For good
AI accessible to all is a claim you have to spend money to make.
§ 01 · The test
"AI for everyone" is the most common sentence in Indian technology and one of the cheapest to say.
Our test is straightforward. Can a person who does not read, does not speak English, cannot see the screen, or has never used a smartphone get value from what we built? If not, it is not accessible, whatever the marketing says.
We fail that test regularly.
We publish what we are doing about it, because a commitment with no accountability attached is just advertising.
§ 02 · Where the work goes
Where the work goes
Disabled users.
Voice-first interfaces in Indian languages for people who cannot use a screen. Tactile and audio-described experiences, already built into our museum installations. Screen-reader-native design, tested with people who use screen readers rather than against an automated score. Indian Sign Language data collection, which is severely under-resourced next to ASL.
Rural users.
Vernacular voice access to services that currently assume literacy, a smartphone and English. Offline and low-bandwidth operation, because a system that needs 4G is unavailable exactly where availability matters most. Agricultural advisory in the farmer's own language and crop context.
Older users.
Interfaces that assume no prior digital experience. Voice-first, patient with hesitation, and tolerant of restarted sentences and long pauses, which most voice systems are not. Fraud resistance built in, since older users are targeted hardest and warned least.
§ 03 · The commitments
The commitments
Five per cent of Marxen engineering capacity goes to accessibility and public-interest work, scheduled like client work with owners and dates.
Free platform licences for government schools, public hospitals under 100 beds, registered disability organisations and rural skilling centres. No seat caps, no features held back, no conversion pressure at month twelve.
The Suite free for registered non-profits and disability organisations, deployed and maintained by us.
Open datasets in the categories nobody funds: Indian Sign Language, speech from older speakers, and speech from people with atypical articulation. Every commercial ASR system in India performs worse for these speakers because the data does not exist.
Contributor terms apply to the free work too. Paid, on schedule, and told what the data is for, in their own language.
We report against all of this annually.
Built in India. For the people actually using it.
Tell us what you are trying to do. Bring the use case, the constraints and the users. We will tell you honestly whether Marxen is the right call, including when the answer is no.