Congratulations! Your first assistant made it into the world, and it can successfully guide your end users through a complete action. It’s now on the path to becoming the dynamic solution you envisioned when you first started the planning process.  

In our post on planning your assistant, we encouraged you to take a walk-run-fly approach. Your assistant is like a toddler that’s just beginning to walk. It needs experience and guidance to learn how to run and, eventually, how to fly. 

Now that you’ve built your first action, the focus will shift to managing your assistant’s lifecycle, which will consist of deploying and improving your assistant’s content.  

Ensuring the long-term health of your assistant requires analysis, iteration, and reiteration. To empower you to achieve this flow, we’ve designed watsonx Assistant with two separate environments:

  • Draft, where you can continuously preview and edit your content
  • Live, which contains the version of your content end users see when interacting with your assistant 

You’ve already gotten a taste of how these environments interact while building your first assistant. Now, you’ll manage your assistant by continuously building, deploying, and improving new versions. The navigation panel is organized along those lines:  

  • Build 
    • Actions 
    • Preview 
  • Deploy 
    • Publish
    • Environments 
  • Improve 
    • Analyze 

Mastering the navigation panel and the relationship of the two environments is the key to a long, healthy life for your assistant. Let’s take the tour. 

Planning and previewing 

Previewing content is woven through the authoring process. The preview panel in the Actions editor lets you measure your assistant’s ability to complete actions while you edit its content. The preview panel saves you time by letting you test actions without leaving the editor. Once you’ve constructed a conversation you’re satisfied with, you’ll want to preview how your assistant will appear on your website. 

The Preview page shows what your users will experience when they interact with your assistant on your site. Click the Preview icon (the play button in the menu) to access a test page with your assistant’s chat panel surfaced, the same way it would be on a live site. Select Change background website and enter your organization’s URL into the field to see the assistant’s chat panel against the backdrop of your home page. 

Customize web chat features a quick link to the web chat editor, where you can edit your assistant’s UI elements to match your brand’s design, update or alter your assistant’s greeting and conversation starters, and enable or disable suggestions. 

This lets you preview how your assistant’s chat panel will appear to your users. The preview panel displays your assistant’s custom name and greeting, the suggestions you’ve added to the home page to funnel your users into actions, plus whatever custom design changes you’ve added. 

The Preview page is also home for the preview link, which you used earlier to share the first version of your assistant with your colleagues. 

Any time you make changes to your assistant’s content in the draft environment, those changes are reflected in the preview URL. Click Copy link to share at the top of the page to copy your preview URL to your clipboard, then paste it into a message to your team. 

The staging ground

When you enter the Draft environment, you see an overview of the architecture of your assistant, starting from the channels where users input requests, leading through to the resolution methods that watsonx Assistant uses to guide users to the completion of actions.

The Channels side of the panel shows you where your assistant is configured and deployed. A reminder, web chat is provisioned for you automatically.  

To see the full suite of available integrations, click Browse catalog above the web chat tab. This will take you to the Integrations page. From there, you can open and edit your web chat in either the live or draft environment, connect your web chat to a live agent integration, and activate the Phone integration. 

The integrations page is also where you connect your assistant to third-party apps like Slack or SMS with Twilio, embed search into your assistant by adding Watson Discovery, and build custom extensions.  

Once you add a new integration, it appears in the draft environment page with a setup status indicator. Adding channels in the draft environment allows you to test channel integrations before deploying them to end users. 

Tip: Channel configurations and environment settings are not updated automatically when publishing a new version. They must be manually configured for each environment. 

The draft environment page also provides a timestamp on your latest draft edits and a quick link to your draft content. Click that link to access your actions editor and edit your content prior to publication. Once you’ve added all your preferred channels and updated your actions, it’s time to progress to publication.  

The ignition button

You received your first taste of the Publish page when you built your first action. You’ll return to it often as you continue to revise and improve your assistant’s content. In addition to being the point of publication, the Publish page provides an overview of both your unpublished content edits and a comprehensive version history. 

Once you click the Publish button at the top of the page, all your modified content is captured in a version that is saved in your version bank. Versions are snapshots of your content. 

Versions are numbered in the order of publication. When you’re ready to create a version of your draft content, a modal will pop up that allows you to give a description to that version and, optionally, assign it to the live environment. 

Your list of unpublished changes is cleared out after you publish, and all subsequent edits to your actions will be saved in the draft environment before you’re ready to publish a new version of your content. 

If you notice any bugs or fixes necessary in a published version, you can use the Revert button to pull content from a prior version back into your draft environment. 

In order to avoid losing any in-progress work, you can choose to create a new version of your content before reverting to a prior one. Click the revert button, then select the version you want to revert to and click Publish and Revert to pull that version back into the draft environment.

Tip: You have the option to add a description for each version before publishing. Adding descriptions makes it easy to differentiate between versions, a helpful feature when trying to decide which version to pull in via the revert option. 

The larger world

The layout of the Live environment is identical to the draft environment, tracking your deployed channels and resolution methods. The difference is that everything you see in this field reflects content and integrations that are available to your end users. 

We’ve made it easy to assign or switch versions of your content in the live environment. The overview shows you which version of your assistant is currently live.

Click Switch version to open a grid listing all your published versions. Click the version you want to switch to and select Switch version. The live version is now switched without affecting any unpublished content. 

Learning to fly

By now you probably understand why we describe MVPs as toddlers taking their first steps. They need guidance (an infusion of your domain expertise) in their early stages, and experience (interactions with users and consistent content iteration) to build on their initial training and become the most mature versions of themselves. 

Now that you’ve built your first assistant and know how to maintain and improve it, you can begin growing your assistant into the complete one that you first envisioned, one that can handle any user’s request at any phase of their journey.  

There’s a wealth of training material available to you as you continue to train and improve your assistant. Check out the watsonx Assistant blog on Medium for articles on common conversational patterns, taking advantage of your assistant’s built-in analytics, and adding integrations and extensions. Also, don’t forget to take advantage of the various enablement materials IBM has provided: 

  1. watsonx Assistant Learning  
  2. The Getting Started blog series 
  3. Welcome to watsonx Assistant in the product documentation 
Was this article helpful?
YesNo

More from Artificial intelligence

Why you should use generative AI for writing Ansible Playbooks

2 min read - Generative artificial intelligence (gen AI) can usher in a new era of developer productivity by disrupting how work is done. Coding assistants can help developers by generating content recommendations from natural language prompts.As today’s hybrid cloud architectures expand in size and complexity, IT automation developers and operators can benefit from applying gen AI to their work. In a 2023 IBM survey of 3,000 CEOs worldwide, three out of four reported that their competitive advantage would depend on who had the…

Empowering the digital-first business professional in the foundation model era 

2 min read - In the fast-paced digital age, business professionals constantly seek innovative ways to streamline processes, enhance productivity and drive growth. Today's professionals, regardless of their fields, must fluently use advanced artificial intelligence (AI) tools. This is especially important given the application of foundation models and large language models (LLMs) in Open AI’s ChatGPT and IBM's advances with IBM watsonx™.   Professionals must keep up with rapid technological changes such as cloud computing and AI, recognizing the integrative power of foundation models, which…

Building trust in the government with responsible generative AI implementation

5 min read - At the end of 2023, a survey conducted by the IBM® Institute for Business Value (IBV) found that respondents believe government leaders often overestimate the public's trust in them. They also found that, while the public is still wary about new technologies like artificial intelligence (AI), most people are in favor of government adoption of generative AI.   The IBV surveyed a diverse group of more than 13,000 adults across nine countries including the US, Canada, the UK, Australia and Japan.…

IBM Newsletters

Get our newsletters and topic updates that deliver the latest thought leadership and insights on emerging trends.
Subscribe now More newsletters