F

A machine learning approach where a model is trained across multiple decentralized devices or servers holding local data samples, without exchanging the raw data. Federated learning enables AI training on sensitive data (medical records, financial data) while preserving privacy, because the data never leaves the local device — only model updates are shared.

A prompting approach where you include a small number of examples (typically 2–5) in your prompt to show the AI the pattern or format you want it to follow. Few-shot prompting is more effective than zero-shot for tasks that require a specific output structure or style.

The process of taking a pre-trained AI model and continuing to train it on a smaller, specialized dataset to improve its performance on a specific task or domain. Fine-tuning allows businesses to customize general AI models for their specific use case, terminology, or communication style.

An AI meeting assistant that automatically records, transcribes, and summarizes meetings from Zoom, Teams, Google Meet, and other platforms. Fireflies can extract action items, create searchable meeting archives, and integrate with CRMs to log call notes automatically.

A large AI model trained on broad data at scale that can be adapted to a wide range of downstream tasks. Foundation models like GPT-5, Claude 4, and Gemini 3 are the base layer on which most AI applications are built. The term was coined by Stanford researchers in 2021.

A feature of AI models that allows them to call external functions, APIs, or tools as part of generating a response. Function calling is the technical mechanism that enables AI agents to take real-world actions — such as searching the web, querying a database, sending an email, or updating a CRM.