Back

AI Assistants: Right now, where to start and what’s coming next

AI Assistants right now - where to start and what's coming next?

As part of Firebrand Talent’s #Digitalks event series, Lisa Bouari, Executive Director at OutThought, recently spoke to audiences on the topic ‘AI Assistants: How to take your business to the next level’. The key areas Lisa focused on are detailed below.

What the market is saying

I’m not sure about you, but I wasn’t surprised to hear the latest stats from Gartner suggesting that 25% of organisations will use Chatbots in their customer service by 2020.  It’s a large number when we consider the thousands of organisations that are yet to adopt this technology – but the take-up rate and activity in the marketplace would suggest we are well on our way. IBM states that 70% of consumers also prefer to message over calling, for customer support.

In the latest State of Chatbots report, users said that the benefits they would most expect to enjoy from Chatbots were 24 hour service, and instant responses and answers to simple questions.

Fig: State of Chatbots, Drift (2018)

Of course, the blockers for those same users were that they had concerns around its ability to help them as effectively as a real-life assistant and the accuracy of results.

Fig: State of Chatbots, Drift (2018)

There is no doubt though, that the business impacts for some of those who have adopted this technology are compelling.

  • The Royal Bank of Scotland has achieved 84% attainment rates (bot handled questions) after a year of their Virtual Assistant, ‘Cora’ going live.
  • Swedbank is handling 40 000 requests per month and experiencing similar attainment rates of 81%.
  • Autodesk has cut their Tier 1 service cost range from $15-$200 per enquiry down to $1 when handled by their Virtual Assistant.
  • Sephora has increased makeover bookings by 11% through their Virtual Assistant which leads to direct product up-sell revenue.

Despite the compelling numbers and success stories in the market, however, organisations are still somewhat sceptical.  The majority of Virtual Assistant projects are being started as trials or as a ‘test’ for Virtual Assistant capabilities,  approaching them on a project by project basis rather than in a strategic and transformational way, which is what they require to ultimately be successful.

Where are we at, REALLY? Hype vs Reality

I thought I’d take a look at the traditional Gartner Hype Cycle curve, and adapt it for Virtual Assistants and where the opportunities lie. For those who haven’t seen this before – each year, Gartner plots a range of emerging and more recently established technologies on a curve that represents consumers expectations of these technologies and their position in the cycle.

It is a powerful visual to show how technology travels along this curve as first use cases are established, and sometimes fail, investment strengthens technology offerings and their gradual uptake turns them into finally being productive for organisations.

In 2016, Virtual Assistants were barely on the curve as emerging and just a few years later we had witnessed the first major public failures of this technology (remember Microsoft Tay?).  We are finally starting to see real results in business.

Fig: The Virtual Assistant Hype Cycle

When we are at the ‘Peak of Inflated Expectations,’ this is a great time to step back, slow down and consider what is REALLY possible with this technology.  It is an opportunity for us to consider the best and most realistic use cases, take the hype away and stop kidding ourselves about what is possible.

The ‘Start Competing’ bubble represents the time where we start to become truly productive with the technology, and that we are now at a point where we should take advantage of innovation. For Virtual Assistants, this might mean combining them with other technology like Robotic Process Automation, more integrations with enterprise applications and so on.

So Virtual Assistants in 2019 are starting to become truly productive for business, are still suffering some of the hype backlash of the last few years but the opportunity is now to take advantages of the proven use cases and get ahead of your competition.  For those who have already been using these use cases productively, it’s time to start thinking about how to take them to the next level.

4 Virtual Assistants to start with right now

Frequently Asked Question Assistants

These assistants are the most commonly adopted assistants because of their ability to automate large volumes of repeating customer enquiries.

Organisations that have a high volume of customer enquiries, or internal inquiries for that matter, are ripe for taking advantage of these assistants.  These assistants usually answer questions online, or provide links to a website or other knowledge base material and are suited to almost any industry or business.

Examples might also include walking a user through a commonly requested process such as password reset, or scheduling a call back from the organisation if there isn’t a contact centre to pass to.

Fig: Ceba, Commonwealth Bank Assistant.

Triage Assistants

Triage assistants are different and I think of these as the Virtual Assistants that ask you more questions than you ask them.  The idea of triage assistants is that they can direct your user to take different actions based on what their inputs are.

For organisations, they can take advantage of these by asking questions of their customers, learn about those customers in the process and recommend a next step, registration, product download, account application process or so on.  Triage assistants can also be used to understand what your customer is after and route them to the right staff member to assist.

Lead Generation

Lead Generation assistants are all about catching, and keeping your prospective leads in your digital sales funnel. Lead Generation assistants capture information about your client that you wouldn’t have been able to if they just arrived to browse your website, and having a mechanism to get back to them with a good understanding about what it is they are looking for.

A great example is this property one which allows me to search for property, collects lots of information about what I’m looking for and allows the organisation to connect with me as a warm lead. 

Fig: Keyes, Property Assistant.

E-Commerce

E-Commerce Virtual Assistants are all about efficiently driving a user to make purchases inside your Virtual Assistant. These virtual assistants often need to be connected to other systems such as payment gateways to facilitate transactions however they are a powerful way to streamline purchases and capture decision points about your customers.

Where are AI Assistants going?

To many, all Virtual assistants seem the same. They communicate back and forward, and they allow a user to select options or type responses. OutThought.co have created a Maturity Model specific to Virtual Assistants.

Level 1 – Basic and controlled input

Level 1 best represents the first chatbots that existed. Rigid, limiting but functional in terms of responding to user button selections for example. These chatbots are cost-effective to develop and still can provide business value if used in the right context as they can quickly direct a customer, capture inputs and provide utility for the right business case.

Level 2 – FAQ & free text

Frequently Asked Question bots are the most commonly implemented assistants today, and this comes with a new set of challenges when we allow users to enter free text into an input box.

Opening this utility to customers means an ability needs to exist to interpret and handle and respond to that request. These Virtual Assistants usually use technology such as NLP (also in level 3) and may be embedded in a website providing dynamic page linking behind the Virtual Assistant to relevant information pages or links.

Most organisations Assistants are currently at Level2 and some may have integration points to other live agent systems for handover or basic integration to applications (level 4).

Fig: OutThought Virtual Assistant Maturity Model.

Level 3 – NLP and Context

Context is what makes a Virtual Assistant’s conversational abilities impressive. Take a quick read of ‘Level 3’ above. The user asks ‘even for my boat?’ mid-way through the conversation.  In isolation, this question would not normally be understood by a Virtual Assistant because the Virtual Assistant does not understand that this question is relative to the ‘context’ of making an insurance claim.  Context is what allows Virtual Assistants to naturally flow through a conversation even when the users input twists and turns in unexpected ways.

It is notoriously difficult to design for across an entire Virtual Assistant for a number of reasons, but the technology now exists and organisations will have to decide at what cost, and how engineered they apply this.

Level 4 –  Personalised and Applications

While it is possible that some Level 2 Virtual Assistants might relate to users by name using authentication parameters and such or have some basic integration points – Level 4 assistants are where we will start to interact with assistants that can effortlessly include information about us from enterprise systems in the organisation.

The example above would return the users vehicle types and ask the vehicle which one they would like to make a claim for.  It would also collect the claim information and store this in the CRM or other enterprise systems against the customers record, starting the actual claims process within the Virtual Assistant.

While this is strictly quite possible now, it isn’t the first capability that organisations are deploying and when combined with the preceding features, we start to see an impressive and truly engaging conversational assistant.

Level 5 –  VA + RPA + Assistant Family

Level 5 Virtual Assistants are not too far into the future. These Assistants will start to combine simple business processes using robotic process automation such as combining linked customer data from separate sources, executing a create claim request or other, and be able to take, and send information to the Virtual Assistant to complete a user’s request or process, end to end for multiple processes.

It is also likely by this time, that organisations won’t have just one Virtual Assistant in customer service. Organisations will have several Virtual Assistants with different roles and capabilities that are all aware of each other and where to route users based on which assistant will serve them best.

Final Thoughts

Organisations who have adopted Virtual Assistants in the past few years, are reaping the rewards and benefits, and most importantly, the learnings and best practice from these rapidly evolving technologies. These same organisations are differentiating themselves from their competition with their ability to streamline customer service, generate revenue, maximise cost savings and engage their customers and employees more effectively.

Virtual Assistants, when approached in the right way, can dramatically transform and benefit organisations. They should be approached in a strategic way, focussing on them as a holistic solution, touching many parts of the business and often benefiting new areas with the new information and customer rich data that is collected.

Back

Subscribe To Firebrand Ideas Ignition Blog

Sign up to receive our new blog posts via email. You'll get all the latest articles on digital, marketing, social media, communications, personal branding — and lots of career advice

Search Website