Release Note

Custom AI Agents

Realease Notes
Custom AI Agents
October 3, 2024

Summary

Announcing LocalClarity’s latest AI-focused enhancement: custom AI Agents. Clients can now incorporate unique instructions, brand content, and location data into AI workflows.

Background

At LocalClarity, we are committed to constantly improving our platform to deliver smarter, more efficient solutions for managing your business’s local presence and store reputations. These new AI agents are designed to work hand in hand with your team. They help you manage large volumes of reviews more quickly, with greater accuracy, while maintaining the personal touch your customers expect.

Details

We have added a custom agent builder workflow to the platform. Initially, you can access it from the Intelligent Auto Response workflow, but it will soon have its own page in Settings. The first page starts with a request for a custom icon to represent the agent, a short name, and then a description that helps to identify the roles in a multi-agent profile (this description is internal and not part of any prompt customization).

Next, you can provide business background details used within the AI model prompt to ensure each response reflects the unique flavor of your brand. Share details on the brand’s history, value proposition, market position, and other rich content. Don’t forget to include geographic elements to help the model respond with appropriate regional language preferences. If the model needs to follow specific industry instructions, like HIPAA rules that restrict sharing patient details, make sure to provide instructions in this section.

Prefer to discuss your unique prompt needs with the LocalClarity Client Success team? Let us know, and we can work together in a “sandbox” environment to explore various prompt and settings sets to trigger the idea responses.


The next step lets you craft custom prompt instructions to set the tone of voice based on the review’s star rating. Suggestions are provided as a default setting that instructs the model to be more serious (and brief) as you walk down the rating scale.


Now that the agent understands your business better, it’s time to set your language preferences and appetite for response variability. Here’s where you choose if all review response suggestions should be in the identified language of the review (the “polylingual” setting), or if you would prefer that all responses be composed in a selected language. (Please let us know if you prefer this fixed approach and don’t see the preferred language choice).

At the bottom, you're invited to set the relative response variability, from 0.1 to 1.0. If you’re going to be sending responses without human moderation, we recommend settings in the 0.2 to 0.4 range. You’ll always be able to come back and edit this setting if needed.

Finally, you can confirm all the details to build your agent. Don’t worry—you’ll always be able to go back in and edit any parameters.

What's Next

The Agents are being directly incorporate into the Auto Response Campaign workflow with today release but the team is also working on agentic listing management.