At Troopr Labs, we are imagining a future where a Smart Assistant (aka Multi Purpose Chatbot) for workplace, powered by our conversational AI platform and deep integrations with business applications, is at the center of every digital workplace. Alexa and Google Assistant have shown us the potential of a personal Assistant at home. Imagine a workplace assistant that is able to answer your work-related questions, perform tasks on your behalf, brief you periodically on what needs your attention, recommend follow-up actions at every step and allows you to teach it new skills for your unique needs. The potential of this future is what drives us everyday.
Why Workplace Assistant?
Mundane work problem: According to a recent McKinsey report, employees spend 6 hours a week in repetitive tasks like data collection, approvals and updates. A smart workplace assistant can help automate many of these tasks.
Context switching problem: We are in the era of SaaS apps and organizations across the world are taking a best of breed approach to workplace IT needs. A typical employee is engaging with more than 4+ apps at workplace regularly*. While this is generally proven good for business, 70% knowledge workers say app switching is affecting productivity*. A workplace assistant can alleviate this problem with a unified smart interface to all workplace apps.
Cluttered communication problem: Historically emails became less useful when the admin work to declutter and find actionable information became too expensive. The spectacular growth of business chat platforms and the proliferation of chatbots are bringing with them similar problems to the chat world. Workplace assistant can address the problem with the potential to
Prioritize and streamline notification from across apps/botsConsolidate fragmented action items across apps/botsAdd context to data from across apps/bots
Key challenge with building a resourceful Assistant for workplace is solving the user interface problem. User needs to be able to interact with the Assistant intuitively without the need to learn a large bot vocabulary. Most chatbots today are command driven peppered with some button driven interactions. Chatbots driven purely by a NLU (Natural Language Understanding) engine have to solve some really hard problems with applying NLU in business use cases. In our experience Beta testing with customers, most were naturally drawn towards the smart button driven interactions complementing the NLU triggers.
NLU driven interaction
Interpreting user intent with natural language is the ideal approach to workplace requests classification. However there are some unique challenges with implementing NLU in the workplace. Typically interactions with a workplace chatbots are ambiguous short sentences that are not fully formed. Users type commands that are many times cryptic and expect quick relevant responses.
A key expectation from an Assistant is to be swifter than the alternative options on getting things done. In that context, it is important for the NLU engine to be trained and tuned to the nuances of typical workplace interactions.
At Troopr we are simulating a lot of these interactions and building training corpus that deliver better accuracies in classification. The key is for machine’s ability to continuously train and retrain with user interactions so it can get better at unscripted interactions. For secondary level of intelligence we introduce human in the loop for continuous round the clock grooming. More on this topic in a separate blog.
Eventually the AI engine will be independent, building knowledge and experience from hundreds of thousands of conversations. Until then a smart multi modal interaction will provide for optimal user experience.
Command and Button driven interaction
In a typical command driven interface, the chatbot publishes a list of commands that it understands and requires user to be familiar with them. Button driven interface work similarly in the sense that they both require the user to classify the request and trigger a specific action in the bot.
Commands triggers rely on a conversational followup to clarify and fulfill user requests. Button triggers rely on additional information in a visual popup form for the same. Buttons lend for easier discovery and quicker interactions and are also well suited for multi step workflow interactions.
The interface scaling challenge
Single purpose chatbots work with a finite number of user commands but with a smart Assistant that is multi-skilled, each skillset serves a set of intents. Scaling a command or button driven interaction across skills is not practical. Buttons can help alleviate the discovery problem by anticipating user’s needs based on user type, team, context etc and serve an optimal set of actions in a persistent fashion.
An interface that leverages the power of AI for anticipation and recommendation and fulfills the user’s request with a swift button assisted interaction is one that is well suited to scale.
Personalizing the Assistant
While Smart Assistant will bundle several skills, only a subset of them will be relevant for a user/team. Assistant’s ability to determine and activate the right set of skills and personalize them for the end user is critical to deliver optimal user experience.
Personalization also extends to how users are notified about events that pertain to them directly or to their team. Smart actionable notification delivery mechanism that is efficient in minimizing noise is key component of the Assistant. More thoughts on an ideal notification delivery framework in a separate blog.
Every team is different in the apps they use and in the workflows and processes they adopt. Workplace Assistant needs to be backed by a powerful suite of tools to connect to any business app, define and automate custom workflow and “conversationalize” interactions with the connected apps. A combination of customizable pre-built skills and a powerful skill builder toolkit are essential components of a workplace assistant solution. More thoughts on this in a separate blog.
Workplace Assistants are good for business
A resourceful smart assistant is a win-win for employees and business stakeholders. Employees get to offload some mundane work and spend less time getting work done. Business succeed because a smart assistant
Improves team productivity and saves a ton of time
Maximizes ROI on existing app investments
Instantly upgrades employee workplace experience
Advent of Smart Assistants in workplace is inevitable. At Troopr Labs our vision is to place a smart assistant at every workplace. Today we have started with focus on serving agile teams that are building software products and using Slack as their business chat platform.
Our customers are helping us shape the future workplace Assistant, one skill at a time. Support us by trying Troopr for your agile team or recommend us to one.
Troopr Assistant is available for trial now. Click here to know more.