Scaling Meaningful Connections in a Rapidly Growing Ecosystem

As ConnectAmericas' business community expanded, so did the need to improve the quality, relevance, and scalability of business connections. Manual partner searches were inefficient, and many potential opportunities went unrealized due to the lack of intelligent tools.The team was also looking to scale their matchmaking model beyond one-on-one meetings at in-person events, enabling virtual and always-available business connections.The IDB set out to transform ConnectAmericas from a business network into a true smart matchmaking platform: precise, personalized, and centered on user experience.

‍

‍

Introducing ADA: AI-Driven Matchmaking with a Human Touch

Onetree was selected as the technology partner to design and develop ADA, an AI-powered business matchmaking solution. The platform was built as a minimum viable product (MVP) in approximately three months, in close collaboration with the Microsoft AI Lab in Montevideo, Uruguay—one of only five global Microsoft Labs dedicated to accelerating AI innovation. This partnership included architectural design, model selection, and an in-person sprint to produce the initial working version.

We envisioned ADA as more than just an interface—it was shaped as a conversational AI assistant, designed with a human-centric and goal-oriented approach. Instead of relying on long forms or static filters, ADA engages users through progressive, natural-language dialogue that uncovers business goals, interests, and priorities in a more intuitive and fluid way.The result was a modern AI architecture powered by vector search and semantic processing, offering much more relevant and efficient results than traditional SQL-based keyword searches.

Key features include:

• Personalized matchmaking algorithms based on sector, location, company size, and specific business needs.
• Conversational interface powered by generative AI, adapting responses based on user intent and context.
• Affinity scoring and match probability, helping prioritize the most relevant connections.
• Continuous feedback module, allowing users to rate suggestions and improve the system over time.
• Custom dashboards and analytics, giving companies actionable insights to optimize their networking strategies.

From a technical standpoint, ADA includes:

• Vectorized company profiles enabling semantic search.
• Normalization and transformation of structured and unstructured fields into AI-optimized descriptions.
• Multilingual capabilities (Spanish, English, Portuguese, French), with responses tailored to each user’s preferred language.
• Three distinct search modes: suppliers, clients, and strategic alliances.
• Justified responses in natural language with curated business cards verified by the IDB.
• Directus was used as both the CMS and backend, providing a flexible, headless infrastructure to manage content and structure business data efficiently.

While ADA uses foundational models like OpenAI, the solution’s real differentiator lies in the structured data prep, prompt design, fine-tuned context, and constant feedback loop—making it far more than a simple "out-of-the-box" AI integration.