Botpress

Botpress Studio allows companies to choose which LLM powers an agent, in addition to developing chats. Instead of being limited to a single supplier, teams can choose from a variety of leading models, plug in a model running on their own hardware, or configure several models for different workflows, utilizing a quicker model where speed is important and a stronger model where accuracy is critical. This transforms model selection into an option that teams may customize, rather than a decision baked into the platform.

Botpress features ‘Autonomous Agents’ that can employ LLMs to decide what to do during a discussion, selecting tools and steps based on the user’s actual needs. A whole conversational flow can be derived from a single prompt or combined with structured logic for phases that must be performed in the same manner every time. These agents also manage several formats in a single interaction, switching between text, pictures, and speech as needed.

Knowledge bases consolidate disparate information into a single location where agents can look when answering a question. Teams can import entire websites and FAQ pages, upload PDFs and documents directly, or manage anything via the API to keep material up to date. Instead of guessing, answers are based on authentic source material.

Tables provide agents with an integrated database for storing and retrieving data during conversations, eliminating the need for a separate backend. Built to contain millions of entries and stay in sync across sessions, they allow teams to query in plain English rather than manually coding filters, which is useful for data that changes frequently, such as inventory counts or appointment slots.