A Three-Step Process to Evaluate the Right Conversational AI Platform

Person using a laptop computer from above

261

Take a behind-the-scenes look at the RFQ process of a Fortune 500 company that is ranked in the top 50 globally and reached over 90% intent recognition in the demo phase.

We show what enterprises demand from their technology providers and how they set up their RFQ.

In this case, we’re looking at one of the world’s largest companies from the finance sector who has set up its selection process so precisely that it could unmask any vendor who was unfit to accompany it on its journey towards success.

Step 1: Identifying the Main Goals

Identifying their goals early on made setting up the process easier and helped them to stay focused during the RFQ process. The mapped scenario is short but precise.

Scope:

90 million calls p.a.

Main Goals:

Automation of 40 million calls (45%)
Speed up service time
Radical cost-reduction

Timeframe:

Reach 45% call automation within two years

Step 2: Creating Must-Have Requirements

Knowing what the achievement should be is a great start, but the second step is to find out what to compare. The above goals can only be met if customers accept and even enjoy the conversational AI experience.

A great user experience is always backed by mature functionalities that guide the user naturally and can solve their requests. Additionally, the flexibility of the platform and a global understanding are very important to be able to adapt easily to new circumstances.

The list of requirements was boiled down to five main categories which were investigated thoroughly.

Compared Categories

Enterprise Ready

Agnostic: compatible with multiple NLU and speech providers
Scalable
GDPR compliant
Training materials & programs

Product Usability & Flexibility

Low-code and feature-rich
Off-the-shelf integrations with other platforms
On-prem support
Self-service-ready to empower business units to maintain bots with minimal tech involvement

Future-Proof

Solution & product roadmap
Cloud-native
Mature provider with a strong customer base
Extensive SLAs
Ability to build AI models in-house
Unsupervised bot-learning

Global Reach

Multiple hosting options
Network of global service integrators
Meet local language and legal requirements

Voice Expertise

Extensive voice (not just chat) experience
Telephony and contact center integration
Large NLU & ASR libraries

Step 3: Validating Through PoC

But having this checklist filled with ticks does not mean to have made it to the end. A demo agent should serve as the final basis for the decision and be able to identify 90% of the callers’ intents and serve or route them correctly.

Cognigy’s consultants started building the demo together with our customer and were in constant exchange every step of the way. This process was able to prove that the technology perfectly suited the needs of the customer.

The demo voice bot identified 92%-97% of the intents!

The result was a mature voice agent with countless features, ready to take customers’ calls.