SugarCRM has announced the latest iteration of its AI-fueled CRM platform with the introduction of generative AI. The vendor promises out-of-the-box value and productivity gains for sales marketing and service teams with the new functionality.
Targeted at the mid-market, SugarCRM has invested heavily in AI, machine learning and predictive analytics. It aims to automate business processes and deliver the ability for teams to be more effective. This will help deliver better revenue and profits to customers.
Generative AI is a game-changer. A McKinsey report, “The economic potential of generative AI: The next productivity frontier”, notes that, “Generative AI is a step change in the evolution of artificial intelligence.”
The report later says, “We identified 63 generative AI use cases spanning 16 business functions that could deliver total value in the range of $2.6 trillion to $4.4 trillion in economic benefits annually when applied across industries.”
The report also noted that 52% of the benefits will fall in customer operations: Marketing, Sales and Service. It is, therefore, no surprise that SugarCRM has invested in technology. Organisations that are not using solutions with generative AI may need to reconsider their decision.
Craig Charlton, SugarCRM CEO, said, “Most businesses are looking to AI to transform experiences and drive greater productivity. Today’s generative AI announcement is the latest evidence of our commitment to make AI accessible to all and to maximize usability for a next-level CRM user experience.
“The generative AI revolution is here, and midmarket sales, marketers and customer service pros can’t afford to be left out of the equation,” he said. “With Sugar, organizations can unleash the potential of generative AI to automate, accelerate and optimize marketing, sales and customer service.”
SugarCRM sees four basic use cases for generative AI within its platform. It has applied to all four across Marketing, Sales and Service. Hildebrand explained them, saying.
“Composing, sales and marketing content, helping marketing professionals or sales professionals to create emails, make them more compelling. Maybe even ask AI to use a specific tone that might resonate more.
“Finding answers and recommendations. You have a problem. It could be a customer service problem. Then, you find an answer to the problem or a recommendation on how to solve it, or the recommendation could be a product recommendation. This is also the area where we have chatbots and personal assistants who’ve been doing similar things already.
“Summarizing things, there’s a huge potential. For example, if you need to visit a new customer that you’ve never visited before, you don’t know anything about it. Now you can go into the CRM system and look at a lot of data, the purchase history, you can go online and find information on Bloomberg to get some firmographic stuff, etc. That takes a lot of time to prep for your first customer meeting. Generative AI can collate all that stuff, and then you just read through it in like three minutes. That’s a huge productivity gain.
“The fourth category is Creating ad hoc reports, that Generative AI helps to set up or configure workflows. These are basically the scenarios where you’re not sharing any data, you let the generated AI based on their training return the SQL query, and then you run the report within your system.”
Sprackett noted that historically, the user would have needed to understand the data structure and where any data is located. Generative AI takes away that complexity with its knowledge.
Generative AI for Sales offers use cases across the four categories above.
Specific items include:
- Composing personalized and compelling emails
- Composing sales copy
- Creating data-driven and persuasive ready-made call scripts
- Creating sales proposals infused with real-time customer intelligence
Generative AI for Marketing helps marketing teams automate and makes possible tasks such as:
- Personalizing marketing campaigns
- Creating landing pages at scale
- Creating personalized emails at scale
- Automatic translation, core Sugar CRM is available in 32 languages. It is also working with the LLM models to support even more
- Improving segmentation with additional, more nuanced information
Generative AI for Customer Service accelerates knowledge and value exchange through:
- Summarizing case history
- Summarizing service tickets from email and chat conversations
- Creating personalized user guides and product documentation
- The natural language interface enables service agents and customers to ask questions and obtain answers to help solve issues quickly
Compliance for generative AI solutions is critical and is at the heart of the SugarCRM offering. Sprackett noted, “One of the foundational principles that we have for our generative AI capabilities today is to keep the human in the loop. We’re not doing automation and sending it out without having humans review it. It’s really more of a tool today for allowing users to be more productive and giving them more information in order to make them more effective in their jobs.”
Where is the data coming from?
One of the key components around generative AI is the data that is used to generate the insights. ET asked Sprackett what the sources of data were.
He replied, “The data sources that are available for this are all of the different activities that exist within Sugar that the customers are entering or acquired through things like email and calendar sync and all of that. We also have additional firmographic information that we layer on, which is pulled from third-party data sources that help to fill in the gaps and see around those corners that might exist in customer data or deal with some of the staleness that can be part of it as well. The large language models also bring a lot of knowledge based on the training set.”
Is data from different customers anonymized and shared?
Sprackett firmly said, “We definitely do not share data between customers. We firmly believe and have been leaders in the field of data privacy around CRM for a number of years now. The data that is provided for a particular customer comes from their CRM, and it’s only used for their purposes. The challenges that we’ve had to deal with are, how do you audit all of that and provide traceability and visibility into what information is being used and how for the customer, so that they can establish that trust that it’s being used responsibly.”
That structure of compliance is important. SugarCRM is delivering explainable AI to users so that they can identify the source of the information used. SugarCRM masks any data sent to LLMs and can provide an audit trail of where the data presented is sourced from.
Another way that SugarCRM is using LLMs is by sending its data model. This allows the coding capability of the LLM to provide the appropriate SQL commands. These will extract the data required by SugarCRM itself. Sometimes, SugarCRM would use a combination of both methods. Sprackett explained, saying, “Building an overview of a business, that would be a combination of our firmographic information that we provide and also some of the general knowledge that the LLM has about a particular company.”
Sprackett recognizes that this could obfuscate where the information comes from. It is why it has built the audit capability into its generative AI architecture to mitigate that issue.
Hildebrand also recognised that there is a risk of accidental copyright infringements or trademark violations. He highlighted that companies are using generative AI to generate content. Also that there is a risk with hallucinations, and AI cannot differentiate between right and wrong in every case.
While this problem exists and the audit can help mitigate it, he gave an honest appraisal of the situation. He stated that “Sugar and software vendors in general and LLM providers need to work on how can you build in a test mechanism which is not available today.”
Enterprise Times: What does this mean?
This is an exciting update from SugarCRM. However, most customers will need to wait a while. When asked about availability, Sprackett answered, “We’re in the closed pilot right now, and we will be generally available in the first half of next year.”
One challenge that midmarket organisations might be wary of is whether they have enough data to feed into a generative AI engine. Spracket noted, “That’s one of the nice things about generative AI. Predictive AI has much stricter volume requirements. We’re able to see benefits on the very first interaction with a customer by leveraging that firmographic information by leveraging the activity information around the customer.
“There’s no, ‘hey, you need 1,000 records before the model can actually work stuff out’. It’s made available out of the box and able to start immediately, providing results to customers, which is amazing. That’s brilliant!”
If SugarCRM generative AI lives up the the expectations portrayed by Hildebrand and Spracket, it could become a game changer for their customers.