Changelog

Occasionally with AI you don't always get an answer you expect, with our improved Inspect feature you can better understand your new AI agent's responses.

Just click Inspect answer then ask your question to better understand the answer.

You can read more on debugging AI replies here.

It’s been a little while coming (and a good amount of work!) but I am happy to announce we have now obtained our SOC-2 Type II certification!

SOC 2 Type II is an independent audit that shows a company is following strict security and privacy practices to protect customer data.

While a Type I report checks that the right controls exist, a Type II report goes further by verifying that those controls were actually working effectively over a period of time.

Hopefully this should help us to:

• Build trust: Proving to you (our customers and partners) that we handle your data securely.

• Reduce risk: Ensuring we have strong processes around security, availability, and data handling.

• Operate maturely: Demonstrating consistent, well-run, industry-recognized internal security practices.

If you would like access to our Trust Center or SOC-2 report, just get in touch with us via our chat.

You can now use AI to automatically tag your Freshdesk tickets and Freshchat conversations.

It can take hours of repetitive work to tag and categorize your support tickets and emails each month.

But now your AI agent can automatically do this classification and categorization for you, directly in Freshdesk, accurately and consistently, so you can focus on understanding what your customers are asking and how they are feeling.

You can also use it to decide which tags or custom fields categories you want your AI agent to reply to, and which you will handle yourself.

Learn more here for Freshdesk and here for Freshchat.

Now you can control which types of conversation your AI agent replies to in Intercom with our block AI reply by tag feature.

For each tag or conversation attribute you can decide whether you want to use your AI agent or not.

Learn more about how to enable this here.

We often don’t chat in the same way we email, so to reflect in your AI agent, you can now specify whether your Communication & Style guidance is for your agent everywhere, or only for email replies.

Go to Improve > Guidance and either enter a new piece of Communication & Style guidance or edit an existing one, then select “Email only” from the drop down at the bottom of the guidance.

If you’re a HubSpot or Gorgias user (especially those with multiple accounts or brands), you can now specify which email accounts you want your AI agent to reply from by allowing accounts.

To do this, go to AI Agent Setup > Customization and scroll to the “Reply from specific accounts” section.

Enter the accounts you want your AI agent to reply from, click Allow and you’re all set!

Make the most of your AI agent by allowing it to execute SOPs or processes on your behalf, taking repetitive tasks like refunds, address changes and more away from your human agents.

Tasks enable you to bring together personalized user data, interactions with your back end databases and product via actions and tools, customized guidance and your AI's knowledge to step a user or customer through a process that requires business logic.

In simple terms what that means is, instead of users having to click through a multistep flow (old world of chatbots), they can talk to the AI that refers to an SOP or process that will ask the right questions at the right time and not be as rigid as something flow-based.

Read about some example use cases for your industry here, or learn more about tasks here.

Our new Test feature takes a sample of your historic support tickets and previous questions asked (if you have integrated Zendesk or Intercom) and/or lets you add in your own test question set and shows you how your AI agent will respond to those questions, so you can see quickly how it will perform on day 1.

You can find Test in your Dashboard under Improve > Test.

Learn more about Test here.