• State of Mind
  • Posts
  • Conversational AI & the 5Ws of Autonomous Agents

Conversational AI & the 5Ws of Autonomous Agents

An overview of where we are seeing the most value in AI * CX

Personalized generative responses are kind of table stakes at this point. Yes it is amazing, but the real value is being able to autonomously trigger workflows in the various systems your team is using. Without the ability to make function calls for Order, Subscription or Returns; it would be like having someone on your CX team who can respond to customers but can’t actually login to any of your systems and make the needed update. It’s great, yes, you can help with product questions but your team may still have 100+ requests (that require some form of update) from over the weekend that have to be manually resolved. What if those 100+ requests on Monday morning were actually resolved perfectly across all of the apps in your stack? There was someone on your team who worked all weekend making sure that on Monday morning those requests were handled… any that were not able to be handled were escalated to the appropriate team member. This is where StateSet is in the development cycle of Conversational UI & Autonomous AI Agents. We can create, deploy and manage autonomous, personalized AI agents that can not only respond to questions but make function calls across the various systems that you are using and provide on brand confirmation messages to customers 24/7.

Actions are the Future of Autonomous Agents in the DTC ecosystem and beyond. There are three areas we are seeing for post-purchase automated actions:

Order Changes - Add Tag(s) to / Cancel / Change Shopify Orders

Subscription Changes - Automatically Cancel, Skip Next Order or Pause a Subscription. Generate a Portal Link for the customer to manage their account and more.

Return / Warranty Exchanges - Automatically process a Return / Warranty Replacement. Use Order Number lookups to assist with return / warranty replacements.

All of these revolve around the Order & Transaction.

It doesn’t matter if it’s Shopify, WooCommerce, Stay, SKIO, ReCharge, Loop, ReturnLogic, Gorgias, Zendesk, ShipStation, ShipMonk; we can integrate with any commerce system and provided an orchestration layer for powering a next-gen CX AI Agent through a conversational interface.

How do we do this? What do we need to get right 100% of the time.

The 5Ws

We are working with multiple 8 & 9 figure brands on deploying autonomous AI agents across post-purchase operations. What is important to consider when determining how to better automate a specific subset of tasks that may be completely manually and costing your CX & Ops time time / money?


Who is sending the message back to the customer? Is the AI Agent named a real persons name? It is Robert or Jessie on the Customer Experience Team? Is it a persona that the customer knows will get back to them with helpful information? Do we say it’s an automated attendant and let them know it’s an AI that is responding? Or do we just say “Customer Care Team”?


Are we responding back with a generative & personalized message based on your company knowledge base, FAQs, tone, macros and more? Or are we sending a 100% deterministic pre-approved copy that is the same exact thing every time based on the intent detected? Is the message a long-form message or should it be short form and concise? What is the tone of the agent, are there brand specific words that we should always use? What are the macros that we want the AI to respond with? What functions / actions should the AI Agent be able to call? What rules should the Agent have to follow anytime it generates a response?


When should the AI Agent actually process this type of request. Is it based on a specific channel? Is it based on a specific intent that was detected? Is it based on when a specific tag on the ticket is present? Should it be scheduled to be sent back after 5 minutes for email versus right away for chat? Should an action be sent only after we receive confirmation from the customer? Determining when a message from the customer should be processed and a subsequent action should be taken is a very important step. Are there any time thresholds that we should be checking before confirming that an action can be processed?


Where should the request be processed? Should it be sent back to the same channel it was received in automatically? Should it be sent back as an internal note that can be copied by another member on the team or should it be automatically sent to the customer? Should the response be sent via eMail, Chat, sms or IG DM?


What are the key performance indicators for our AI Agent. Are the KPIs for an Order Agent different than those for a Subscription Agent? What are the outcomes that we expect customers to experience? How does an AI Agent help with not only external facing customer experience, but reporting / intelligence for our internal teams?

The ability to have actions combined with AI agents with context aware state of order history, shipping statuses, subscriptions, returns and warranties history. Conversational AI Agents can create a next-generation customer experience based on a unified customer profile across all of your apps. This combined with a next-generation workflow orchestration layer for automation is what creates the future of autonomous commerce operations.

For more info on what we are building reach out at [email protected] or setup time for demo here.