Skip to main content

Overview

kapa.aiย makes it easy for developer-facing companies to build AI support bots for their community. Teams at Docker, OpenAI, Prisma and NextJS use kapa to level up their developer experience and reduce support.

Want access within 12-24 hrs?

Head to kapa.ai to sign up and the team will be in touch within a few hours to get you set up with a demo instance.

What is kapa?โ€‹

kapa.ai lets you create LLM based chatbots that can answer developer questions about your product. Once created your kapa instance will not talk about anything but your software. This is possible because kapa is a retrieval augmented generation system (RAG) optimized for answering technical questions.

The kapa.ai platform can index all of your technical knowledge sources and then start answering questions using your content. The most common use case of kapa is to directly deploy it to your community of developers. With kapa your developers can find instant answers to their questions and solve their problems faster.

Checkout these live examples to understand uses cases and examples of where you can deploy kapa.

How does kapa work?โ€‹

The kapa.ai system consists of multiple key components:

Data Pipelinesโ€‹

A large part of kapa.ai are specialized data pipelines that ingest your various technical knowledge sources. These pipelines connect to your data sources, pull their content, transform and index them using embeddings. Once your content has been indexed it is semantically searchable. These data pipelines also refresh your content periodically so it is always up to date.

After the kapa team initially gets you onboarded, you can manage your data sources in the kapa dashboard (access granted after sign-up). See this guide on how to add knowledge sources to your kapa instance.

RAG Pipelineโ€‹

When a developer submits a question kapa performs retrieval augmented generation (RAG). In a first step kapa searches the index of all your content to find the most relevant pieces of information. In a second step kapa uses an LLM to answer the question based on the retrieval result. When answering questions kapa sticks closely to your content which puts you in control and minimizes hallucinations. However, kapa can infer a complicated answer from multiple pieces of separate information.

Deployment Integrationsโ€‹

kapa.ai is an end-to-end solution and comes with the integrations you need to immediately deploy kapa to your community of developers or internally within your company. The following deployment integrations are available:

  • Website Widget: Create an embeddable kapa.ai website widget that can be integrated in 1-line into any website.
  • Slack Bot: Create a kapa.ai Slack bot that can either auto-reply to forum posts or answer questions when mentioned.
  • Discord Bot: Create a kapa.ai Discord bot that can either auto-reply to forum posts or answer questions when mentioned.
  • Zendesk Agent App: Use the Kapa Zendesk support app to reduce ticket solution time by using instant reply drafts.
  • Build on API: Build on top of the kapa.ai core infrastructure and leverage the API to answer technical questions automatically.

What can I do with kapa?โ€‹

Deploy kapa to your community of developers

You can deploy kapa directly on your documentation page, in your Slack or Discord community to answer questions. The knowledge sources of most software companies are vast and spread across many different places. This makes it difficult for developers to find what they are looking for. So instead of manually searching for the right piece of information your community can just ask kapa and get an instant answer based on the most relevant content. This helps new developers onboard faster and improves everybodys experience.

Docker has kapa deployed for 13m monthly documentation visitors.

Deploy kapa in your company internally

Many of your employees have technical questions about your own product. For example sales and customer success have to respond to technical questions from your customers. Deploy kapa internally in your company so your internal teams can find answers faster without asking engineering causing them to context switch. This also helps new employees onboard faster.

Use kapa to answer support tickets faster

You can install the Zendesk Agent App inside of your Zendesk workspace. Kapa can then understand the history of your support tickets and draft potential replies to them making it easier for your agents to answer. In drafting replies kapa can leverage all of your past tickets as well as your other knowledge sources. With kapa your internal support teams will get back valuable time for other high value tasks like creating content.

Mapbox uses kapa inside Zendesk.

Deflect support tickets

Developers often open support tickets because they can not find an answer by themselves. In many cases their questions could have been answered using the documentation. Kapa will start deflecting support tickets by helping developers find answers by themselves.

kapa handles 25% of Astro's support questions.

Improve your documentation

kapa offers a suite of analytics that help you improve your documentation. You can understand what questions are commonly asked and whether they are documented well or not. You can also understand what parts of your documentation are used the most to decide what areas to focus on.

How do I get started with kapa?โ€‹

Currently, the kapa.ai team will help you onboard onto the platform. Want access within 24 hrs? Head to kapa.ai to sign up and the team will be in touch shortly to get you set up with a demo instance.