Architecture

K Pro’s architecture

An agentic harness designed for intelligence, reliability and interoperability

Explore K Pro’s architecture

K Pro

Owkin’s AI scientist - built to power smarter pipeline decisions

K Pro integrates flexibly and securely to provide powerful support to biopharma users.

Smart

K Pro can call data, tools, agents and skills from user systems, Owkin’s expert catalogue, and third parties to answer a broad range of questions.

Reliable

Expert-written skills guide K Pro’s reasoning to ensure reproducible scientific results.

Interoperable

MCP integration allows K Pro to be called through a user’s existing internal system, as well as operating stand-alone.

Secure

Deploy on your infrastructure and behind your chosen firewall - ISO27001 and GDPR compliant.

K Pro brings data, tools and skills together through an agentic AI harness optimized for biopharma

Dig deeper into how it works

K Pro’s Architecture

The K Pro agent reasons with state-of-the-art AI and can call data, tools and skills.

LLM reasoning

K Pro reasons through a state-of-the-art enterprise LLM - that can be substituted according to need.

Tools & data

It can call tools, customer-built agents, third-party services, and internal and external data platforms, through MCP and internal platform data access.

Expert skills

Skills provide instructions that help K Pro orchestrate tools more effectively and reproducibly.

MCP gateway

K Pro integrates with enterprise data systems and APIs to connect to an organization’s systems rather than replacing them.

Code executor

To address tasks that cannot be predefined, the K Pro can generate new programs that are executed in a sandbox environment.

Multiple agents

For deep research, campaigns use parallel autonomous sub-agents that execute independent research on multiple different angles of a research question at the same time.