Salesforce is rounding out its plans to infuse generative AI technology across its product range as its World Tour roadshow reaches New York today. As well as extending the capabilities that Einstein GPT brings to Sales Cloud and Service Cloud, it is launching Slack GPT, which will bring native AI capabilities into the team messaging platform, along with connections to Einstein GPT. Meanwhile, to help meet the implementation and change management challenges enterprises will face in adopting this nascent technology, Salesforce has teamed up with tech consulting group Accenture to create an acceleration hub for generative AI.
Slack GPT marks a significant step up from the existing Slack integrations with generative AI apps ChatGPT and Claude, which currently simply allow users to submit prompts and follow-ups from within a Slack channel. It will be available natively in Slack to provide generative AI capabilities such as summarizing a messaging thread or a huddle conversation, or drafting a message or Slack canvas document, and customers will be able to specify their own choice of language model, including their own in-house models.
A new Einstein GPT app for Slack will allow Slack users to query real-time customer data held in Salesforce using the CRM-specific capabilities of Einstein GPT. For example, a sales rep could request a summary of information about an account, including primary stakeholders, next scheduled meeting, and other key facts.
The newly released Slack platform becomes AI-ready, allowing developers and users to connect AI actions into no-code workflows. Customers will be able to integrate a range of Large Language Models (LLMs) in this way, ranging from those supplied by OpenAI and Anthropic, to models developed in-house, to other third-party models including those developed by participants in the Salesforce Ventures generative AI fund, and in the future, Salesforce’s proprietary LLMs.
No-code automation, with added AI
Customers will be able to get their hands on the first of these new capabilities this summer, when Slack GPT AI connectors for Workflow Builder will become available. There will be a longer wait for Slack GPT and the Einstein GPT App for Slack, both of which are currently in development with no date yet given for when they will become available. Ali Rayl, SVP of Product at Slack, says:
Developers today are able to start building these building blocks. Then later this summer, anybody will be able to pick up these blocks and start putting them together. The goal of all of this is to bring no-code automation to everybody in the business, supercharging that with generative AI.
For example, a user could create a workflow that starts by with a prompt created within Slack that asks ChatGPT to create a tweet on a specific topic, posts it in the channel for review with an approval button, and then pushes the approved tweet to a social calendar in Google Sheets. Or once a Pagerduty incident has closed, there could be a workflow that uses the LLM to automatically summarize the incident from the message thread — including start and end times, who was involved, root cause, how it was solved and so on — and then record the summary in a canvas document. She adds a final example of time-saving automation in a sales context:
The idea of a new lead coming into Salesforce prompting ChatGPT to conduct qualification questions, draft a prospecting email, consolidate that in a Google Doc, send it to a channel and tag a sales development representative to get involved.
This is normally the kind of work that can take a sales development representative a full half hour to pull together.
Choice of LLMs
The ability to bring in other LLMs is particularly relevant in many of the verticals that Slack customers operate in. Rayl says:
Some people want to use ChatGPT. Some people want to use Claude. Some companies are already going out and building their own custom models. Think about, for example, a legal firm that is training their own custom model against their entire case history. Or think about a company on the leading edge of biotechnology and medical research. They’re going to want to have a very specific LLM and training model to their domain, their expertise, and their information …
With the platform and the recomposable building blocks, not only can companies choose which LLM they want to integrate with, but they can also choose different models for different purposes … On the back end, whoever’s creating those workflows could tie it to the correct LLM.
This allows companies to not only choose to bring in the publicly facing LLMs like ChatGPT, but also to bring in their private LLMs, which maybe another company is hosting for them [or] maybe they’re hosting it in their own infrastructure. Really this is all about … giving companies the choice and the options to use the model or models that are relevant to them and their businesses.
Developing strategies for AI-powered automation
The broader partnership between Salesforce and Accenture has five main elements:
- Working with customers to develop customized AI strategies for productivity and growth, using new accelerators for Einstein GPT, user interfaces and process automation.
- Developing new use cases to showcase how the use of generative AI in sales and service contexts can help improve productivity and customer experience.
- Building industry-specific AI models for companies in financial services, health, manufacturing and the public sector.
- Using Salesforce data cloud, exploring how generative AI can learn from customer data to create more personalized experiences for customers.
- Develop learning resources to help teams develop and enhance their generative AI skills.
The need to press ahead with implementing new forms of AI-powered automation is highlighted by the findings of a new Slack State of Work Report published today, based on a global survey of more than 18,000 desk workers. It finds that those who have adopted AI in their work are 90% more likely to report higher levels of productivity than those who haven’t, while over three-quarters of all employees (77%) are keen to automate routine tasks to improve their productivity. But the report finds that barely a quarter of companies (27%) are using AI tools to do so.
There’s much more in the report beyond the impact of AI and automation, including a finding that over a third of workers (35%) cite spending too much time in meetings as their top productivity challenge, while 43% feel meetings could be eliminated with no real adverse consequences. Yet only 19% of leaders are encouraging asynchronous alternatives for collaboration. Employee experience is also highlighted, with four out of five workers (82%) saying that feeling happy and engaged with their organization would improve their productivity, while 43% of people managers cite helping their team stay motivated as a top challenge.
The rapid march of generative AI into enterprise applications continues at pace. Team collaboration and knowledge work are areas where the benefits are self-evident, provided the accompanying risks aren’t overlooked. But the tie-up between Accenture and Salesforce highlights an important point — technology alone doesn’t solve anything unless it’s introduced in a way that allows people to adapt to it, reskill, and maximize the benefits of new processes and working patterns. There’s a crucial role here for consultancies across the Salesforce ecosystem and beyond to take a lead in helping make the most of these new generative AI tools.