Cognigy.AI v4: Our Most Advanced Conversational AI Platform Yet

8 min read
Philipp Heltewig
Authors name: Philipp Heltewig August 30, 2020

Cognigy is proud to launch the most intuitive-to-use yet most powerful conversational AI tool on the market – Cognigy.AI v4.

The new version advances the simplicity of creating virtual agents across the enterprise, enabling non-technical users, developers and conversational designers to improve their customer experiences faster and more easily.

WATCH RELEASE SHOW

Those new to conversational AI will find it simple to get started with Cognigy.AI. Those experienced in conversational development will find that it ingeniously addresses pain points found in every conversational AI platform on the market today.

We have received very positive feedback from our customers and partners. Andreas Winter, Business Unit Management Enterprise Software Engineering at valantic Financial Services Automation shared their feedback: "With version 4 of Cognigy.AI we will be able to implement high-quality conversation automation projects with our customers even more successfully. The new user interface, the improved NLU or the redesign of the flows consistently reflect the further development of the product for many years."

The four areas of improvement of Cognigy.AI V4

1. The editing experience is now more intuitive than ever

Highlights:

• 100% revamped user interface

• Newly designed Flow Editor

• Pre-built Extensions

• Improved Multi-language support

flow editor cognigy.ai v4

The goal: The driving force behind the re-visioned Cognigy.AI user interface (UI) was to enable users of all skill levels, including non-technical people, to create powerful conversational experiences with ease.

Starting on the login-screen, users will experience a clean look & feel and modernized UX patterns. From there, users familiar with the platform will immediately notice a new layout for conversational flows – merging intent-driven dialog with linear processes, all embodied in a single interface. The Flow Editor is where users create conversations (shown here). As flows can become quite large, we built in pan & zoom, drag & drop, collapsing of branches, and other features to make handling flows simple.

With Cognigy.AI V4, the Cognigy Integration Framework has evolved into Cognigy Extensions. They allow non-technical users to easily extend Cognigy.AI’s capabilities by simply uploading custom Extensions via a web UI. New functions seamlessly blend in with Cognigy.AI’s native capabilities and are just as easy to use. All of our out-of-the-box nodes are based on the exact same framework and written as Cognigy Extensions. As a result, the full set of UI capabilities available to our developers is now available to our customers as well.

Cognigy has always supported bots in multiple languages, with out-of-the-box NLP in 20 different languages and variants. Cognigy.AI V4 takes multi-lingual bots far beyond NLP, allowing for truly multi-lingual dialog management. Flows can now be translated, use fallback languages and share business logic and backend integrations.

The result: Faster to learn, more intuitive to use, enabling more complex use cases in any language.

 

2. The developer experience is now more powerful than ever

Highlights:

• Snapshot capability

• Command Line Interface (CLI)

• Optimized for CI/CD pipelines

• New API generation

new API generation cognigy.ai v4

The goal: To provide a best-in-class tool kit for developers and architects.

With Cognigy.AI V4 we introduce Snapshots, an easy way to bundle everything that makes up a Conversational AI – like Flows, Lexicons, Connections, logic, content, configuration, trained NLU models and even Extensions – and transport it with ease across projects, systems and environments, all sync’d with your release cycles.

Because developers want control on a code level, we are introducing the Cognigy Command Line Interface (CLI). CLI allows users to manage the source code of VAs directly within any IDE (e.g. VS Code, Eclipse, etc.), including Flows, Intents and even the Typescript Code of Transformers. This enables integration in complex CI/CD pipelines and source control like GitHub or Visual Studio Team Services.

What enables the CLI is the completely rebuilt API. It covers 100% of our features, meaning that everything which can be done through the UI can be done through the API. It is REST-based and comes with full OpenAPI support making the API intuitive to use and easy to learn.

The result: A perfect fit for enterprise projects where application development and conversational development walk hand-in-hand.

 

3. The conversational experience is now smarter than ever

Highlights:

• Redesign of the on-board NLP engine, producing outstanding performance benchmarks

• Hierarchical intent mapping

• Direct intent feedback

• Multi-language capabilities

multilanguage experience cognigy.ai v4

The goal: To create a next generation of NLP that leads the market.

Natural Language Understanding (NLU) is a core component in any Conversational AI platform. Essentially, it is what makes a Virtual Agent smart. For Cognigy.AI v4, we’ve rebuilt the NLU engine from the ground up. While developing, we evaluated every step of the process and measured if it contributes to better results. The outcome: A 10-30% improvement of intent recognition accuracy compared to Cognigy.AI v3.

We followed standards within the academic community to benchmark Cognigy.NLU against other vendors. (You can actually go and perform the tests yourself.) The result shows that the accuracy of v4 outperforms the three best-known NLU vendors significantly. And that is the case for small training sets as well as larger training sets.

But we have not only changed the interior of the machine, we also improved working with the NLU capabilities significantly.

The new Hierarchical Intent Transformers allow users to group intents into clusters of a similar category. That has two main effects: 1) it makes it much easier to organize large intent collections 2) it improves accuracy.

Our Direct Intent Feedback provides real-time confidence scoring to business users for every piece of new training data they put into the system. It evaluates the quality of new inputs relative to the existing intent model so that it can guide users proactively through the training. Business users can focus on improving the NLU process and get significantly better results with less work.

The result: Highest accuracy, improved intent recognition, a better end-user experience, even with large intent collections.

 

4. The voice experience is now more accessible than ever

Highlights:

• Out-of-the-box connectivity to SIP and WebRTC voice communication services

• Advanced call management, including transfer to agent, call recording, call disconnect and more

• Transferring of calls to a live agent with full transcripts and context

 

The goal: To dramatically shorten the time-to-ROI for Contact Center Automation.

Cognigy Voice Gateway brings the power and the benefits of automated conversational experiences to phone lines. It plugs into existing contact center solutions and enhances the standard capabilities with our AI-driven technology.

Our Voice Gateway is a pluggable solution that delivers results fast, without the need for re-platforming. Organizations can start where they are, with small but impactful AI support, and gradually expand to broader use cases.

The result: A turnkey, scalable solution for Contact Center Automation, allowing customers to find answers to their questions before reaching a human agent.

Watch our 20-minute release show to learn more!

WATCH RELEASE SHOW

Start a free trial of Cognigy.AI at cognigy.com.

image119-1
image119-1
image119-1