Nairobi, Saturday, November 7: teams and the problem statements
This page explains how teams work, which problems they choose from, and what they share at the end. For where and when, see the schedule.
Your team
There are six teams of seven. Each has five makers, chosen for a balance of skills, gender and age, and two co-designers with lived experience of disability, allocated by United Disabled Persons of Kenya (UDPK). Each team also gets a mentor who knows the toolkit. The co-designers define the problem and share every decision. On the day, a few more co-designers with other access needs are on hand, so teams can test with people beyond their own team.
What you build
At the information session on October 31, each team picked one of the problem statements below. On the day, it builds a working solution with the toolkit and a coding agent, tests it with its co-designers, and logs what they find.
Problem statements
Senses Hub and UDPK chose these problem statements together, after talking with organizations of people with disabilities across Kenya. They show the gaps people meet in real life, both the barriers and the things that help, when they use websites, reach public and private services, and go about their daily work. Teams first heard them at the information session on October 31.
Three problems came up strongly in every community:
- The services that matter most are government services as they move online, mobile money and payments, and opting out of services and subscriptions.
- People move constantly between English, Swahili and Sheng, which breaks screen readers, captions and plain-language tools built for one language.
- Most people use low-end Android phones with patchy data, often shared with family, so anything that runs in a browser, works offline where it can and respects privacy on a shared device fits best.
Each statement was checked against the toolkit’s three building blocks and its two themes, which are adapting browsers and mobile apps, and improving wearables that already exist.
Blind and low vision
Blind and low-vision people in Kenya are an organized community, and active online. Most use a smartphone with TalkBack every day, and many use NVDA on a laptop. Their tools work until a page has unlabeled buttons, text inside images, a CAPTCHA or a timer; then the phone goes to someone else, usually at the payment step. Pages that mix languages are hard to read with the assistive technology people have, and apps that describe scenes and text say too much, in American English, and miss handwriting and Swahili. With mobile money everywhere, asking for help with every transaction makes financial independence hard.
- Mixed-language pages. A screen reader set for one language mispronounces mixed English, Swahili and Sheng, or skips parts of it, so people can’t read notices, banking information and much else on their own. How might we adapt a Kenyan web page so that a blind or low-vision user can listen to mixed English, Swahili and Sheng content in their chosen way, without having to hand the phone to another person?
- Government forms. When fields, one-time codes, payment prompts or receipts aren’t accessible, a blind user may not be able to finish a multi-step government form alone, and asking for help exposes private information such as payment details or a PIN. How might we help a blind user complete a realistic government-service form through an accessible, step-by-step web flow, while ensuring that submission and payment happen only after the user explicitly confirms?
Deaf and hard of hearing
Deaf Kenyans use Kenyan Sign Language (KSL), which is a language of its own rather than signed English, and many read English as a second language. Most hard of hearing people, including most older adults with hearing loss, manage without hearing aids: fewer than one in ten people who need one has one. The channels that work are WhatsApp and SMS. Customer service means a phone call, with nothing captioned and nothing in KSL, and where automatic captions exist they’re in English only and never show when they’re guessing. In medical, legal and financial conversations, human interpreters are still essential.
- Captions for Kenyan video. Kenyan video and audio online, whether a presidential address or a university lecture, has no captions or English ones only. Captions break down on Swahili, names and accents, and never show when they’re unsure, so a Deaf viewer can’t tell a good caption from a guess. How might we give Deaf and hard of hearing people captions for Kenyan video that switch between English and Swahili and show honestly when the caption is uncertain, with a route to human KSL for the sections that matter?
- Voice-only helplines. Banks, hospitals, utilities and government helplines are voice-only, and more of them now use AI voice agents. A Deaf person has to find a hearing relative or go in person, and loses privacy either way. How might we let a Deaf person handle a routine bank, hospital or utility query in writing, without a hearing relative, and with every answer shown before it is accepted?
- Phone alerts that matter. Default phone alerts don’t make important events clear or easy to tell apart for a person who is blind, Deaf or hard of hearing. Low-priority alerts bury the ones that matter: money leaving an account, a message from a named person, something urgent. How might we let a person define a small set of important phone events and receive a clear alert in their chosen format, while keeping all rules and message content private and under their control?
Physical and motor disabilities
This group includes people with cerebral palsy, spinal injury, the effects of a stroke, arthritis, tremor or limb loss, and wheelchair users, for whom where the screen sits matters as much as what’s on it. They meet small targets, drag gestures and timed inputs, identity checks that need a steady hand, voice control that mishears Kenyan English and ignores Swahili, and screens mounted for adults who are standing. One mis-tap must never send money or submit a form, and nothing should time out while someone is still working.
- Reaching the screen. A wheelchair user, or someone using a phone in bed, may find a touchscreen hard to reach or use, because of small targets, poor placement and the need for a precise touch. How might we adapt a browser-based service for a wheelchair user so that its controls, text and input method match what the person can comfortably reach and operate?
- Browser games. Browser games assume players can see the screen, use a mouse, react quickly and play without pausing. That shuts blind players, and people with motor or other access needs, out of games with friends and family. How might we adapt one open-source browser game so people with different access needs can play together using the controls, pace and feedback that work for them?
- A board game for everyone. A blind player, or someone who can’t easily move pieces, is left out of a shared board game because they can’t see the board, keep track of positions or move pieces on their own. How might we make a simple physical snakes and ladders game playable by blind and sighted players, and by a player who cannot move a piece, while keeping each move visible, understandable and under the player’s control?
Cognitive, learning and psychosocial disabilities
This group includes people with intellectual disabilities, dyslexia, ADHD, autism or an acquired brain injury, people living with anxiety or other psychosocial disabilities, and older adults new to smartphones. What gets in the way is language and overload rather than perception: dense legal English, dozens of steps, deadlines hidden in notifications, and systems that assume a guardian will do it. Journeys, clinic visits and interviews can’t be rehearsed. People draw clear lines: nothing submitted or signed on their behalf, nothing shared with a helper without asking, nothing hidden when text is simplified, and nobody deciding for them when they’re unwell.
- Official forms. Filing a simple tax return, registering for a service or accepting a loan means legal English, error codes and penalties for a wrong click, and most of it now happens online. People sign what they can’t read, or hand it to someone else. How might we make it possible for someone with a learning or intellectual disability to complete a simple official form on their own, understand what they are about to submit, and know what was left out when it was simplified?
- Rehearsing a journey. Some people find the trip to a stop, a hospital reception or another public service confusing, stressful or physically risky. Learning the steps ahead of time would help, at their own pace and with control over how much it asks of their senses. How might we adapt a short virtual rehearsal of a Nairobi journey or service visit to a person’s access preferences, so they can practice at their own pace and exit whenever they choose?
How projects are judged
Projects are judged on four things:
- How well it fits the chosen problem statement and the person it serves (30 percent).
- A working solution, demonstrated by a co-designer (30 percent).
- Its use of the toolkit or AI agents, and what it contributes back to the toolkit (20 percent).
- Clear documentation, and the potential to continue (20 percent).
Senses Hub may still adjust these criteria until the event starts.