Salesforce World Tour returned to London this week. Whilst I was unable to attend in person, I’ve been catching-up on Salesforce +. Want a quick summary? You’ve come to the right place! Here are my key take-homes from World Tour London: AI Day.
Investing in the UK market
Salesforce is investing $4 billion into the UK business over the next 5 years. Areas of focus will be AI, innovation, customer success and growth.
Summer of AI
At AI Day: New York, Salesforce referred to this being the ‘Summer of AI’. Instead of the normal Spring, Summer, and Winter releases, new features are being released when they are available. With 16 AI feature drops in the next 3 months, there is plenty to keep an eye on!
AI Day: AI Cloud & Generative AI
The focus is on rolling out generative AI capabilities onto Salesforce. Salesforce’s vision is that by infusing generative AI throughout the platform, it can help end users, admins and developers be more productive. Very powerful stuff!
Sales, Service & Marketing GPT
There are many potential use cases for generative AI in Salesforce. During the sessions, various features were overviewed. The main items covered are as follows:
- Activity 360: Introduced in May, this automatically logs interactions and provides real-time activity reporting with actionable insights in a single place. More information available here.
- Buyer Assistant: Personalised AI driven conversations with prospects on a website to help with lead qualification.
- Conversational AI: Supporting 16 languages. AI driven call transcription, summaries, next steps and call coaching (e.g. Zoom, Teams, Google Meets), etc.
- Sales Emails: Empower team members with AI generated emails. Check out the ‘Prompt Studio‘ slides for more information.
- Sales Assistant: Tool to help sales reps complete prospecting, research, carry out follow-ups and update CRM data.
- Service Replies: Generate personalised, grounded responses to customer interactions from any channel.
- Work Summaries: Wrap-up a case automatically based on case data and history.
- Knowledge Articles: Automatically generate knowledge articles (e.g. based on Slack Swarms, Chats, etc).
- Mobile Briefings for Field Service GPT: Automatically generate briefings for front-line workers; info for a specific service appointment (i.e. assets, customer data, information from previous interactions, etc).
The case study from John Lewis Partnership demonstrates how the top three of these can be used. Examples of the top three features are shown in the ‘Prompt Studio’ slide deck.
NOTE: There is a follow-up webinar on Service GPT on 19th of July 2023. Click here for more info.
Einstein GPT Trust Layer
You may be wondering what is at the heart of all of these generative AI features? The answer is AI cloud, which is powered by the ‘Einstein GPT Trust Layer’. This provides the tools to:
- Harness generative AI securely (e.g., data masking, secure gateway and zero retention).
- Three deployment strategies for Large Language Models (LLM): shared trust, Salesforce hosted or bring your own model.
- Protect against generative AI risks, such as hallucinations, bias, and toxicity. Click here for more info.
It is also worth noting the Audit Trail provided by the Einstein GPT Trust Layer. This will allow admins to see how prompts are being used and their effectiveness. Overall, the solution keeps ‘humans in the loop’; empowering teams to review generated content and amend prompts as needed. For more information, click here.
Data Cloud (formerly Salesforce Genie / Customer Data Platform) featured throughout the day. Case studies such as Santander, Heathrow & F1 showcase how Data Cloud has the potential to harmonise fragmented data. This can then be fed into Einstein GPT to generate personalised, targeted content.
The following updates were of note:
- AI driven segmentation: Use Natural Language Processing (NLP) to segment data in Data Cloud
- Data Sharing with Google BigQuery: Data Cloud can query data in Google BigQuery, without copying the data.
- Bring Your Own model with Google Vertex AI: Use Google Vertex models across the Salesforce Platform. Click here for more information.
A prompt is a question and/or instruction given to an AI tool to generate an output. Providing a clear and ‘grounded’ (i.e., contextualised) prompt is critical. After all, an LLM cannot generate a useful response, if you have not given it the context it needs. Users could type in long, detailed prompts. However, this is time intensive. This is where ‘Prompt Studio’ comes in.
Prompt Studio allows admins to create prompt templates. These can be used throughout the platform (e.g., Apex and Flows). This will likely become an essential tool, with more information following in due course. Check out the slides below for more information.
Before we start to wrap-up this post, it is important to take into account this piece of advice:
Given all the excitement about AI, it is important to remember model outputs can only be as good as their inputs. The Einstein GPT Trust Layer provides many valuable tools. However, Salesforce teams and system users must work together to keep an org clean. This will help ensure everyone gets the greatest results from AI cloud.
It is easy to focus solely on the technology from the day. However, World Tour also provides an opportunity to reflect and recognise the contributions of others in the ecosystem. Here are two pieces of news which caught my attention.
A new golden hoodie recipient
Levelling the Playing Field: Equality Matters
Last but far from least, there was a very thought-provoking session on equality. The panel discussed a variety of key topics for any business. Among the key themes covered were:
- Accountable: Ensure that there is transparency of key metrics and on-going monitoring.
- Intentional: It is important to actively reach-out to under-represented communities.
- Measurability: Define clear and transparent measures which can be objectively considered.
It is clear that Salesforce’s ‘Summer of AI’ is in full-swing. This post has covered a variety of features and announcements, ranging from the Einstein GPT Trust Layer, through to Prompt Studio. As we head towards Dreamforce and beyond, we’ll see more innovations become available. Whilst it is still relatively early days for generative AI, there is already plenty to explore.
Until next time, thanks for reading!
Bonus Penguin Fact
In a previous post, I wrote about Little Penguins (Eudyptula minor). This past week, I visited Sea Life Weymouth, home to the only colony of these penguins in Europe! In honour of these penguins, here’s another fact about the smallest living species of penguin.
Little Penguins are native to Australia and New Zealand. Phillip Island (Australia) is home to the largest Little Penguin colony in the world. It has become famous for what is known as the ‘Penguin Parade’. At dusk, the Little Penguins return from their hunting and waddle back to their burrows. This can prove quite the spectacle!
Found this article useful? Why not share!
Or check out some of our other articles: