Amazon Lex V2 Now Has A Visual Conversation Builder!
Amazon Lex V2 has received a significant upgrade with a Visual Conversation Builder being introduced. This update addresses a material vulnerability in Lex, the absence of a no-code flow designer and builder.
The TL;DR
- On 17 August 2022 Amazon released an additional avenue to build and manage conversations via a visual dialog builder.
- This update is aimed at improving Lex’s chances to make it into the leading quadrant of Gartner for Enterprise Conversational AI Platforms.
- Currently Lex lags in the bottom quadrant, lacking in completeness of vision and ability to execute.
- The caution from Gartner centred around lack of equivalent capabilities and users being vulnerable to cost and added complexity having to leverage additional AWS services. This deployment remedies the Gartner caution to some degree.
- This development is designed to put Lex on a path to exit the niche player status and enter challenger status.
- Lex V2 can now accommodate increased complexity whilst remaining no-code. While Lambda services are still available as a pro-code route.
- This development moves Lex V2 away from being a development environment for command and control applications only, or just an order fulfilment tool; and becoming a conversation design interface.
Please follow me on LinkedIn for updates on Conversational AI and more. 🙂
Below you can see where Lex finds itself currently in terms of the Gartner Magic Quadrant.
Amazon Lex V2 now has three levels of application development. The intent editor can be used to create a basic order fulfilment conversational interface, or a command & control speech interface.
The visual conversational builder can be leveraged for more complex multi-turn conversations. With this visual dialog flow builder Lex V2 is aligning itself with the Gartner leaders.
Lastly, Lambda code can be used in isolation to facilitate conversation state management, or it can be used as an extension to the Visual Conversation Builder.
Assessment Of Amazon Lex V2
Improvements
1️⃣ In a previous article I listed two points of criticism of Lex, the first was that the Lex Editor is not interactive (intent editor). It reminded me of the Oracle Digital Assistant canvas, and how Nuance Mix implemented their dialog design canvas. The nodes are not fully interactive, in the sense of dragging and dropping elements.
2️⃣ I also previously pointed out that it does feel a bit debilitating to not be able to drag, drop and manipulate nodes in the canvas by moving them as this is a natural inclination, to want to directly interact with the design canvas. The visual conversation builder remedies these ailments to a large degree.
3️⃣ The fact that the Intent editor remains largely intact means that users have that flexibility to continue with current ways of work, or migrate to the new environment.
4️⃣ Below a view of the design canvas, which is extremely intuitive and responsive.
5️⃣ The approach of Lex is still centred around intents, and each intent has a flow attached to it. Digression is made possible via the Go to intent node.
6️⃣ Lex has very much the same approach as Nuance Mix with a limited amount of conversational nodes, which currently stands at 10 as seen below. Users can drill down into settings for each node, this top-down approach in complexity reminds of Nuance Mix.
Please follow me on LinkedIn for updates on Conversational AI and more. 🙂
Considerations
1️⃣ It would be really helpful if pre-sets or templates existed for extending from the no-code Visual Builder into a pro-code Lambda environment. For example, Cisco MindMeld does a great job of code block pre-sets and templates for easy adoption of the pro-code way of work.
2️⃣ Lex has not added any structure to their intents as we have seen with Cognigy, Kore AI, HumanFirst and a few others. Specifically considering hierarchical or nested intents.
3️⃣ Collaboration in flow design and development is key, and the extent to which collaboration between multiple uses are possible remains to be seen, Voiceflow is most probably the leader in this department.
4️⃣ The interface seems simplistic, keeping in mind that this is a first iteration.
5️⃣ The build process feels long for small prototype applications.
In Conclusion
Amazon Lex V2 seems to be pursuing parity with the Gartner leaders, with differentiation and capturing market share still looming ahead.
However, this development does make the decision easier for AWS incumbent users to justify opting to make use of Lex.
The two biggest challenges or tests for the Lex Visual Conversation Builder will be scaling and collaboration.
This development does beg the question, could Amazon not have done more to differentiate Lex with market leading innovation?
Please follow me on LinkedIn for updates on Conversational AI and more. 🙂
I’m currently the Chief Evangelist @ HumanFirst. I explore and write about all things at the intersection of AI and language; ranging from LLMs, Chatbots, Voicebots, Development Frameworks, Data-Centric latent spaces and more.
