where the real value happens.
- Watch out for the "Speed Gap." If one person gets super-fast with AI, but the rest of the team and the customer stay slow, you create a bottleneck. The whole process won't get faster just because one person is faster.
The paper concludes that AI is already changing the workplace faster than our companies, teams, and customers are ready to adapt. The technology is here, but the human rules for how we work together are still playing catch-up.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.