Synsira unveils Kind, a private desktop AI search tool

Synsira has launched Kind, a desktop application that uses artificial intelligence to search a user's own files and answer questions from that material rather than the open internet.

The product centres on local data. Users upload material into the app and ask questions against that collection. Kind returns answers based on those files and points to the source location within the stored content.

Kind positions itself as an alternative to general-purpose AI chat services that draw on online information. Those services can produce errors and may invent details when they cannot confirm an answer. Kind uses only uploaded data and responds with "I don't know" when the files do not support a query.

How it works

Kind supports a range of file types and personal archives. Users can organise and search content such as textbooks, research papers and articles, as well as media including audio files, photos and video.

The app presents responses with references showing where the information appears in the user's collection. The goal is to make clear what evidence supports each answer-and what content the tool did not use.

Synsira describes Kind as privacy-first because it runs on a user's computer and searches only that user's files. It is positioned as a way to keep sensitive information out of online services that require uploads to external systems.

Synsira also ties the desktop model to environmental impact, saying the software relies less on large data centres because it operates from a personal computer and local data.

Target users

Synsira is pitching Kind at people who manage large volumes of personal documents and reference material. The company highlights use cases ranging from freelance writing and content production to travel planning and academic work.

In one example, a copywriter could store drafts, creative assets and client background documents in the platform, then surface previously approved material when drafting new byline articles. A traveller could upload reviews, recommendations and personal media, then pull up preferred hotels and tips when preparing an itinerary. A professor could store research, lecture materials and presentation decks, then query the collection while preparing a speech.

Origins in education

Kind traces its origins to a tool built by founder Jonathan Schaeffer during his time teaching at the University of Alberta. The initial program answered student questions using course documents and videos, and later became the foundation for the commercial product.

Schaeffer is a former Canada Research Chair in Artificial Intelligence. Synsira says his research generated two Guinness World Records.

"People want AI to be helpful and trustworthy," Schaeffer said. "Current online-based AI tools have their flaws. Kind was built to overcome those challenges and is based on a simple idea: your data should work for you and only you. By grounding every answer directly in a user's curated files, we've created an AI experience that users can rely on for accurate information."

Languages and pricing

The interface supports English, Spanish, French, Portuguese (Brazil), German, Japanese, Simplified Chinese and Traditional Chinese. Synsira says the application also supports additional languages for user files and searches.

Users can download a limited version for free. A paid subscription, Kind Pro, costs USD $18.88 per month and is sold via Shopify.

Synsira says it plans additional features and updates in future versions.