How do you write up to 65 chapters from 1000+ data points, with many people, and without losing track?

Role: Sole Product Designer, with 1 PM, 3 engineers & CTO.

Context

Mavue is an early-stage B2B SaaS company that helps small and mid sized manufacturers in Germany create their annual sustainability report, from collecting ESG data and spotting risks to setting goals and writing the report itself. Throughout the year, they collect data on Environment, Social and Compliance topics and use it to identify risks, challenges and opportunities. Once a year, much like a financial report, they turn these insights and their goals into one report.

Problem

Writing long reports is used to mean jumping between guidance, data tables and separate drafts. It is also a team effort: chapters have different owners, because only HR can write about HR topics, and many companies have no central sustainability manager to cover environment, social and compliance. People working on the same report could easily overwrite each other.

Scope

• Up to 65 chapters to report on

• 4 areas: general, environment, social, governance

• 1000+ data points behind the report

  • 1/3 = numerical data (energy usage, gender equality)

  • 2/3 = written data (policies, risks, action plans)

Constraints

• Early stage, few engineers: The product was built mainly to prove product market fit.

• Build fast: We used existing components from a component library, plus a few new elements kept simple so their function is clear.

• Mixed data: The collected data points are partly quantitative and partly qualitative. Showing both in one interface without overwhelming writers was a real challenge.

• A lot to cover: The report draws on 1000+ data points, and the workspace had to support the whole writing cycle: sharing chapters, tracking status, drafting, commenting, reviewing each other’s drafts and a first AI draft from existing data.

Solution & Key Decision

The chapter became one workspace: guidance, data and an example sit right next to the editor, with comments and history alongside. To avoid overwhelming writers, it shows only the data that is relevant to the chapter, right next to the text. Given the constraints, I chose an edit lock per chapter over live co editing, which avoids overwritten text and keeps commenting open for everyone.

The big question was speed versus ownership. The fastest option is AI that writes straight into the chapter, but writers want to stay in charge. So they choose the sources, tone and length, preview the draft, and insert it only when they are happy, with Undo always available.

Process

1. Understand: I gathered technical limitations and minimal requirements from the engineers, interviewed users about their needs and challenges and looked at how other tools handle this.

2. Ideate: I initiated creative workshops with the team to discuss the gathered research and generate ideas.

3. Prototype: Based on the workshops, I drafted 2 interactive prototype solutions in Figma. Option A inserted the AI draft straight into the chapter with Undo. Option B showed it as a preview first and inserted it only after confirmation.

4. Test: I tested them with 10 users of two kinds: sustainability managers, who know the topic well, and colleagues from other departments, like HR, who are less used to sustainability topics and to the platform. Option B won, because testers felt more in control, even though it costs one extra click.

5. Iterate & hand off: I iterated on the feedback, created the final high fidelity prototype and handed it over to the tech team.

Outcome

The user group for testings was small and only consisting of 10 users, so I don't have hard numbers, but the feedback was very positive:

• The AI draft was a huge help to get started, and it saved a lot of time, especially for sustainability managers who own the whole report.

• Sharing chapters, seeing progress and reviewing each others work made the process transparent.

• Having everything in one place was very helpful.

• The balance between AI draft and the manual control of the writer felt well.

"Having the guidance, the data and the draft in one place makes the report feel very manageable. The AI draft gets me started, and I still decide what goes in.” (Sustainability Manager)

"I'm not a sustainability expert, but I could see which chapters were mine and where everyone else stood. The AI draft was a huge help.” (HR Department Head)

Learnings

• Control makes AI feel helpful: Testers valued the balance between the draft and their own say, more than raw speed.

• The AI draft matters most for whoever owns the whole report. Sustainability managers write or oversee the entire report and have to turn the collected data into text. They are not experts in every specific field, so for them the AI draft was especially helpful: it saves a lot of time and gets them to a finished report faster.

What I would do differently today

AI tools were still very early when I worked on this. Today I would:

• Synthesize research faster: Use Claude Projects to collect interview notes and feedback and synthesize themes, while I decide what matters.

• Explore and test earlier: Use Claude Design and Figma Make to try several directions quickly with still being consistent with the existing design and put more variations of clickable prototypes in front of users even sooner.

• Prototype the AI itself: Use Claude Code to build a working Draft with AI flow with real outputs, so sources, tone and the preview step are tested with real drafts.

Note: The screens shown were recreated, anonymized and adapted: data, copy and branding are invented to protect confidential product details.