Swapcard AI isn't a bolt-on feature. It's foundational.

Sherlock is Swapcard's AI assistant, powered by large language models (LLMs). It provides instant answers to participant questions about sessions, speakers, exhibitors, and venue logistics. Sherlock works in any language, understands each participant's role and activity history, and is available on web, iOS, and Android.

Cortex is an intelligence layer that integrates generative AI into Swapcard's event tools. It makes Sherlock model-agnostic, meaning the platform can select from multiple LLM providers—OpenAI, Gemini, Anthropic, Mistral, and Meta—based on the task, audience, and performance requirements.

Sherlock answers questions in real time. For example, if an attendee asks, "Which sessions cover sustainability in supply chains?" Sherlock surfaces the most relevant sessions, explains why they match the query, and offers to add them to the attendee's schedule.

Gismo is Swapcard's recommendation engine, built on non-generative AI algorithms. It powers smart matchmaking, personalized meeting schedules, recommended leads, and engagement insights. Gismo is open-source and uses a custom variant of TF-IDF mixed with a random walk approach.

Gismo recommends the most relevant sessions, exhibitors, and people to connect with based on profile data and real-time behavior. It generates personalized meeting schedules automatically and adapts continuously—the more an attendee uses the platform, the more precise the recommendations become.

Gismo surfaces the most relevant attendees as recommended leads, scored by interaction and profile fit. It matches exhibitors and sponsors with the right participants before, during, and after the event, and captures and enriches lead data automatically. Exhibitors using AI-powered matchmaking generate 2.5× more qualified leads.

Gismo provides real-time engagement insights across content, networking, and participation. It generates transparency into how recommendations are made, so organizers can configure and adjust matching logic in the platform backend.

Non-generative AI scales efficiently across thousands of participants in real time, processes data faster, and provides stronger privacy controls than LLM-based approaches. It's optimized specifically for the high-volume, real-time nature of event networking.

Swapcard's Cortex layer can select from multiple LLM providers—OpenAI, Gemini, Anthropic, Mistral, and Meta—depending on the task, audience, and performance requirements. This ensures the platform always benefits from the most capable reasoning engine available.