It was an honor to be invited to sponsor and attend this year’s HOSS, an exclusive annual event SF holds for customers. The Heart of Service Summit is an intimate, invite-only event targeting Director and Service champions and decision makers. This year the agenda included thought leadership, customer stories, product roadmap and networking.

The highlight is always being in an environment where we can hear directly from Salesforce customers. In addition to the customer panel, I had an opportunity to participate in round tables  and speak 1:1 with customers. A few key themes:

• Central to the discussion was EinsteinGPT and in addition to the Service Cloud product roadmap, it was interesting to see how many companies are experimenting with EinsteinGPT, deploying internal tools: some successfully and some who had to pull back already due to user feedback and security concerns.
• Service delivery models span the spectrum of maturity. Some companies remain on traditional email and phone channels and some have deployed full digital channels. There is still so much opportunity to improve service delivery across channels, improving agent experience and productivity, actionable analytics and service quality.
• CRM is hard. Speakers highlighted that rolling out crm is a team sport and requires a healthy partnership between IT and the business and that Change Mgmt is central to success. Some advanced features like personalized and predictive service require good data and a willingness to revise the approach to get right.

AI and ChatGPT

Keynote Presenter: Renee Richardson Gosline of MIT presented on How Generative AI Will Transform CX. 

Renee advocated that Friction is necessary in AI to create an effective customer experience, companies should work to eliminate bad friction and add good friction.

Good friction is providing customers a clear choice in the journey, for example to make a decision about how cookies are being used in their experience vs automating the decision completely out of the process.

Bad friction are processes that are disingenuous and can introduce harm. For example, Terms and Conditions that if not accepted result in limited use of Apps or make it difficult to terminate and exit a service.

Finally, Renee left us with findings that show that when AI as a co-pilot is used, end-user trust in the content is significantly increased. This is simply implemented in the process to enable contact center agents to check the AI generated answers, confirm or update the response before it goes to the customer – and inform the customer the answer is “human verified”. This human-in-the-loop approach is being taken by Salesforce in EinsteinGPT for Service Cloud. I would put this in the good friction camp.

Salesforce EinsteinGPT for Service Roadmap

I am super excited about the EinsteinGPT for Service Cloud Summer release features because they address some of the most important but also challenging aspects in customer service.

• Service Replies: Generate personalized responses grounded in relevant data sources at any touchpoint, on any channel.
• Work Summaries: Drive agent productivity by generating wrap up summaries based on case data and history
• Knowledge Articles: Summarize support interactions to save agent time and formalize institutional knowledge, and drive shift-to-scale
• Search Answers: Surface generated answers to agents’ & customers’ questions that are grounded in your trusted Knowledge base directly into your search page

I’ll have more about these features in an upcoming blog.

One area that I would have liked to get more focus on is data hygiene and cleanliness since AI relies so strongly on good data to be accurate, effective, and build the trust of users. Companies will need to put in place good data governance for the AI learning models.

It was nice to wrap up the day with a casual networking reception sponsored by the Thunder team.