Langdock GmbH, a startup with an enterprise chatbot that allows users to change the underlying language model, has raised $3 million in seed funding to support its growth efforts.

TechCrunch reported the investment today. General Catalyst and La Famiglia led the round with contributions from about two dozen other backers. According to Langdock, Y Combinator and the founders of several venture-backed German startups were among the participants.

Berlin-based Langdock provides a chatbot platform of the same name for business users. The software can generate text, search the web for information and automate mundane tasks in third-party applications. Langdock says that the data users enter into the interface isn’t used to train the underlying large language model.

Exactly which LLM the platform leverages to process queries is determined by the user. Langdock offers a choice of models from OpenAI, Anthropic PBC and Aleph Alpha GmbH, an artificial intelligence startup backed by more than $500 million in funding. The chatbot also works with several open-source LLMs including Meta Platforms Inc.’s Llama 2 series.

According to Langdock, workers can change the AI that its platform uses to generate responses in a few clicks. The company says that the ability to quickly change models helps customers avoid becoming locked into a single provider’s LLMs.

Besides its core set of chatbot capabilities, Langdock also offers a number of complementary features. A built-in collaboration tool allows members of a team to share LLM prompts with one another. After Langdock responds to a prompt, it suggests follow-up questions that the user can enter to obtain more information about the topic at hand.

The chatbot lends itself to retrieving data from both the public web and a company’s internal systems. According to Langdock, its platform is capable of analyzing data stored in Google Drive, Confluence and other popular productivity applications.

The platform’s integrations with third-party applications also power a capability called Workflows. According to Langdock, it allows users to create custom versions of its chatbot optimized for specific tasks. Members of a product development team, for example, could set up a workflow that summarizes feature suggestions submitted by users.

Such workflows include an interface tailored to the task they’re designed to automate. According to Langdock, the interface provides controls that enable users to customize settings such as the length and tone of LLM responses. If a model requires additional input from a worker to generate a response, the platform can check that the input has been entered correctly before processing the query.

Langdock’s installed base includes more than a half dozen customers including pharmaceutical company Merck and GetYourGuide, a travel startup that received a $2 billion valuation last year. The chatbot maker says those companies have rolled out its platform to tens of thousands of workers. 

Image: Unsplash