Introducing Instant Assistant—powered by Voiceflow's AI

In 2021, Deloitte said it took six to nine months to build the average assistant. Given how time-consuming and labor-intensive the process was, it made sense that many companies couldn’t afford to invest in building a chatbot.

But with the advent of large language models (LLMs), and a lot of work behind the scenes to streamline this process, we’re thrilled to announce that you can now deploy a functional assistant in minutes instead of months.

Meet Instant Assistant

We’re all familiar with the hottest debate in the industry: Will LLMs replace natural language understanding (NLU) systems? Our verdict is that they are complementary, not competitive.

With Instant Assistant, you can now unlock the best of both worlds. Design with NLU to capture user intent, then leverage LLMs to generate responses that are directly informed by your institutional knowledge.

Things to know

Instant Assistant is not an engine for content generation—it’s powered by an LLM but informed by your institutional knowledge. Using your existing knowledge base, Instant Assistant can quickly and accurately answer questions about your business. A few examples:

Pssstttt... wondering what a knowledge base is? Good question. A knowledge base is like a virtual hub for all the important information related to an organization. It's a go-to reference for people seeking information, making it easier to find answers and get things done faster.

Choose your own path

Depending on your use case, you can tailor how you use this release.

If you’re creating your company’s first assistant, explore Option 1 and build from scratch. It takes minutes to build a robust assistant, making it perfect for those who are just getting started.

If you’re a mid-market or enterprise company with an existing assistant, explore Option 2 and integrate it into your existing workflow. Voiceflow can be a secure sandbox for you to experiment with LLMs—without putting your company at risk.

No matter which option you choose, you can select the LLM that best suits your needs: GPT-3, GPT-3.5, GPT-4, Claude, and the newly launched Claude Instant—just to name a few.

Option 1: Build an Instant Assistant from scratch

If you’re looking to spin up an assistant for the first time, follow these three steps:

1. Connect your company’s knowledge base. Upload PDFs, text files, and/or URLs to your Knowledge Base data source.

2. Test your assistant to preview generated responses. Leverage our prototyping tool to build confidence in your assistant's ability to serve your customers.

3. Push it live and invite your customers. Ready to deploy your assistant? Hit publish, copy your code snippet, and drop it into the <body> tag of your website. A similar process can be followed to deploy your assistant on a supported channel, such as WhatsApp, Twilio SMS, or Microsoft Teams.

Option 2: Integrate Instant Assistant into an existing assistant

Want to control how your assistant responds without manually defining what should be said? Leverage our Knowledge Base data source directly from within Voiceflow’s AI steps to create guardrails that allow you to safely experiment with LLM-generated responses. It’s a workflow-agnostic feature, meaning you can choose where in the conversation to leverage your institutional knowledge.

The capabilities you'll unlock along the way

Use Voiceflow to build an Instant Assistant, and you’ll unlock some compelling benefits:

  1. Safely experiment with LLMs. While there’s a lot of buzz about LLMs in the industry, we’ve observed that intelligent controls are a much-needed feature, particularly for enterprises. By leveraging the Knowledge Base data source in your AI steps, you can experiment with generative AI without putting your organization at risk.
  2. Accelerate your workflow. Say goodbye to crafting custom logic and flows. By leveraging the Knowledge Base data source, you can build an assistant in months instead of months.
  3. Supercharge your customer conversations. With Instant Assistant, serve your customers on autopilot—answering their most common questions about your company, team, products, and services with human-like responses.
  4. Reduce your operating expenses. Budgets may be tight, but you’re still expected to achieve your goals. That’s why we’ve made the features that power Instant Assistants available on our free-forever plan.

What will you build?

Whether it’s an assistant tackling Cheesecake Factory’s monster-of-a-menu or something to streamline your internal documentation, the opportunities are endless. Here are a few tutorials and templates to inspire you.

🏗️ Build a restaurant menu bot

For this example, we couldn't think of a better restaurant menu to tackle than the Cheesecake Factory's textbook-length menu. If you want to make a restaurant menu bot that helps customers make a decision, upload a menu into your Knowledge Base data source. Here's how our own Peter Isaacs tackled the Cheesecake Factory menu:

🏗️ Build a support chatbot powered by your sources

This template lets you create a chatbot that can provide product recommendations, get user information, conduct sentiment analysis, and run an API summary. It’s a perfect starter template to explore the range of capabilities Instant Assistant unlocks.

🏗️ Create a Shopify chatbot

With this template, you'll have a running Shopify chatbot for your e-commerce store to let you automate FAQs, suggest product lines, and boost your customer experience.

🏗️ Create a Slackbot

This template empowers you to instantly create a Slackbot with Replit that answers questions from any internal documents you upload.

🏗️ Build a Discord Bot

Use this template to build a Discord Bot powered by GPT4, Replit, and Voiceflow.

Let's go

Whether you’re looking to build an assistant on the fly, or experiment with LLMs risk-free inside your existing workflow, we encourage you to take this new feature for a test drive. Ready, set, sign up for our free-forever plan.


Introducing Voiceflow's GPT-powered AI Builder

No items found.