Chatbots accessible with AI for people with disabilities

In the age of artificial intelligence, chatbots can be a powerful tool to improve digital inclusion. But to achieve this, it is essential to design taking into account the specific needs of people with disabilities visual, auditory, cognitive, motor, or speech.

Here we will share some recommendations based on our experience— to develop chatbots truly accessible, as well as useful tools and a reflection on the current challenges.

1. Design conversational inclusive
Using clear language, straightforward, and simple.
Avoid jargon or ambiguous sentences.
It offers multiple ways of interaction (text, voice, buttons).

2. Compatibility with screen readers
Make sure that the chatbot is accessible with technologies such as NVDA, JAWS or VoiceOver.
Correctly tag the elements and prevents contents may change without notice.

3. Accessibility visual
Uses good contrast between text and background.
It allows customization of font size.
Avoid relying solely on emojis or visual elements.

4. Options multimodal
Provides input and output voice for people with impaired vision or mobility.
Use subtitles and transcripts if the chatbot has a voice.

5. Reduction of cognitive load
Presents clear steps and segmented.
Use buttons or answers guided when possible.
Provides context-sensitive help.

6. Validation with real users
Involving people with disabilities in the design, and testing.
Applies principles of universal design and co-creation.


During the development of prototypes accessible, we have made tests with platforms such as Voiceflow, especially focusing on the creation of chatbots controlled by voice.

The result? To this day, has not been satisfactory.
The assistants by current voice still showing important limitations in the natural understanding of the language, in response to different accents or tones of voice, and the seamless integration with assistive technologies.

This type of technology, although promising, it is not yet at the height of what they need to people with disabilities, especially those that depend on the voice as the only channel of interaction.

We need this technology advance, not only in technical abilities, but also on criteria for accessible design, so that you can to fulfill its promise of improving the lives of all, without leaving anyone behind.


Google Dialogflow + Vertex AI
Powerful AI conversational.
Compatible with voice and text, adaptable to accessible web sites.
You can train with inclusive language.

Microsoft Azure Bot Framework
Accessible templates.
Integration with Azure Speech and other tools cognitive.

BotPress (open-source)
Total Control over design and accessibility.
Customizable to meet the standards of WCAG.

Tidio / Landbot
Platforms-code for quick deployment.
Require manual adjustments to ensure accessibility.

Accessibility resources
W3C Web Accessibility Initiative: https://www.w3.org/WAI/
Guía WCAG 2.1: https://www.w3.org/TR/WCAG21/
Inclusive Design Toolkit: https://inclusivedesigntoolkit.com

Digital accessibility is not an addition: it is a right. The chatbots can be a solution inclusive, but only if the design from start thinking of all the people.

Our experience gives us a clear conclusion: much remains to be done in the chatbots based on voice, and we need to encourage the development of technologies that really serve to those who can benefit most from them.

Do you have experiences or ongoing projects related to accessibility and IA? We'd love to hear them and share learnings! InteligenciaArtificialparaONG.org

Related articles