Building a Cognitive Center of Excellence: Part 2

6 minute read

Cognitive technologies promise to be a transformative force for enterprise, but it also requires a new way of thinking about business.

A little over a year ago, I wrote a blog post about how to establish a Cognitive Center of Excellence (CoE), a working organization within a business focused on driving cognitive technology adoption, to support the deployment of Amelia. The post explored lessons learned based on observations of early customer experiences implementing Amelia and other supporting technologies. The blog went into organizational aspects, team structure, skill-sets, and processes we recommend, as well as the available tools and templates to assist customers in getting started on the right path.

Fast forward to today and the basics remain the same, but there are several new recommendations worth exploring. The good news is that, in general, things have become much easier. However, some areas also require additional sophistication and preparation to get the most out of your CoE.

One of biggest changes since last year was the introduction of Amelia v3, which greatly expands the scope of her capabilities and the potential for business impact. Before getting into new CoE strategies, let’s explore what’s changed in the past 12 months.

What has changed with Amelia and our customers?

Simplified learning: With Amelia v3, our focus was ease-of-use, especially around automated learning. Our expectation is to eventually have 60-80% of the teaching done by business Subject Matter Experts (SMEs), rather than cognitive engineers who would perform this task with past Amelia versions. This is possible because Amelia has more “awareness” of her own knowledge and performance, which helps the implementation teams pinpoint weaknesses and where additional training is needed.

Increased out-of-the-box knowledge: Over the past year, our teams have continuously expanded the volume of built-in skills, capabilities, and knowledge, so Amelia can begin working for our clients right away.

Sophisticated use cases: As Amelia gets smarter and customers become increasingly adept in training her, there’s a growing list of elaborate use cases, such as integrating her directly into the fabric of the digital eco-system to drive business transformation, or using her new predictive analytics module (which in turn necessitates more complex skills in implementation teams or the CoE).

Scale of deployments: Finally, just as use cases have increased in sophistication, our production deployments are getting larger. With expanded scope (one client has trained Amelia in more than 800 processes), it has become increasingly in our customers’ interest to take control of the entire lifecycle themselves in order reduce the dependence upon IPsoft.

The journey toward the ideal Center of Excellence

Sophistication and scale, as mentioned above, are perhaps the most important drivers to establishing a mature, structured CoE. However, as in all aspects of business, one size does not necessarily fit all.

The graphic below visually captures these elements to aid our customers and partners with designing CoEs.  We’ve divided the functional aspect of the CoE and delivery teams into four major categories, and the time dimension is depicted by initial pilot project (blue), a transition step (mixed colors), and finally when the CoE becomes fully operational (purple).

Initially, IPsoft – or one of our implementation partners – owns most of the functional areas. By the transition phase, customers start building up their expertise and begin taking control. Then finally, most areas are handed off to the client. Here are some of my recommendations across the functions:

Project control: This function is core to the CoE’s raison d’etre and incorporates project management, steering committee oversight, business cases, and roadmaps etc. IPsoft and our partners generally control the pilot phase in order to guide use-cases and priorities, but ideally control is transitioned to the customer shortly thereafter in order to align with business priorities and strategic goals.

Content creation: Content creation is the main role of the implementation teams. Initially, customers lean on IPsoft or our partners for this function, but as teams become proficient in the core aspects of Amelia, they are freed to take over more of these roles, as outlined below. Following the initial phase, some CoEs may still require assistance around linguistics, analytics, and journey design. Some customers may choose to maintain a cognitive engineer as a supplement to their own staff or as a means to preserve a direct connection with IPsoft for continuous knowledge transfer.

Dev/Ops: As Amelia learns skills, increases integrations, and takes on new roles in the organization, the lifecycle of the application components begins to require an integrated dev/ops mentality. In the beginning, this is obviously done by our teams and in some cases, we will maintain that role. However, most customers will want to take full ownership of the content lifecycle and only tap our team for assistance around major Amelia upgrades.

Support: Support is closely related to monitoring. If IPsoft is responsible for monitoring your Amelia, we will likely continue to also provide support. However, in instances where the monitoring of Amelia is performed by an internal IT Operations team, we will help train them to be fully capable. In all instances, we remain closely connected with the internal team to ensure service levels are met.

Having a plan, measuring progress, and adjusting expectations are important traits of managing the creation and evolution of a Cognitive CoE. This is – and will likely remain – an evolving process for quite some time.

People and skills changes

During the past 12 months, we’ve seen changes to the overall composition of Cognitive CoE and implementation teams, which are training Amelia on a day-to-day basis. Many of these changes are focused on designing the right customer or employee journey.

  • Conversation experience designers: Some of our linguists have become “conversational experience designers.” This is not just a fancy title upgrade (we still use linguists as part of the Amelia localization efforts). However, when implementing and training Amelia, we’ve found the overall conversational journey is closer to a design process than a traditional linguistic exercise. Depending on the size and nature of the project, a conversational experience designer can be a part of the CoE working with multiple teams or part of an individual implementation team.
  • Data scientists: Data scientists have also become more frequent in our customers’ CoE. With Amelia’s new Predictive Analytics module, real-time collection of data-points can be run against predictive models, which guide Amelia’s behavior and allow her to present analytics back to a customer or employee for informed decision making. We expect this role in the CoE to continue to grow.
  • UX designers: UX designers have become important parts of the process as well. While not usually part of the cognitive CoE, these designers are a shared resource working closely with the implementation teams and journey designers to develop the visual components which help guide the end user through the process.
  • QA engineers: QA engineers remain vital, especially in implementation teams when developing new content. However, over the next few months, we will introduce enhancements and tools to make QA a part of the dev/ops process and much more automated via cloud conversational testing, so stay tuned.
Processes and Tools changes

Agile still rules from an implementation process standpoint, however our governance framework has been enhanced in several areas to incorporate learnings from multiple large projects. This includes fortnightly steering committee meetings – in addition to the weekly project status meetings – with clear agendas and reports to drive decisions around reducing risks and improving velocity of the implementation teams. We have also improved the structure and agendas around team breakout sessions to better guide participants.

From a tool perspective, we have internally made changes to facilitate improved agility and communication. We’ve employed a more prescriptive use of documentation related to our projects (in our case, we’ve found Confluence to be a superior model for sharing information); we’ve standardized document sharing with OneDrive; and finally we changed our project, task, and resource management to SmartSheets. These changes have improved, simplified, and standardized our internal communication and meeting systems. It’s not important what tool you select, just make sure you have tools to support effective collaboration and communication.

What’s next?

Cognitive Centers of Excellence don’t happen by themselves. They need to be planned, budgeted, and maintained. They will evolve over time as the technology and use-cases change. We recommend a staged approach to the creation of your CoE.

The first stage or level, is a “seed” level. This means it doesn’t require a huge investment with lots of staff, but just enough to have a presence within your business. We recommend establishing this shortly after the first Amelia production use cases.  Next step (Level 2) focuses more on the strategic control of the projects, which can begin about six months after the initial seed. The final step (Level 3) is where the CoE assumes operational control, with knowledge transfer and team build-up commencing after another six to nine months.

Cognitive technologies promise to be a transformative force for enterprise, but it also requires a new way of thinking about business. There probably isn’t one best way to go about this process, but there are strategies that have proven successful – and we know establishing a Cognitive CoE is among the most effective. Through posts like this, we want to share learnings and best practices that can help your company incorporate cognitive abilities in the most efficient manner possible. As always, we welcome any comments, feedback and experiences from the AI pioneers implementing these technologies, or questions about a Cognitive CoE.

Previous Next

The Intelligent Contact Center

Companies have spent decades implementing Interactive Voice Response (IVR) systems in their call and customer care centers, but they've proven unable to keep up with customers' expectations.

In this white paper, we examine the benefits of an Intelligent Contact Center, where companies utilize Conversational AI-powered virtual agents to provide first-line resolution and support for customers, and augment human employees through AI and automation.

Download our paper to learn the benefits of this approach and why current IVR systems simply will not cut it in today’s hyper-paced digital landscape.

Learn More