Author
Dustin Dye
Strategy
January 9, 2022

Building the ultimate Google Cloud chat stack.

Botcopy is a Google Cloud Partner. We like Google for many reasons, but the main one is the tech. The NLU engine, machine learning, and ease of use make Google Cloud Platform (GCP) the ideal place to start building the ultimate chat stack.

(Quick reminder who we are: Botcopy connects Google Cloud Dialogflow to your website. We’ll soon be integrating with more frameworks. For now, let’s focus on Google’s fantastic NLU engine and why GCP may be the right choice for your chat stack.)

Let’s first take a GTO. (Grandma Time Out.) What the heck is chat stack?

Here’s how I’d explain it to my grandma. (Or grandpa.) A chat stack is the combination of tools you need to communicate effectively through automated chat. I like to use the analogy of a coffee stack.



Ultimate coffee stack

Coffee grounds, coffee maker, filter, cup…you get the point. The analogy is: the coffee is like the chatbot UX, the coffee maker is the NLU/NLP framework, the cup is the database to store the chat history, etc.

Adding milk and sugar is like adding branded designer touches and rich content that enhance your bot’s UX.

Sorry for the extended metaphor. The bottom line is at Botcopy, we make excellent “coffee.”

More importantly, so can you.

The ultimate GCP and Botcopy chat stack

The ingredients you need to create a perfect virtual assistant: data storage, NLU/NLP, webhooks, analytics, and a platform for deployment.

Here’s an infographic for how a solid chat stack fits together. Below that, let’s dive into each ingredient in detail.

Data storage — Google Cloud Project.

Store the chat histories in a safe, secure way, with tons of affordable space. It’s Google, people! You trust them with your emails and docs. You can’t go wrong with Google for keeping your agents and chat histories safely stored. GCP’s cloud security is compliant enough for the world’s biggest companies — banks, governments, healthcare, you name it.

NLU/NLP Framework — Dialogflow CX or ES.

NLU framework is a biggie — so it will be the most detailed aspect of the ingredient list. In conversational AI tech, Google Dialogflow is without peer. It’s a powerful tool designed for enterprises that are building advanced virtual assistants.

Its NLU (natural language understanding) is the best we’ve tried and is crucial for creating an intuitive and enjoyable experience. You can create a UX that understands what users are saying. Your bot can take non-linear paths and keep context, just like a live agent. With great NLU, the power to scale the chat experience is virtually endless. 🚀

What sets Google apart — to this day — is the intent mapping.

Not to get too technical on you, but when an end-user types or says something, Dialogflow leap into action to find the best match to the query. This is also called intent classification.

Dialogflow scores possible answers with an intent detection confidence known as the confidence score, ranging from 0.0 (complete uncertainty) to 1.0 (totally certain). After the outcomes are scored, Dialogflow makes smart matching decisions and files away this learning to weigh future scores. It’s applied statistics, which is a less-fancy synonym for machine learning (ML).

GCP still leads the pack in ML. Machine learning helps with accuracy >> accuracy helps with people not getting annoyed with your bot >> if people like your bot they will use it >> if people use your bot, your efficiency explodes, expense plummets, revenue skyrockets >> you win. 🏆

ML helps with accuracy >> accuracy makes your bot more likable >> if people like your bot they will use it more >> if people use your bot more, your efficiency explodes, expense plummets, revenue skyrockets >> you win. 🏆

Webhooks — Google Cloud Functions.

No chat stack is complete without the ability to automate end-to-end solutions and personalize experiences. With Google Cloud Functions and the Dialogflow and Botcopy stack, you can pass a user query or event to your framework, where an API call is routed to your database. Data is then retrieved and served back to the end-user. We’ve created a custom webhook template on our Github page to make this process easier for you. Read about it here.

(To see examples of Botcopy and Dialogflow fulfillment coming to life, see our ref parameter detection (RPD) video below. You don’t need RPD to leverage Dialogflow fulfillment, but it’s a quick way to demonstrate the power of webhooks.)

Analytics — Dialogflow.

Aside from search, Google is famous for Google Analytics. So it’s no surprise their bot analytics are great, too! Enjoy a birds’ eye view of how your chatbot is being used.

No chat stack is complete without a way to monitor bot performance and learn from your findings. Tools to measure performance, and find opportunities to improve your bot for better results, are not optional, they’re required. GCP’s Dialogflow analytics keep getting better, and it has always shown which intents are triggered — something other frameworks, e.g., Amazon Lex, lacked until as recently as late 2019.

You’ve got options

Botcopy has always had native integrations with Chatbase and Dashbot, tools designed for advanced bot analytics. And good news — Chatbase (a Google product btw) is officially migrating to Dialogflow. 👏 So using GCP, all your analytics needs can be met in one place.

To sweeten the coffee, Botcopy just released a tapered-down analytics page within the Botcopy portal, in case you want to glance at key metrics throughout the day without a million bells and whistles.

The platform for deployment — Botcopy.

Google Cloud Dialogflow is wonderfully omni-channel so you can deploy your agent to many well-known messaging apps. BUT, to get your agent on what we consider to be THE most important place — your website — Botcopy’s SaaS is your best friend. It connects Dialogflow to a rich, custom web chat in one click.

The way we see it, Google has the best AI, Botcopy has the best UI, which is why we call Botcopy the UI of AI.

A glimpse at pricing

The whole shebang costs more than a cup of coffee. But it costs less than you might think. Dialogflow ES offers an extended free tier, so you can kick the tires and create your Dialogflow & Botcopy POC without a commitment. In the beginning, you can even deploy a production-ready chatbot live without paying a penny.

Botcopy’s native integration and partnership with Google allows you to create a free Botcopy account and connect your Dialogflow agents to Botcopy in less than 5 min. In the Botcopy portal, you’re free to style the frontend interface to fit your brand’s needs with several adjustments to the shape, colors, headers, animations, input bars, full-screen mode, language drop-downs, and many more options, too numerous to put here.

After you get the look-and-feel right, dive into more advanced intelligence features of your new interface. These include event triggers, window listeners, bot prompts, ADA compliance, TTS, custom response types, building web-views for login screens, forms, etc. live chat confirmation with any open API, and so much more. See info and videos on these features at botcopy.com/features.

Getting locked & loaded

Once you fall in love with the ultimate chat stack, GCP offers numerous deals and incentives to unlock more technologies and help you ramp up your program. Botcopy offers our Gold and Pro plans, unlocking deep volume discounts and expert services to ensure that all the bot program pieces sync together nicely and follow best practices.

Now THAT’S a chat stack.

If you’re looking to gear up your virtual assistants for 2021, and want to revamp your support operations with an AI-first mindset, let’s find time for a quick call. We’d love to help you out.

Stay tuned for Botcopy’s 2.0 Product Hunt launch coming November 12th!

We look forward to supporting you so that you can better help others. 👊

That’s all, folks.