Demo Night Recap — Fear Detection to Production-Ready AI Agents

Last week, the TorontoAI community came together for another hands-on Demo Night — this time focused on practical, real-world AI applications rather than hype. Despite snowy weather and a smaller in-person turnout, the evening delivered deep technical insights, candid startup stories, and live demos that showcased how AI can move from experimentation to meaningful impact.

The event featured two main demos:

1. A live walkthrough of a FearSense application built using Falcons AI's fear-mongering detection model.
2. A deep dive into Moorcheh.ai, a platform designed to help teams build scalable, production-grade AI assistants and agents using Retrieval-Augmented Generation (RAG).

TorontoAI — Growth Partner for Startups

The evening opened with an introduction to TorontoAI, a community founded to bridge the gap between developers, startups, and applied AI use cases. With over 6,000 members across platforms, TorontoAI focuses on demo nights, panel discussions, and applied learning — especially for people building and deploying AI systems, not just talking about them.

A special thanks to Shift Labs for hosting the event and supporting the local AI ecosystem by opening their space to the community.

Demo 1: FearSense — Detecting Fear-Mongering with AI

Fear-driven content is everywhere — news, political speech, social media, and even children's content. While sentiment analysis is common, fear detection is a more nuanced and underexplored area, especially when considering its psychological and societal impact.

This demo explored a simple but powerful question: Can AI reliably detect fear-mongering content, and what can we do with that insight?

Falcons AI is a lean, developer-focused AI company whose models consistently rank among the most downloaded on Hugging Face, despite being built by a small, bootstrapped team. Their success comes from solving specific, real problems — not chasing generic AGI narratives.

The FearSense demo leveraged a DistilBERT-based model, fine-tuned specifically for fear-mongering detection. Unlike large LLMs, the model:

• Can be deployed locally or on small cloud instances
• Prioritizes determinism and explainability

The FearSense application was built as a Streamlit app that:

• Accepts YouTube URLs or raw text transcripts
• Scores each chunk for fear intensity
• Visualizes fear peaks and distributions

GitHub: torontoai-hub/fear-monger-detector

Despite some real-world demo friction (cloud tunnels, missing Python dependencies, and live debugging), the audience got a realistic look at what actual AI development looks like — not polished slides, but real systems being deployed and fixed on the fly.

Healthcare Applications

One of the most compelling discussions centered around healthcare applications:

• Correlating fear-heavy media consumption with heart rate or stress data from wearables
• Studying impacts on vulnerable populations, including seniors and children
• Providing researchers with tools, not conclusions — allowing them to explore correlation vs causation responsibly

The key takeaway: This was not a "medical diagnosis" tool, but a research-enabling prototype designed to spark deeper investigation.

Demo 2: Moorcheh.ai — Production-Ready AI Agents

The second half of the evening shifted from model-level demos to production AI systems.

Many teams struggle when moving from a basic prototype to a scalable, accurate, and cost-efficient AI assistant. Common pain points include expensive vector databases and re-rankers, and difficulty exporting prototypes into real applications.

Moorcheh.ai positions itself as an infrastructure abstraction layer for AI assistants — reducing complexity while maintaining performance and accuracy.

DoctorPal AI — A Healthcare Use Case

One featured use case was DoctorPal AI, a healthcare assistant built on top of thousands of pages of medical and nutrition documents.

Using Moorcheh.ai, the team was able to:

• Upload large document sets (PDFs, websites, structured data)
• Automatically chunk, embed, summarize, and index content
• Enforce strict relevance thresholds to prevent hallucinations
• Provide citation-backed responses
• Control which questions the AI is allowed to answer (kiosk mode)

The result: a domain-specific AI assistant that only answers based on verified source material, not general internet knowledge.

Key Platform Features

• Namespace-based knowledge isolation
• Built-in re-ranking and relevance scoring
• Toggleable kiosk mode to block irrelevant questions
• Model flexibility (Claude, LLaMA, Bedrock-native models)
• API-first design for embedding assistants into real products
• Serverless, cloud-native architecture for cost efficiency

A major differentiator highlighted was the ability to export AI assistants into production apps, unlike tools that remain locked inside notebooks or playgrounds.

The demo concluded with an advanced walkthrough showing how Moorcheh.ai can be used to build dynamic AI agents — agents that:

• Dynamically change questions based on user responses
• Use RAG not for documents, but for decision rules and instructions
• Produce structured summaries for human review

The entire workflow — from knowledge base to UI — was assembled in hours, not weeks.

Key Takeaways

• Small, focused models still matter — especially when accuracy, cost, and deployability are critical.
• AI demos should reflect reality: debugging, tradeoffs, and iteration.
• RAG is no longer optional for serious AI products — but it must be done carefully.
• The future isn't just chatbots — it's context-aware, task-driven AI agents.
• Community-driven learning accelerates real innovation far more than polished marketing.

---

Thank you to Falcons.AI, Moorcheh.ai, and everyone who attended and participated in discussions.

See upcoming events →