Whenever any cutting-edge technology hits the mainstream, there’s an explosion of creativity as people experiment with designs, try to find solutions for old problems, and create new solutions looking for a problem. This is especially true in the accessibility industry. It happened with the internet, web 2.0, the blockchain, and now it’s AI’s turn. Three quarters of these products disappear without ever gaining adoption. But the ones that survive change the world forever.

In the case of AI, I feel like I’ve tested every single one of the available accessibility solutions. But these are the three that stuck; they made a real difference to my personal daily life, and I find myself coming back to them, even after trying other solutions.

BasiliskLLM

Named after the famous Roko’s basilisk, BasiliskLLM is a Windows desktop app that integrates with multiple AI’s, through multiple API providers. It offers audio transcription, image description, access to web search, and the ability to integrate with the NVDA screenreader through an addon. It sits in the system tray, and can be activated and dismissed via a shortcut key. It also has a tabbed interface allowing quick switching between multiple conversations, profiles for different prompts and presets, and allows switching models mid-conversation. I use it with OpenRouter so I can get access to all of the well-known AI models; that lets me, for example, quickly ask two different vision models the same question, if I want to verify the answer. I use BasiliskLLM for describing images, answering quick questions, testing models, and verifying the state of inaccessible interfaces. The only tool I wish it had that it’s missing is retrieval augmented generation (RAG), to make it easier for me to give it folders of manuals and documentation.

KAI###

KI is the beta multi-step assistant offered by the KAGI search engine as part of it’s premium plan. While not quite as accessible as Chat GPT, it’s still perfectly usable. The primary issue is that switching models with the keyboard can be slightly tricky. But it can read results as they come in,the sources are easy to find and check, and the conversation interface is easy to navigate. I find that the far superior search results offered by KAGI make all the difference when I’m doing AI research. Garbage in, Garbage out, after all. For anything detailed, I also find KI to be much better than using the web search functions of OpenRouter with BasiliskLLM. As well, it offers access to all of the major AI providers (google, openAI, deepseek) for a single subscription, and allows me to switch between them whenever I like. That plus Kagi’s excellent search engine makes it the winner of my monthly AI subscription dollars.

Seeing AI

There are dozens of AI accessibility apps on IOS, each with their advantages and disadvantages. But no matter what I try, I keep coming back to Seeing AI. First off, I like that it offers different AI’s for different functions, instead of trying to do everything with one model. This makes the results more accurate, and allows me to focus directly on what I’m doing: navigating the world, getting a description of a photo, scanning barcodes, identifying currency, or any of the dozens of other features it offers. It’s biggest disadvantage compared to other apps with AI functions (like Be My Eyes) is that it doesn’t include the ability to reach out to a real human to verify results. I keep Be My Eyes installed on my phone as a backup for that reason. But Seeing AI’s speed, accuracy, stability, and dozens of available functions has kept it as my primary mobile AI app for years now. I did, however, obtain a pair of Meta Smart Glasses recently. But I’ve only had them for a month, so it’s too early to say if or when they’ll replace my beloved Seeing AI.

Tools I’m Testing

These are the current AI based tools I’m testing; it’s too early to know if these will earn a spot in the toolbox or not. Expect to read more about them in the coming months!

  • Innosearch AI: using AI to make shopping more accessible and inclusive
  • Guide: have AI control your computer to help you use inaccessible apps and websites
  • Metta Smart Glasses
  • Glidance: an AI device to augment the cane. Not launched yet, but I’ve pre-ordered.
  • TypeAhead: similar to Guide for OSX

Conclusion

There are dozens of accessibility related AI apps and products in the marketplace, and it feels like new ones launch every day. The above three tools are just the ones that I go to daily; the industry is moving fast enough, though, that this list could look completely different in a year. So what tools are you using? Are your toolbox picks different from mine? Did I miss any upcoming tools I should be evaluating? Let me know!