How AskVet is Transforming Veterinary Care with AI and RAG Technology

Dave Kearney, CTO of AskVet, joined a recent episode of RAG Masters to discuss AskVet’s AI adoption journey and unpack the technical innovations that made it possible.

Watch the introduction from the episode introducing AskVet and Kearney's AI explorations and background:

The Evolution of AI in Pet Care

Kearney recalls the company's early beginnings, having been founded in January 2014: 

"AskVet has been around for about 10 years. We've done an amazing job of connecting veterinarians with pet owners who want to get their medical questions answered."

But AskVet didn't stop at simple connections. They saw the potential in AI early on. 

"We were an early adopter of artificial intelligence," he notes, "using tools to deliver capabilities like Amazon Alexa and Google chat for answering pet health questions."

This foresight laid the groundwork for a technological leap that would set AskVet apart.

RAG: The Game-Changer

The real breakthrough came with the implementation of RAG technology. But before we dive deeper we’ll talk about what RAG is, and how it works in the context of veterinary care.

How RAG Works in Veterinary AI:

1. Information Retrieval: The system accesses a vast database of veterinary knowledge.
2. Context Understanding: It interprets the user's query and relevant context.
3. ‍Answer Generation: Using retrieved information, it generates a relevant, accurate response.
4. ‍Continuous Learning: The system improves over time through feedback and new data.

Kearney emphasizes the impact: "It wasn't until our VERA came along and we integrated large language models with RAG that things really started to accelerate for us."

This technological advancement allowed AskVet to handle a significant portion of customer inquiries without human intervention.

"Now we can answer questions with the large language model, without involving an actual veterinarian."

Here's some other advice from Kearney about implementing an effective system:

The 'Mood Bucket' System: Customizing AI Responses for Audiences

One of VERA’s most innovative features is its 'mood bucket' system for customizing AI responses depending on the audience it speaks with.

Here's a simplified look at how it could work, illustrated by a brief python code example:

def get_response(user_type, query):
    if user_type == 'pet_owner':
        empathy_level = 'high'
        info_depth = 'moderate'
    elif user_type == 'veterinarian':
        empathy_level = 'low'
        info_depth = 'high'
    
    return generate_response(query, empathy_level, info_depth)

This system recognizes the varying needs of different users.

Kearney explains, "We divide into two empathy levels between the pet owner and the professional. The professional wants just the information they're asking for, whereas the pet owner responds better to an empathetic entity."

By integrating this system directly with RAG, VERA is able to tailor its responses as effectively as it retrieves the accurate information. This combination ensures the user has a personalized experience that is custom-fitted to their specific needs.

Here's Kearney describing the technology behind VERA in more detail:

VERA Pro: AI Assistant for Vets

Building on their consumer-facing AI, AskVet developed VERA Pro for veterinary professionals. "It's a way for vets to quickly access information and answer questions during appointments," Kearney describes.

A Day in the Life with VERA Pro:

As Kearney walked through the benefits of VERA Pro, he discussed how a typical day might go as it works alongside a veterinary team to streamline everyday processes and improve the effectiveness of each team member.

The VERA Pro system gives answers to questions very quickly while delivering handouts and summaries to people in the office, such as providing information to the pet owner that they need to walk home after an appointment. The veterinarian has complete control of the information they can then share with the pet owner.

Here’s an example of areas the VERA Pro system is designed to support on a daily basis:

1. Morning: Vet reviews patient histories, AI summarizes key points.
2. Consultations: AI provides real-time information during appointments.
3. Documentation: AI assists in generating detailed medical notes.
4. After Hours: Reduced paperwork thanks to AI-assisted documentation.

Challenges and Future Implications

While RAG has proven powerful, it's not without challenges. Data privacy, especially with sensitive medical information, remains a top concern. Additionally, ensuring the AI doesn't overstep its boundaries and knows when to defer to human expertise is crucial.

As Kearney describes, the potential applications of this technology are vast. From healthcare to finance, legal to education, any field that relies on accessing and interpreting large amounts of specialized information could benefit from similar AI implementations.

For developers interested in exploring RAG technology, similar implementations as AskVet can be accomplished using a combination of technology like EyeLevel’s GroundX APIs or No-Code tools, along with open-source libraries and database technology. The key is to focus on high-quality, domain-specific data and to continuously refine retrieval and generation processes.

Conclusion

AskVet has transformed from a pet telemedicine service to an advanced implementer of AI techniques that power their VERA system. Their innovative use of RAG technology, combined with their unique datasets and implementations, provide a blueprint for how AI can be effectively implemented in niche markets.

You can read more about AskVet's success with implementing RAG technology here.

As many industries move forward, the challenge for developers will be to create AI systems that not only automate tasks but also enhance the quality of professional services. There are many opportunities for those who can harness the power of RAG and similar AI technologies.

Ready to explore RAG for your industry? Start by identifying your domain-specific data sources and considering how AI could augment human expertise in your field. The next AI success story could be yours.

You can watch the full episode of RAG Masters below: