Overview

Fortune 50 company with dozens of global brands (one of the largest diversified brand portfolios) with ~300,000 employees. Since the 1960’s this company has made a name for itself, and has recently been investing in AI and advanced technologies to stay on top.

Problem Description

• The company portal that services HR, IT, marketing, and branding communications with employees had low user satisfaction scores and required automation to scale.
• Having multiple chat platforms led to low adoption and user confusion.
• Disparate legacy platforms led to employees needing to remember where to go for what, resulting in confusion and the need for increased support staff.
• They had built solutions with 6 different conversational AI vendors and reached the limits with each. They spent a year exhaustively evaluating vendors that wouldn’t end in failed solutions.

Solution Identification

• Informed by feedback from users of the employee portal, they decided to automate portal processes and communications (updating passwords, retrieving 401K materials, etc.)
• They identified that a conversational AI interface would help users navigate to useful information. Use cases spanned from simple FAQ to complex virtual assistants.

Vendor Selection Criteria

• Strategic Platform Approach – After failing to scale with 6 different conversational AI vendors, they realized they needed a platform, not a portfolio of point solutions. They wanted a platform that could handle use cases across the enterprise, enable different teams working together (not just developers), and that supported  iteration and expansion of solutions without heavy technology lifting.
• Train on a single vendor – They prioritized creating a center of excellence (COE) to gain training on a single vendor.
• Omnichannel and multimodality – They needed access to all channels, the ability for a single conversation to span multiple channels, and the ability to design using conversational memory. Voice was a particularly important channel that needed to be supported.
• Citizen development with a No-code platform – They prioritized the ability for their entire organization to create solutions.
• Integrations with existing enterprise tools – They needed something that would connect with all of their current solutions and legacy systems. This included connecting to existing contact center providers via SIP.
• Demonstrated ability to build experiences quickly
• Support wide range of internal use cases – They needed a platform that supports internal use cases as well as contact center solutions with integrations into Microsoft Teams and Salesforce.
• Multiple NLU solutions – They needed flexibility with NLU providers that enabled them to use the best technology  on a case by case basis (e.g. employee asks about product, MSFT returns answer with 90% confidence and Rasa returns answer with 60% confidence, so chatbot responds with MSFT solution).

Vendor Selection Process

Selection Criteria Development Steps

• Issued an RFI with 200 questions
• Issued an RFP with an additional 100 questions
• Meetings focused on a range of topics, from composable architecture, governance, business outcomes, pricing, and usability.

Vendor Selection Finalization Steps

• Conducted Bake-Offs and POC with top 3 vendors
• Co-created solutions with top 3 vendors
• Dozens of people involved, but the decision was made by a small group of experts who took the time to understand the domain rather than just attending sales presentations.

Rollout

Rollout Planning

• They had a total of 29 different chatbot experiences outlined from migration to new bot automation, and used agile scrum based prioritization.
• Used a co-creation and federated training approach (different tracks for learning including things like conversational design, flow builder learning, advanced training for developers, etc.). Enabled users through single sign on.
• Dedicated intern time towards supporting the rollout.

Rollout

• Staged in phases over the course of12 months, updating monthly as new projects surface.
• Started by providing employees with access to training videos for using the OneReach.ai platform.
• Created a business request process for bot creation so that business units could ask for specific bots.
• Migrated existing, legacy chatbots over to OneReach.ai with an increased focus on improving user experience.
• CTO shares progress on the conversational AI rollouts during a monthly town hall meeting.

Current Status

• They have a dedicated conversational AI team with both OneReach.ai and their own resources (15 people total). This team serves as a center of excellence and team of enablers that encourages adoption across the enterprise.
• Conversational AI team shares demos to various business units for continued adoption and planning.