Knowledge Agent

Bring your knowledge to life

Knowledge Agent is the engine underneath all of Kundo. It learns from your real sources, powers chat, forms and agent drafts with the same knowledge, shows the source in every answer and learns what it couldn't answer.

Book a demo

The next-generation knowledge engine:

Connect your sources

Connect your help center, website, past tickets and CRM. Build the AI from real conversations, not just uploaded files.

Visible sources

See the source of every answer, so both customers and your team can check where the information comes from.

Self-learning

Unanswered questions get flagged and learned. The AI gets better the more it's used.

RAG architecture

Ground every answer in your own data. Semantic search in your index first, then answers generated from what was found.

One knowledge base

Maintain knowledge in one place and make it available in chat, forms, drafts and summaries.

Feedback

See which AI answers got negative feedback from customers and fill the gap where it is.

Support for more languages

Answer in the customer's language from knowledge written in yours. One knowledge base covers every market.

External data

Integration

Bring order status and customer profiles from external systems into AI answers, read-only, with no transaction risk.

Build your AI on real conversations

Most AI solutions are trained on uploaded documents that are out of date the moment they're uploaded. Knowledge Agent reads what actually happens in your customer service.

Connect multiple sources

Build the AI from your help center, website, past tickets and CRM.

Let knowledge keep up with your business

Keep knowledge current as your business changes, without a separate maintenance project.

Ground the answer in your data

Ground every answer in your own data instead of a general model.

Maintain it in one place. Use it everywhere.

The same knowledge powers your AI Chat, your AI Forms and the reply support in ticketing. Customers get the same answer wherever they ask.

Give customers one answer, not four different ones

Answer the same way in chat, in the form and through the agent.

Automatic conflict handling

Knowledge Agent spots right away when two knowledge sources contradict each other and brings in a person who can decide which one is correct.

AI is only as good as the knowledge it's built on

When AI in customer service goes wrong, it's rarely the model. 95 percent of AI failures trace back to fixable gaps in the knowledge base (AI4SP, 2025). Air Canada lost a court case in 2024 after its chatbot described a policy that didn't exist. The problem wasn't that the AI talked. It was that it didn't have the right things to talk about. Knowledge Agent is built around that. It searches your own knowledge first, then answers from what it found, and shows the source. Whatever it can't find, it flags, so your team can fill the gap and the next customer gets the answer. The same knowledge sits underneath chat, forms and agent drafts, and it grows every time a good answer is published from a ticket. Knowledge is our core at Kundo, and you can see it in the quality of all our products.

Knowledge is what sets you apart

95%

of AI failures are caused by fixable gaps in knowledge management, not by the models (AI4SP, 2025).

80%

of common customer service issues will be resolved autonomously by agentic AI by 2029, at 30 percent lower operating costs (Gartner, 2025).

87%

of customers say companies using generative AI must offer a way to reach a person (Gartner, 2026).

Everyone has access to the same language models. What decides whether the AI answers correctly about your business is the material it has to go on.

More than 500 customers in the Nordics:

"We're seeing a 274% increase in chat conversations compared with last year, while fewer tickets are passed on to customer service."

Jessica Svärd, Customer Service Manager, Impecta Fröhandel

359%

more customers helped through AI Chat in six months at SKB.

Tradera Toyota Financial Services Arla Foods SKB Vasakronan Sveriges Radio SMHI Plantagen Hemglass Carlsberg

Frequently asked questions about Knowledge Agent:

What is Kundo Knowledge Agent?

Knowledge Agent is the AI engine underneath all of Kundo. It learns from your own sources, powers AI Chat, AI Forms, AI Drafts and AI Summary with the same knowledge, shows the source in every answer and flags questions it couldn't answer. It isn't a separate product. It's what makes Kundo's AI features answer the same way wherever the question is asked.

Which sources does Knowledge Agent learn from?

Published help center content, your website, past customer tickets and conversations, and CRM data. That way the knowledge reflects what is actually asked and answered in customer service, not just what someone had time to write down in a document. Through integrations, external data such as order status can also be brought into answers, read-only.

What is RAG and why does it matter?

RAG stands for retrieval-augmented generation: the AI retrieves relevant text from your own knowledge source before it writes the answer, instead of answering from its general training. It's the industry standard for AI in customer service because it grounds the answer in your data, makes the source traceable and lets you update knowledge without retraining the model.

What happens to questions the AI can't answer?

They're flagged as unanswered, so your team sees exactly where the gaps are and can fill them. Once the answer is published, the next customer gets it right away. In the meantime, the customer is offered a way to reach a person, which 87 percent of customers expect from companies using generative AI (Gartner, 2026).

Do we need to build a knowledge base first?

No, but it goes faster if you already have published content. Knowledge Agent can start with your website, existing help center and past tickets, and the knowledge then grows in daily work as agents publish answers from tickets in one click. The quality of AI answers follows the quality of the material: 95 percent of AI failures come down to gaps in exactly that (AI4SP, 2025).

Where does the model run and how is our data handled?

Kundo is hosted in the EU, and the language model runs via OpenAI in Microsoft Azure's EU environment. Your knowledge is used to answer your own customers and isn't shared between Kundo customers. The source reference in every answer also means you can always see exactly which content an answer is based on.

Is Knowledge Agent a separate product we buy?

No. Knowledge Agent is the engine included underneath Kundo's AI features: AI Chat, AI Forms and the AI support in Kundo Workspace. Different products are included in different subscription tiers.

How does Knowledge Agent avoid making up answers?

Through RAG architecture: it searches your own index first, generates the answer only from what it found, and shows the source. Questions without coverage get flagged instead of answered with a guess. That's the difference from letting a general language model answer freely, which is what cost Air Canada its court case over its chatbot in 2024.