How Nendo Uses AI to Visualise Data (and How You Can Too)

We made this film for the screens at the launch of How Kenyans Use AI, to play on a loop as guests arrived. It has no presenter and no video file. It is 1,161 dots of light, one for every person in the survey, moving through the report’s biggest findings in about seven minutes.

We made it with an AI assistant. Here is the film, how we made it, and how you can use the same method to make your own data easier to digest.

The film plays without sound. Use the buttons to pause it, turn on the soundtrack or go full screen. On a phone, go full screen and turn it sideways. If your device is set to reduce motion, the film waits for you to press play.

What you are watching

Every dot is one person who took part. When a share of them gave the same answer, that share of the dots lights up. 52.8% of respondents use AI every day, so 613 of the 1,161 dots glow. Only 2.5% say they came to AI by accident: that is 29 dots.

The scenes move through the report in order:

• AI has already arrived. 87.8% use AI daily or several times a week, and 52.8% use it every day.
• A toolbox, not one tool. 78.6% used two or more AI tools in the past 30 days. ChatGPT and Meta AI are the most common pair (57.3%).
• Then it gets personal. Research (77.3%), education (67%) and professional work (58.6%) lead. But 72% have used AI for emotional support or a personal situation, and 46.3% use it for things they haven’t told others about.
• Trust depends on the question. 58% mostly or completely trust AI’s advice on careers. For family matters, it is 40%.
• Value and anxiety at once. 91.2% say AI has made them more effective at work or study, yet 39.2% are both more effective and worried about their jobs or income.
• Does it understand Kenya? 72% would use AI more with better Kenyan cultural and language fit. Only 34.9% rate its Kiswahili or Sheng responses very good or excellent.

How we made it

1. We started with the report, not the AI

Every number in the film comes from the report’s slides. Before any design work, we wrote the numbers into one list that the film reads from, with a single rule: nothing on screen may contradict that list. The film even carries a list of the only percentages it is allowed to show.

2. We gave every dot a meaning

One dot is one person. That rule keeps the charts honest: 2.5% can never look bigger than 29 people. When the film shows the five most-used AI tools, the five clusters are the same size on purpose, so the picture never suggests a share that the words don’t state. The map of Kenya marks only Nairobi and Nyeri, the two places named by the respondents we quote.

3. One idea per scene

The report has 66 slides. The film has 12 scenes, and each one lands a single finding, with only a few words on screen at a time. If an idea needed a paragraph, it stayed in the report.

4. We briefed the AI like a collaborator, with rules

The brief covered where the film would play (screens in a room filling up before a launch), how long it should run, how the brand looks, and the rules above. From that, the assistant built the film as a single web page: no video files, no plug-ins, and a soundtrack generated in code. The decisions about what to show, what to cut and what each number means stayed with people.

5. People checked everything

AI drafts quickly. People decide what is true. Before publishing the film here, we checked every number, quote and label against the slides again. That check matters. AI assistants tend to smooth quotes into paraphrase, and a number can slip in that your source never states. Our check caught both: two figures the report doesn’t contain, a trust figure that needed correcting, and respondent quotes that had drifted from their exact words. All of them are fixed in the version above.

6. We designed for where it would be seen

The film fits any screen, loops on its own, starts without sound and has a calmer mode for people whose devices ask for less motion. On this page, it only animates while it is on your screen.

How you can do it too

You don’t need an animation studio. You need a clear set of facts, a few rules and the patience to check the result.

1. Pick one dataset and one audience. A survey, a sales report, a programme evaluation. Decide who will watch and where: on a phone, in a meeting room, on a conference screen.
2. Write your facts down first. List every number with its source, such as the page or slide it comes from. This list is the only thing your visual is allowed to say.
3. Choose one visual rule. One dot per person, one bar per answer, one colour per group. A simple rule keeps the visual honest and easy to read.
4. Storyboard before you build. One finding per scene and a handful of words per scene. Cut anything that needs a paragraph.
5. Brief an AI assistant. Give it your fact list, your visual rule, your storyboard, your brand colours and fonts, and where the result will be shown. Ask it to propose a plan before it builds anything. The prompt below is a good start.
6. Check everything, then test it on real screens. Compare every number and quote with your source. Watch it on a phone and on the big screen. Show it to someone who hasn’t seen the data, and ask what they took away.

One tip from our own process: ask the assistant to keep all the numbers in one place in its code, and to tell you which numbers appear in each scene. It makes the final check much faster.

See the full findings

The film is the short version. The full findings, a five-minute, game-style story and the free 66-slide report are on nendo.co.ke/ai. Download the report, and if your team wants to turn access to AI into real capability, see how Nendo training can help.