AI Tools for Disability in 2026: The Annual Update on What Actually Works

AI accessibility moves fast enough that one review will not hold. This is our first annual update on AI tools for disability use. The goal is honest evaluation of what is actually useful, what is overhyped, and what is starting to deliver after the 2024-2026 generative AI wave. This will be repeated annually.

The conversation about AI and disability has changed shape in the past two years. The 2023 narrative was that AI was about to transform disability tech overnight. The 2024 reality was more uneven. The 2025 reality was that some tools delivered substantially, others stalled, and a few new categories emerged that nobody had predicted. The 2026 reality is that AI for disability is a real and useful category, the early hype has settled, and you can tell the tools that genuinely help from the ones that do not.

This is our honest assessment, by disability category and by tool, with the plain-spoken evaluation the marketing rarely gives you.

A note on prices. Every figure in this article is in US dollars, taken from the maker’s own site and checked in July 2026. Almost every tool here is priced in USD, so a Canadian reader should expect to pay the exchange rate on top, and card issuers usually add a foreign transaction fee. Prices and free-tier limits in this category change every few months. Trust what you find on the maker’s site over what you read here.

Each tool below gets a short verdict in plain words. “Start here” means try it first, and it usually means it is free. “Worth it if” means the tool earns its price only for a particular need, so check that the need is yours before you pay. “Approach with caution” means the category is real but the risks are too.

What changed in 2025-2026

Headline shifts since the last comparable review:

Multimodal AI matured. GPT-4o, Claude 3.5 Sonnet and successors, Gemini Pro and successors all gained strong image understanding. The disability-relevant consequence: AI-driven image description (for people who are blind or have low vision), AI-driven scene understanding for navigation, and AI-driven text recognition all got dramatically better. The 2026 versions of these features are usable in ways the 2024 versions were not.

On-device AI grew. Apple Intelligence, Google AI Edge, Microsoft Copilot for Windows on Snapdragon X processors brought meaningful AI to consumer devices without cloud round-trip. The disability implications: faster response times for accessibility features, better privacy, more reliable function in low-connectivity contexts.

AI captioning standardized. Live captioning is now built into iOS, Android, Windows 11, and macOS. Quality is significantly better than 2024. Cross-language live translation is meaningful in 2026 in ways it was not before.

AI speech recognition for atypical speech improved substantially. Voiceitt, Whispp, and Microsoft’s accessibility speech recognition work have all advanced. The 2026 versions actually work for many people whose speech the 2024 versions could not process reliably.

AI productivity tools became cognitive accessibility tools. ChatGPT, Claude, Gemini for cognitive support (ADHD, executive function, learning disabilities, brain injury) has become substantial. Specific tools designed for the use case (Goblin Tools, Reclaim.ai, others) have grown.

AI companions and conversation became more serious. Replika, Character.AI, and the emerging category of AI partner tools have become more sophisticated. The implications for socially isolated people with disabilities (specifically those whose disability or care context limits social opportunities) are real and contested.

AI hype settled. Several over-promised tools quietly stopped marketing aggressively as the gap between promise and delivery became visible. You can now judge a tool on what it actually does, not on what it promised.

Vision aids

Microsoft Seeing AI

Still the standard free option, and on Android as well as iOS since December 2023. The 2025 updates brought significantly better object recognition, better text reading in challenging lighting, better scene description. The 2026 version is genuinely useful as your primary way of interpreting what is in front of you if you are blind.

Start here. It is free.

Be My Eyes (with Be My AI)

The human-volunteer model plus the AI-first option. Be My AI uses GPT-4-class image understanding for visual interpretation; you can escalate to a human volunteer if the AI does not resolve the question. The combination is one of the most useful disability tech experiences of the past two years.

Start here too, alongside Seeing AI. Both are free.

Aira

Human-interpreter subscription service. Still useful for situations that require sustained human interpretation (complex tasks, travel, specific workplace contexts). Read the plan structure carefully, because it is sold in minutes rather than in months. Aira’s own subscriptions page lists Silver at $26 USD for 20 minutes a month, $50 for 35 minutes and $84 for 55 minutes; Gold runs $132 to $356 for 90 to 245 minutes; Platinum runs $528 to $1,160. Twenty minutes a month is roughly one grocery shop. Aira is also free to use inside Access Partner locations, which is worth checking before subscribing, and it lists gift cards and employer or vocational rehabilitation funding as routes for people who cannot carry the cost personally.

Worth paying for if you need sustained human interpretation. Work out your realistic monthly minutes first, then pick the plan, and use the free Access Partner locations where you can.

OrCam MyEye 3 Pro

Head-mounted wearable that clips to a spectacle frame. The current model is the MyEye 3 Pro, and OrCam lists it at $4,250 USD, with 12, 18 and 24 month payment plans. That is above the range this guide previously gave. For people who need always-on visual assistance and can fund it, it is useful. For most people, smartphone-based alternatives now cover most of the same ground at no cost, so try those first and price the OrCam against what they leave undone.

Worth it only if the free smartphone tools leave a real gap. For most people they do not.

Envision Glasses

Smart glasses with AI text reading, scene description and the Ally AI assistant. Envision lists the Home Edition at $2,499 USD. One thing to know before buying: Envision Glasses are built on Google Glass Enterprise Edition 2, which Google discontinued in March 2023. Envision continues to sell and update the product, and Ally runs on it, but the underlying hardware is out of production, so ask about repair and replacement before you commit.

Worth considering, but price it against the OrCam and against the free smartphone tools, and ask the repair question before you buy.

Google Lookout

Android equivalent of Seeing AI; significantly improved in 2025.

Start here if you are on Android. It is free.

AI image description in ChatGPT, Claude, Gemini

If you want detailed verbal descriptions of complex images (charts, infographics, art), these models can do what the dedicated apps cannot. The use case is different from on-the-go visual interpretation but is genuinely useful.

Worth using for detailed image analysis, document review, and engaging with art and other visual content. Free tiers are available.

Hearing accommodation

Live Captions (iOS, Android, Windows, macOS)

All four major operating systems now have built-in live captioning that works well in 2026. The quality varies by context but is usable for meetings, video content, in-person conversation in many situations.

Turn this on first. It is built in and it is free.

Live Transcribe (Google)

The dedicated Google app for transcription is still useful, particularly for situations where the built-in tools do not cover.

Worth adding. It is free.

Otter.ai

Meeting transcription with speaker identification, search, and editing. Has held up as a meeting tool over the years.

Worth it for meeting transcription specifically.

Microsoft Teams / Zoom / Google Meet captions

Built into the major meeting platforms; quality has improved substantially.

Use the platform’s built-in captioning by default.

Voiceitt

Speech recognition trained on atypical speech (dysarthria, post-stroke, cerebral palsy, ALS). Strictly this belongs with speech tools rather than hearing ones, but it lands here because it is how many people get understood in a conversation. Significantly improved in 2025-2026, and it now works for many people whose speech mainstream recognizers cannot process. Voiceitt is subscription software: $49.99 USD a month or $499.99 a year, with a 30-day free trial. Take the trial before you commit; it learns your speech, and how well it does is individual.

Start here if your speech is atypical.

Whispp

Real-time voice clarification for atypical or whispered speech. Newer than Voiceitt; emerging usefulness.

Worth comparing against Voiceitt for your own speech.

Real-time translation

Vasco Translator, Pocketalk, Google Translate live translation. For deaf and hard-of-hearing people moving between languages, real-time translation has become genuinely useful.

Worth it if you regularly need conversation across languages.

Speech and AAC

Proloquo2Go, TouchChat, LAMP Words for Life

The major AAC apps. These predate the recent AI wave but have integrated AI features (predictive text, voice cloning for personalized synthesis, smarter symbol selection). All three are mature and competitive.

Any of the three is a sound choice. Which one suits you is an individual decision.

Dedicated AAC devices

Tobii Dynavox Indi, PRC-Saltillo Accent. Still relevant if you need dedicated hardware. Tobii’s eye-tracking remains the standard for people with significant motor disabilities.

The right answer if you need dedicated hardware rather than an app.

Voice cloning for personalized AAC voices

Acapela my-own-voice, ElevenLabs, and Apple Personal Voice. Acapela is a Tobii Dynavox subsidiary, not a Microsoft product, and my-own-voice builds a synthetic voice from about 50 recorded sentences. People with progressive conditions (ALS, Parkinson’s, late-stage MS) can bank a voice that keeps their own voice identity after speech is gone. This is one of the most significant assistive tech advances of the past two years, and the timing is the part that matters: banking works best while speech is still clear, which is usually earlier than anyone wants to think about it.

Act on this if you are facing progressive speech loss. The window for banking your voice matters.

Cognitive support and executive function

ChatGPT, Claude, Gemini, Copilot

The general-purpose AI tools have become significant cognitive accessibility aids. Use cases include:

• Task breakdown where executive function is the barrier (ADHD, autism, TBI, post-COVID brain fog)
• Summarization of complex documents (cognitive load reduction)
• Email drafting and editing, if written communication is taxing
• Schedule management
• Memory support (knowledge external storage)
• Decision support (working through choices with structured guidance)

Worth using if you have cognitive access needs. The free tiers are useful and the paid tiers add capacity.

Goblin Tools

A set of small tools built by developer Bram De Buyser for neurodivergent executive function (autism, ADHD). “Magic ToDo” breaks a task into steps, with a spiciness slider controlling how far down it goes; the “Estimator” guesses how long something will take; the “Compiler” turns a brain dump into a structured plan. The web version at goblin.tools is free, with no signup and no usage limit. The iOS and Android apps cost about $3.99 USD, a one-time purchase and not a subscription.

Worth using, and built for autism and ADHD executive function specifically.

Reclaim.ai and Motion

AI scheduling tools that absorb calendar complexity. If scheduling is where your executive function goes to die, handing that off is worth real money. Cal.ai appeared in earlier versions of this list and has since changed: Cal.com retired it as an email assistant and relaunched it as a voice agent aimed at taking bookings, which is a different job from managing your own calendar. It is no longer the personal scheduling assistant described here.

Worth it where scheduling is the thing that costs you most.

Glean

Note-taking for learning disabilities and ADHD students. Captures audio, integrates transcription, supports active engagement during lectures. The 2026 version has substantial AI features.

Worth it for students with learning disabilities or ADHD.

Read&Write (Texthelp)

Comprehensive literacy support for dyslexia, learning disabilities. Long-established; integrates AI features.

Worth it if you have dyslexia or another learning disability.

Otter.ai and Read.ai for meetings

AI meeting assistants reduce note-taking cognitive load.

Worth it if your work is meeting-heavy.

Notion, Obsidian, Roam Research with AI features

External-memory systems for people managing cognitive load (TBI, post-COVID, ADHD). The AI features in the 2026 versions of these tools help with the cognitive overhead of maintaining the system.

Worth trying. Which one fits is very individual.

Mental health

Calm, Headspace, Wysa

The major mental health apps. AI features have been added across the category; measured effectiveness for clinical mental health support remains modest, but for general support these tools are useful. Woebot is gone. It appeared in earlier versions of this list, and it was the best-evidenced product in the category, but Woebot Health announced in April 2025 that it was closing the direct-to-consumer app, which came off the market on 30 June 2025, and pivoted to clinical and enterprise licensing. Its own FAQ now says Woebot is available to new users only in the United States, and only through a study or an access code from a provider, employer or partner. If you were using it, it is not coming back in that form, and that is worth knowing rather than discovering.

Useful alongside therapy, not instead of it.

AI therapy chatbots

A contested category. Some people find them genuinely useful for between-session support, or where therapy is not available at all. Some clinical research raises concerns about AI in mental health, particularly for people in crisis. The evidence is mixed.

Approach with caution. This is not a substitute for professional support.

AI companions

Replika

The longest-running AI companion app. Used by some socially isolated people with disabilities as part of their social and emotional life. The relationship-mode features are contested among the people who use it.

Worth knowing about, with clear eyes about the limits. The category is real, but whether it helps depends a great deal on your own circumstances.

Character.AI, Kindroid, Nomi

Newer AI companion platforms. Each has its own approach, and the use case overlaps with Replika. One change matters more than any feature: from late November 2025 Character.AI barred users under 18 from open-ended chat, using age assurance to route minors to structured creative tools instead. The company made the change after lawsuits, a US Federal Trade Commission inquiry and proposed federal legislation on AI companions and minors. If you are recommending this category to a young person with a disability, that is the first thing to check, not the last.

Approach with caution, and note that Character.AI is now closed to under-18s for open-ended chat.

Daily living and home

Smart home AI (Alexa, Google Assistant, Apple Siri)

Voice-controlled home automation has matured. If you have mobility, vision or cognitive limitations, the gain is substantial.

Worth setting up if voice control removes a barrier for you.

AI in smart appliances

Voice-controlled ovens, AI-supported cooking, AI medication dispensers. The category exists; specific products vary in usefulness.

Judge these one product at a time rather than as a category.

Education and work

AI writing assistants for accommodation contexts

ChatGPT and similar for students and workers with writing difficulties (dyslexia, learning disabilities, brain injury, others). The accessibility benefit is substantial.

Worth using.

AI for translation in accommodation contexts

For deaf workers and students using ASL or LSQ, AI sign language translation has improved but remains imperfect.

Improving, but not yet a match for a human interpreter on complex content.

AI meeting summary and transcription

Otter.ai, Read.ai, Microsoft Teams Copilot, Google Workspace AI. If you have hearing or cognitive access needs, a post-meeting AI summary is genuinely useful.

Worth it if you sit in a lot of meetings.

What is overhyped

A few categories that have not delivered on their hype:

AI prosthetic control: the promise of brain-computer interface controlled prosthetics has been around for years and remains mostly in research settings. Some products are commercially available but the cost-benefit relative to advanced mechanical and myoelectric prosthetics is unclear.

AI-driven adaptive equipment “intelligence”: smart wheelchairs that route automatically, AI-powered medical equipment that anticipates needs. Some real products exist; many are early-stage and not yet meaningfully useful for daily life.

General-purpose disability assistance robots: the social-companion robot category has existed for years; consumer-level products have not delivered substantially.

AI symptom checkers for medical guidance: the category exists (Ada, K Health) but for people with complex disability needs an AI symptom checker is not a substitute for clinical care. Babylon appeared in earlier versions of this list and no longer exists: its US operations filed for Chapter 7 bankruptcy in August 2023 and the UK business was sold. Its symptom checker had already drawn regulatory scrutiny and published accuracy criticism before the collapse, which is a useful reminder that a confident interface is not a clinical validation.

What is emerging and worth watching

On-device AI for accessibility: Apple Intelligence and equivalent platforms have started to integrate accessibility features (Personal Voice, smart routing of notifications, AI-driven content adaptation). The 2027 versions of these features will probably be substantially more capable.

AI-driven sign language translation: improving but still imperfect; worth annual evaluation.

AI in eye-tracking communication: Tobii Dynavox and others are integrating language models with eye-tracking input to predict and complete utterances. Substantial improvements in usability for ALS and people with significant motor disabilities.

AI in continuous glucose monitoring: diabetes management AI is improving the experience for people with diabetes (a frequent secondary condition in disability populations).

AI for wheelchair navigation and obstacle avoidance: research-stage but maturing.

AI in mental health and crisis support: controversial but evolving rapidly.

What to spend on and what to skip

If you have $0:

• Microsoft Seeing AI (free)
• Be My Eyes / Be My AI (free)
• Built-in OS live captions (free with OS)
• ChatGPT free tier or equivalent
• Google Lookout (free on Android)
• Goblin Tools (free)
• OS-level smart home features

If you have $20-$50 USD/month:

• Paid tier of ChatGPT or Claude (better cognitive support)
• Otter.ai paid (better meeting transcription)
• Voiceitt at $49.99/month if atypical speech recognition is what you need, and it will use most of this band on its own. Goblin Tools does not belong in this band: the web version is free and the mobile app is a one-time $3.99, not a subscription.

If you have $100+ USD/month:

• Aira, where $132 buys 90 minutes a month and the free Access Partner locations cover the rest
• Multiple paid AI tool subscriptions for cognitive support
• Specialty subscriptions (Read&Write, Glean, others)

Hardware ($500-$5,000 USD):

• AAC device where dedicated hardware is needed
• Tobii eye tracker for people who cannot use touch or switch access reliably
• Envision Glasses Home Edition at $2,499 or OrCam MyEye 3 Pro at $4,250 for sustained visual interpretation
• Voice banking while speech is still clear, for people facing progressive speech loss

What to do if you are starting from scratch

If you are new to AI accessibility tools:

1. Start with the free OS-level features. Live Captions, smart speakers, basic AI assistance. These are immediate, free, and often surprisingly useful.
2. Add one general-purpose AI assistant. ChatGPT, Claude, or Gemini. Free tier is enough for many uses.
3. Add condition-specific tools as needed. Voiceitt for atypical speech, Goblin Tools for executive function, Read&Write for dyslexia, etc.
4. Add paid tiers selectively. When a tool has proved useful and the paid features matter for your specific use, upgrade. Do not pay for tools you have not tested.
5. Connect with community. AI accessibility tools have active user communities. Other people with your specific access needs often have specific recommendations.

What this guide does not do

This guide does not replace your OT, AT specialist, audiologist, or other accessibility professional. Individual recommendations require individual assessment. The guide does not address every AI tool or every disability category; the field is large and moving fast.

It does describe the actual picture as of mid-2026: what works, what does not, what is changing, what to spend on, what to skip. The next annual update (mid-2027) will assess what changed in the intervening year and update the recommendations.

The AI accessibility space is one of the genuinely useful applications of recent AI advances. The tools that work are worth using. The hype is worth ignoring. The annual update is to keep the practical knowledge current.

Sources

Prices and product status below were taken from each maker’s own pages and checked in July 2026.

• Microsoft Accessibility blog, on Seeing AI reaching Android
• Aira, Personal Subscriptions, for the Silver, Gold and Platinum plan minutes and prices
• OrCam MyEye 3 Pro and OrCam on financing the cost of a device, for the $4,250 price
• Envision Glasses, and 9to5Google on Google discontinuing Glass Enterprise Edition 2
• Voiceitt FAQ, for subscription pricing and the free trial
• Tobii Dynavox on voice banking and message banking, for Acapela my-own-voice
• Goblin Tools, and the iOS and Android store listings for the one-time app price
• HLTH on Woebot Health closing its consumer app, and Woebot Health’s own FAQ on who can still get it
• Character.AI Help Centre on the changes for users under 18, and CNBC on the timing and the reaction
• Cal.com, on retiring and relaunching Cal.ai
• TechCrunch, on Babylon’s bankruptcy

What this article does not claim: there is no systematic trial evidence behind most of these recommendations. This is a practical read of what works, not a research review.

Frequently asked questions

What changed in 2025-2026?

Headline shifts since the last comparable review: Multimodal AI matured. GPT-4o, Claude 3.5 Sonnet and successors, Gemini Pro and successors all gained strong image understanding.

Which AI vision aid is the standard free option?

Microsoft Seeing AI: still the standard free option. The 2025 updates brought significantly better object recognition, better text reading in challenging lighting, better scene description.

Which operating systems have built-in live captioning?

Live Captions (iOS, Android, Windows, macOS): all four major operating systems now have built-in live captioning that works well in 2026. The quality varies by context but is usable for meetings, video content, in-person conversation in many situations.

Which are the major AAC apps, and have they added AI features?

Proloquo2Go, TouchChat, LAMP Words for Life: the major AAC apps. These predate the recent AI wave but have integrated AI features (predictive text, voice cloning for personalized synthesis, smarter symbol selection).