Skip to content

Latest commit

 

History

History
90 lines (54 loc) · 3.36 KB

faqs.mdx

File metadata and controls

90 lines (54 loc) · 3.36 KB
title description
FAQs

Kontext21 is the AI infrastructure layer for user context.

K21 SDK makes it easy to personalize software by capturing & retrieving end-user context from sources like their screen, browser and many more.

See our [Use Case Gallery](https://kontext21.com/use-case-gallery). Currently many use cases around task & time tracking as well as knowledge creation. We are focused on MacOS and Windows. 
With Linux and mobile coming soon. Please reach out if you want to build for them.
You can always re-invent the wheel. Kontext21 provides model agnostic infrastructure layer to build context aware software applications. Using a model is only part of the work to be done. 
* Easy experimentation with different models	

* Receiving structured output

* Merge with different sources of context both from within your software (known entities such as contacts, projects etc.) but also from outside your software (see our list of integrations

* Both experimentation as well as testing & observability are crucial.

* Remove PII information

* Optimize for cost
1. Experimentation 
2. Implement

3. Beta Launch - Make it on-demand, side-by-side installer

4. Production Launch - Include in main installer

5. Continuous Improvement
With computer use and browser use there is usually no way to get back structured data about what's going on for the user. Kontext21 can be used to trigger computer use or also to build your own, highly customized version of computer use. Depends on the use case but less than 10% when running with 1 fps. We focus on affordable, value based pricing. We both have a business to run. Our focus is creating more value for your business through your customers/users. 
Check out our [pricing](https://kontext21.com/pricing) and [license](https://kontext21.com/license) for more details.
* Pattern: Inform the user about what data is collect and why 
* Pattern: Inform end-user when there data is accessed = prevents “spying”

* Pattern: Keep the user in control - Create local drafts before sharing the data with other users. sharing with consent, Chinese wall (nothing from one side leaks to the other. As the user I can be sure that what gets analyzed locally is not made available to my employer & colleagues).

* Pattern: Let user inspect the recorded data

* User trust often means “local first”
* OCR & Document OCR 
* Screen Scraping & Web Scraping

* RPA

* Computer & Browser Use