Brief IA

Lorikeet: The Humble AI Transforming Customer Support

🛠️ AI Tools·Tom Levy·

Lorikeet: The Humble AI Transforming Customer Support

Lorikeet: The Humble AI Transforming Customer Support
Key Takeaways
1Lorikeet is developing an AI agent that recognizes its limits and hands off to humans when necessary.
2Lorikeet's dual-agent architecture includes a Concierge to manage tickets and a Coach to optimize the system.
3Integration with Zendesk and Intercom allows Lorikeet to complement these platforms without replacing them.
💡Why it mattersLorikeet is redefining customer support by combining AI and human intervention, providing an adaptable solution for regulated industries.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

Lorikeet's Innovation: An AI That Knows When to Ask for Help

In a world where artificial intelligence is often praised for its ability to solve everything, Lorikeet takes a different approach. The startup aims to create an AI customer support agent that can recognize its limits and request human intervention when necessary. This philosophy of "AI humility" is at the core of their development strategy.

During an episode of "Just Now Possible," Teresa Torres spoke with Jamie Hall, co-founder and CTO of Lorikeet, along with product engineers Xharmagne Carandang and Rona Wang. Their goal is to design an agent that not only resolves customer issues but also knows how to gracefully hand off when the situation demands it.

From Initial Idea to Functional Prototype

The journey to an effective AI agent has not been without its challenges. The Lorikeet team initially explored several unfruitful avenues, such as brainstorming tools and information dashboards. However, it was a startup in the healthcare sector that highlighted the real need: reducing inbox volume. Lorikeet's first prototype was a simple command-line script that provided results in CSV file format.

Today, Lorikeet has evolved to include two distinct agents: the Concierge, which manages customer tickets end-to-end, and the Coach, which assists in configuring and continuously improving the system.

A Dual-Agent Architecture for Optimized Customer Support

Lorikeet's dual-agent architecture is designed to provide optimal customer support. The Concierge is responsible for ticket management, while the Coach focuses on the continuous improvement of the system. This approach allows for a combination of AI strengths with human expertise, ensuring effective problem resolution.

Lorikeet has also developed client-configurable safeguards, allowing customization according to the specific needs of each business. This flexibility is crucial, especially in regulated sectors like cannabis, where requirements can vary significantly. A cannabis company has highlighted the need to revisit their initial approach to safeguards.

Integration with Existing Tools

Rather than seeking to replace established tools like Zendesk and Intercom, Lorikeet integrates seamlessly with them. This strategy enables businesses to leverage Lorikeet's advanced features while continuing to use their preferred platforms.

The evolution of Lorikeet's user experience, shifting from a workflow builder to a conversational interface, illustrates their commitment to making interaction with AI as intuitive as possible. However, even with these advancements, managing an empty chatbox remains a persistent challenge.

"Resolution in the Loop": A Human-AI Collaboration

One of the key concepts introduced by Lorikeet is "resolution in the loop." This method allows human agents to unlock the AI without taking full control of a ticket. This ensures that the AI continues to learn and improve while providing human assistance when needed.

Domain-specific safeguards are another essential aspect of Lorikeet's approach. For instance, a cannabis company highlighted the need to customize these safeguards to meet unique regulatory requirements. Clients can set their own evaluations and safeguards via the Coach interface, allowing them to tailor the system to their specific needs.

Diagnosis and Learning Culture

Lorikeet also utilizes AI to diagnose failure modes in logs and automatically suggest corrections. This self-correction capability is a major asset for continuously improving the system.

At Lorikeet, learning is at the heart of the product engineering culture. Every engineer is encouraged to learn from customer interactions, fueling a continuous improvement loop. Each week, engineers ask themselves what they have learned from a customer, thereby reinforcing the innovation and adaptability of their solutions.

In conclusion, Lorikeet is not just developing a high-performing AI; it is redefining how businesses interact with their customers by combining advanced technology with human intervention for exceptional customer service.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.