Introducing AI Engine 2.0: Faster, Smarter and More Reliable
Alexander Wijninga
| 3 min
Today we’re launching AI Engine 2.0, the biggest upgrade to Watermelon Agents since the introduction of Generative AI.
This isn’t just a model update.
We’ve upgraded the entire AI engine behind every Agent. From knowledge retrieval and search technology to crawling, transparency, tool selection and response speed.
The result is simple:
Better answers
Faster responses
Smarter conversations
More reliable automation
What has changed?
Most AI platforms improve when a new model becomes available.
With AI Engine 2.0, we’ve gone much further.
We’ve upgraded:
AI models
Search technology
Knowledge retrieval
Web crawling
Tool selection
Source transparency
Performance
Together, these improvements help your Agent find the right information faster and answer with greater accuracy.
Up to 70% faster
One of the most noticeable improvements is speed.
The average time until the first response has decreased from approximately 10 seconds to 3 seconds.
Overall response times have been reduced by an average of 7 seconds.
This creates a significantly smoother experience for both customers and support teams.
Better answers through Hybrid Search
Finding the right information is one of the most important parts of any AI Agent.
Previously, Agents relied primarily on semantic search.
With AI Engine 2.0, we’ve introduced Hybrid Search.
This combines multiple search techniques to improve retrieval quality.
When a customer asks a question, the Agent can now:
Search based on meaning
Recognize important entities such as product names and SKU numbers
Re-rank results to prioritize the most relevant sources
The result is better source selection and more accurate answers.
Better performance for large product catalogs
Many companies work with hundreds or even thousands of products.
For example:
Product variants
SKU numbers
Technical specifications
Large inventories
A common challenge for AI systems is distinguishing between highly similar products.
With AI Engine 2.0, Agents are significantly better at recognizing specific products and matching them with the correct information.
This reduces confusion and helps prevent inaccurate responses.
Improved knowledge retrieval
Retrieval Augmented Generation (RAG) is the technology that allows an Agent to search through your knowledge before generating an answer.
AI Engine 2.0 improves every part of this process.
Agents now:
Retrieve more relevant documents
Use more context
Select better sources
Base answers more strongly on available knowledge
This leads to more reliable responses and fewer hallucinations.
Smarter conversations
Good automation isn’t just about giving answers.
It’s also about knowing when additional information is needed.
With AI Engine 2.0, Agents are more likely to ask targeted follow-up questions when information is missing.
For example:
Instead of immediately answering:
“Can I still change my order?”
The Agent may first ask:
“Could you share your order number?”
This helps improve accuracy and creates a better customer experience.
A better web crawler
We’ve also rebuilt our web crawler.
The crawler now processes websites, documentation, knowledge bases and product feeds more accurately.
While crawling itself isn’t necessarily faster, the quality of the collected knowledge has improved significantly.
This allows Agents to work with cleaner and more complete information.
More transparency
Understanding why an Agent provides a particular answer is important.
AI Engine 2.0 introduces better visibility into source usage within the Tester.
Users can now see more relevant sources and gain better insight into how answers are generated.
This makes it easier to optimize knowledge and build trust in automation.
New models available
AI Engine 2.0 introduces support for:
GPT-5.4 Nano
GPT-5.4 Mini
Both models are optimized for speed, reliability and customer support use cases.
For most customers, we recommend switching to one of these models to take full advantage of AI Engine 2.0.
What should you do next?
To get the best results from AI Engine 2.0, we recommend the following steps:
1. Re-crawl your content
Re-crawl:
Your website
Documentation
Knowledge base
Product feed
This ensures all knowledge is processed using the latest crawler.
2. Switch to GPT-5.4 Nano or GPT-5.4 Mini
These models are optimized for the new AI engine.
3. Re-test your conversations
Use the Playground to test your most important customer scenarios again.
Many customers immediately notice improvements in:
Speed
Accuracy
Relevance
Automation rates
Available now
AI Engine 2.0 is available today for all Watermelon customers.
We’re excited to see what you’ll build with it.
Ready to experience the difference?
Log in to Watermelon and activate AI Engine 2.0 today.
Related articles
Product updates
Introducing: Watermelon Pulse!
Why LLMs are important for chatbots: LLMs (Large Language Models; the future to significantly improve chatbots, making them more intelligent, efficient and user-friendly.
Product updates
Introducing the Next-Gen Website Widget
In the ever-evolving landscape of customer engagement, keeping up with technological advancements is crucial. Today, we are thrilled to introduce our revamped Website Widget