Bee Partners' Industry Insight: NLP on the Verge of Utility
Natural Language Processing (NLP) on the
Verge of Utility
September 2016, revised March 2018
Seed-stage venture capital firm investing in
passionate Founders with deep market insights
The ability of consumers (and enterprises) to conduct
e-commerce dynamically through the use of texting
or spoken natural language.
What is Conversational Commerce?
3rd party Application
Virtual Assistant Platform UI
Agent / Skills / Action Taker
Ex: Bot built using SDK
Knows Intent + Takes Action +
Knows Intent + Takes Action
Bee Partners NLP Stack Diagram
Here’s Why It’s Happening…
• We are in the midst of a (continuing) paradigm shift in how humans interact with
• We are now using human conversation (voice or text entry) to seamlessly obtain
information, goods, and services on our various computers.
• This leads consumers to desire “One Place” to dynamically conduct all our
business and social activities.
• And yet this “One Place” is ubiquitous: Anywhere and everywhere - voice, Siri, in-
car, in-home, messaging apps – the distribution channels become infinite
This Drive Towards One Place Effectuates the
‘Stackless Interface’ …
A single human-computer interface (HCI) that connects
the user to any end requirement
… Which Often Manifests Itself in a “Chatbot” to
A chatbot is a computer program, driven by natural language processing (NLP) and
machine learning (ML), designed to simulate intelligent conversation around a
A chatbot is a computer application that provides a consumer with a convenient way
to interact (interface) with a business via online texting or voice.
A chatbot delivers “conversation experiences” with the enterprise
A chatbot is to a messaging app = Website : Browser
A chatbot can conduct conversational commerce
better than humans in some aspects and worse in
Chatbot Pros and Cons
It is better than humans because:
(1) it can chat with thousands of people
simultaneously, autonomously, and
(2) it can hold vast amount of knowledge in
its memory, and
access it quickly.
(3) it can handle asynchronous
(4) it never gets tired, is consistent, and
And not as good when:
(1) talking about subjects outside its
(2) answering dynamic questions, or
generating speech (NLG).
(3) being witty or thinking outside the
(4) linking commerce across platforms or
How Did We Get Here?
The intersection of three trends is driving this evolution of human computer
interaction (HCI) towards a conversational commerce interface.
1) Consumer Trends
• Messaging is the go-to [millennial] way of communication.
• More people today are using messaging platforms than social networks.
• Messaging is no longer limited to personal communication, but is widely accepted
as a means of business communication.
• Consumers want on-demand access to information services with minimal barriers
– the ‘stackless interface.’
• Native apps are dying and consumers are shifting away from websites and app
• Asynchronous communication is now 7/24/365.
There is a global movement towards messaging platforms and
2) Business Evolution
Enterprises must focus
on delivering personal
customer service at scale
LOW PERSONALIZATION & HIGH
HIGH PERSONALIZATION & LOW
& HIGH SCALABILITY
3) Technology Advancements
Advancements in Artificial Intelligence
• Advances in machine learning such as deep neural networks (DNNs) have allowed
bots to learn as they are fed more data. This improves the utility beyond keyword
Access to new messaging and speech platforms
• Big tech companies have opened up their speech and messaging platforms to
Distributed platforms of IoT
• Voice platforms like Siri, Cortona and Okay Google have become ubiquitous in
phones, laptops, and cars.
Artificial intelligence and access to new messaging and speech platforms have
enabled the emergence of conversational UI.
So What Does This Mean for Investors?
Bee Partners Preliminary Guide to Evaluating the
Conversational Commerce Paradigm
1) Effective Conversational Commerce Must
Leverage Existing Platforms
We do not expect adoption of a chatbot that forces users to embrace a new or
application specific platform.
2) Effective Conversational Commerce Must Have
Functional Expertise and Vertical Depth in a
• CONTEXT IS KING: if you go deep enough in any vertical (the most data with
the best lexicon), the application is going to be differentiated and deliver a
• It’s more difficult to imagine a functional horizontal chatbot. The bot would
need to be able to have a conversation across a diverse set of topics.
• It’s difficult to pick a winner in horizontal enterprise or broad enterprise that
still requires the client to customize.
• Current horizontal opportunities do exist at the “Handshake” or
infrastructure layer where the application connects to the platform.
3) Effective Conversational Commerce Leverages
the Best of Both Consumer and Enterprise
• Bots and personas must feel human but drive business (conduct commerce).
Consumer and enterprise simultaneously.
• Team must pair deep technical knowledge in NLP, AI and machine learning with
vertical business experience.
• Must intimate direct, custom relationships at scale.
• Consistent with the consumerization of enterprise SaaS.
4) Effective Conversational Commerce Must Add
Value Beyond Just Being a ‘Stackless Interface’
• Price arbitrage, assisted selling, previously unavailable channel or way to convert
• Chat is the new search; must immediately get customers deep in the
application/company to find relevant information.
• Ability to close the transaction with the same seamless experience; this may
differ between platforms’ technical capabilities.
5) Effective Conversational Commerce Often
Looks Internally within an Enterprise
• Using NLP in a non-customer-facing way can make employees more effective in
searching for requested information.
• Additional employee capacity drives better outward customer support and
better customer outcomes.
• Customer service is the marketing of the future; brands will struggle to be
competitive without an effective service layer.
The Result of All This Is...
• Continued Democratization of Information
• Additional Friction Layers Dropped
• Approaching “One Place”
And Finally, the Future!
• Your conversational searches reveal immediate answers (faster).
• Everyone has an effective personal assistant (more accessibility).
• Every online brand interaction has a conversational (bot) element.
• Your phone only has 3 apps!
– Phone app
– Social media
Maybe… it won’t even have a keyboard!
What might the world look like 10 years from today?
Bee Partners’ Core Positions Include Application,
Handshake, and Infrastructure Layers
Find and book the perfect hotel room over a delightful
[Exited] The next generation of voice-driven and context-aware human-
computer interaction (HCI). The company has pioneered Speech Team as a
Service (STaaS) around its Deep Neural Network-based speech recognition
A creative technology company, offering a language-based artificial
intelligence platform for sports fan engagement.
Solving EHR complexity with specialized clinical AI and NLP
We welcome introductions to experienced Founders with a deep vertical insights
• have deep vertical domain expertise; strategic thinking about what this new UX
• whose thinking is consistent with the 5 points of Bee’s thesis on effective
And of course... please use and evangelize Cohort, SnapTravel and StatMuse!