Conversational tools have become one of the most widespread use cases in AI, with chatbots promising instant access to machine intelligence.
Designer Paul Faivret, however, saw the trend in a different light. For the French product and UX designer, chat tools could take care of small, one-off work tasks, but he questioned how well they could properly serve those in professional settings.
Faivret has explored that question through his work on V7 Go, a platform built for legal, insurance, and finance teams. Instead of a chat window, he designed a structured system, a table that turned outputs into organized, reviewable data. Now he’s bringing this mindset to software startup Interfere, where he focuses on building products that people can understand in an intuitive manner.
Realizing the Limits Behind Chatbots
The wave of automation that followed ChatGPT’s release flooded workplaces with chatbots. Every new product promised a conversational shortcut: an assistant that summarized, generated, or advised at a keystroke.
At V7, where Paul Faivret served as lead designer, he began feeling like this new technology could come with certain limits. The company’s new product, V7 Go, aimed to help lawyers, insurers, and financial analysts process thousands of documents at once.
The team initially experimented with conversational prompts, imagining professionals typing questions in their day-to-day. But the more Faivret studied his users, the less he felt it was a proper fit. His users were expected to deal with large chunks of data, often being sent in bulk, and, as such, needed organized templates they could organize it into, meaning that constantly engaging in conversation with a model might slow them down instead of empowering them.
“Chat is great for personal use cases or for where you’re exploring,” he said, “but not when you’re executing. Professionals don’t want to guess what a system means — they want to know what it’s doing.”
Switching to a Table-Based Interface
The idea for V7 Go’s interface was based on a classic tool in business software: the spreadsheet. When Faivret began sketching solutions that could process large and lengthy documents, he turned to the structure people already trusted. A table, built on the standard rows and columns model, could go through thousands of files, display progress, and let users set their own logic to get a big-picture view of their work.
The interface that emerged let professionals upload entire document sets, create custom columns such as “Is this clause valid?” or “Does this policy meet compliance?,” and run checks across them all. Results appeared in organized cells, editable and comparable, which users could edit and organize in whatever ways suited them most.
What distinguished the system was that it presented AI as a mechanism for inspection. Every action was visible, and every output had context. The design didn’t try to humanize the model; it framed it as a precise, scalable instrument. “Lawyers have hundreds or thousands of documents,” he explains. “With the table UI, they can drag all their documents in, add columns with requests specific to their workflows, and do due diligence document processing at scale.”
His Holistic View on Design
Faivret’s approach to design is anchored in precision. A self-taught designer with a deep interest in typography, he sees balance and spacing as essential to establishing an internal structure, the quiet framework that determines whether a product feels trustworthy. He considers every line, margin, and type weight contributes to how clearly information reads, and he treats those decisions as a core part of the product’s logic.
He describes good design as “invisible work”, the kind users rarely notice because it feels instantly intuitive to them. That’s why he focuses on making it so that every choice serves how people already think and move through a screen. “It’s about making things that feel simple even when they’re complex underneath,” he said.
That philosophy shapes his definition of quality. A product earns its polish not through visual effects but through consistency, rhythm, and restraint. Faivret believes usability grows out of these underlying fundamentals: typography that reads cleanly, interfaces that hold their structure, and details that stay consistent under pressure.
Structure as the Future of Human–AI Collaboration
At Interfere, Faivret’s focus is on turning theory into infrastructure. As the company’s founding designer, he’s been developing the interface for a platform that seeks to find and solve bugs automatically. The end goal is to build a system that watches its own behavior, diagnoses issues in real-time, and improves through use — a form of automation that strengthens precision as time goes by.
In this new role, Faivret is extending the same logic that guided his work on V7 Go. Where the table interface made complexity visible, Interfere’s interface aims to make maintenance invisible. The principle is the same: technology should work clearly, consistently, and with minimal intervention.
He views this direction as a natural step in AI’s evolution. The next generation of tools, he believes, will move away from imitation and toward interpretation — systems that, beyond talking back, can also pay attention and pick up underlying context.
That shift demands design that can make sure models (and their interfaces) have an internal logic users can understand, helping them feel they can depend on its intelligence. In his own words, “Software should get better the longer it’s used.”
A Precise Approach to Design
Through his work in professional software, Paul Faivret’s focus on structured design can serve as an example of the role designers play in helping the tech industry better ship tools that communicate clearly. To him, the next phase for AI will belong to systems that explain their reasoning as naturally as they perform their tasks.
