kukula.ai

The Kukula AI platform

On the platform, AI agents work with your documents and with Microsoft 365, and Kukula AI takes care of operations.

Is our coating approved for galvanized steel parts on a south-facing facade?

Source:Technical data sheet, Resistance and Substrates sections
Ask the agent

Capabilities

What the agents can do

Several agents can be set up on the platform, and each of them works with the documents and systems approved for its task. In Microsoft 365, an agent only sees what the person currently using it has access to.

  • Answers from your documents and company knowledge, with a reference to the source
  • Connection to Microsoft 365 with email, calendar, SharePoint and OneDrive
  • Product advisor on your website that answers visitors' questions from approved documents
  • Agents that split large tasks into subtasks and work on them in parallel
  • Scheduled tasks that an agent runs daily or on set days of the week

Data protection

How your data is handled

Before a company hands documents and emails to an AI application, it needs to be clear where that data is stored and processed. On request, Kukula AI switches on EU mode for your entire company, so that it applies to all of your agents.

  • Platform and data storage in the EU
  • On request, EU mode keeps documents, chats, memory, audio and images in the EU
  • In EU mode, the language models run exclusively on EU endpoints
  • Model calls without data retention at the provider (zero data retention)
  • No training on your data

Operations

Running the platform

An AI application needs ongoing care after launch as well, because documents change and new language models appear regularly. On the platform, Kukula AI takes on this care for all agents that your company uses there.

  • Monitoring, maintenance and updates by Kukula AI
  • New language models become available without a separate project
  • Your IT only provides access to documents and systems

How a project runs on the platform

Because the technical environment already exists, a project is mainly about your documents and the processes of your team.

  1. 01

    Clarifying the task and documents

    At the start, you work out with us which task the agent should take on and which documents and systems it needs for that. Before launch, we agree with you on how you will measure the value of the agent later on.

  2. 02

    Setup and test phase

    Kukula AI sets up the agent on the platform and connects it to the agreed documents and systems. Colleagues who know the difficult cases then test the agent with real requests from their daily work. Where answers are incomplete, we add documents or instructions before the agent is released to everyone involved.

  3. 03

    Team rollout and ongoing operation

    After the release, we train your team so that it is clear which questions the agent can answer and where its limits are. In ongoing operation, Kukula AI monitors the agent and reviews with you after launch how the agreed metric develops. Further use cases can then be added on the same platform without building the technical environment again.

Is the Kukula AI platform right for your project?

In the intro call, we look at which systems your team works with today and what requirements your IT sets for data protection. At the end, you receive a first assessment of whether an agent is suited to your task and which documents it would need.

Let's get in touch→