Salesforce takes workflow into the generative generation with new Einstein GPT and Data Cloud integrations

Salesforce has announced plans to integrate Einstein GPT and Data Cloud with its workflow automation offering Flow, so that users can create and adapt automated workflows using conversational instructions.

As per the official announcement: 

Data Cloud for Flow links the company’s workflow automation with structured and unstructured data stored in a lakehouse, which is a cross between a data lake and a data warehouse. Customers can combine data from Salesforce’s databases and other sources to harmonize customer views and apply low-code tools to create automated processes and workflows.

Einstein GPT for Flow helps business users and admins work more efficiently to build automation powered by AI with better usability and accessibility, and increased speed.  Business users and admins could describe what kind of flow they want to build and see it built for them in near real time, instead of having to build each flow manually, step by step.

At Salesforce’s World Tour event in Washington, Chief Technology Officer and company co-founder Parker Harris drilled down on the importance of automation, which he argued had been part of the underlying focus for product development since the firm's inception: 

In 1999, it was just salesforce automation. It was a web browser and the internet and people were like, ‘Wow, that's incredible. I can go and log in and I can enter information in a web browser and track my leads, track my opportunities, forecast my sales’. But what was important is that we helped you automate that, that you don't have those manual processes. 

So we added declarative workflow with workflow rules. And then [customers] said, ‘Well, that's great Parker, but I need a little more power’. So we said, ‘Fine, we'll give you a programmatic workflow with Apex. We'll give you approval processes. We’ll give you Flow Builder. We’ll give you Robotic Process Automation. If you've got mainframes, we can connect those, as well as green screens, and we can connect it as a slow orchestration.

All of this has now come under the umbrella of Flow, he went on, pitched as ‘automation for everyone’. It’s a message that’s resonating with customers, Harris argued, citing 44 billion Flows happening every day, saving over 109 billion hours a month, and generating over $2 trillion in business value each month.

Flying high 

Backing up the importance of automation was a use case centering on NASA. Aris Mondey, Platform Services, Ocio Application Chief, said: 

Automations are the key of how we're going to digitally transform as an agency. The Salesforce platform is at the heart of how we're going to digitally transform the ability to embed capabilities inside of our customer business units, to be able to leverage the power of the platform and the automations in order to make that transitive move to more modernized and digitized environments.

The Federal Government sector has had a difficult time when it comes to maximizing IT investment, he went on: 

[It’s been] focused on historically monolithic systems that were fundamentally expensive to maintain and support security-wise and otherwise, getting top heavy. [We are] trying to turn that over with automations and the ability to onboard things in a quicker more efficient fashion. That then allows us to recoup some of that capital to reinvest back into the platform, to create more capabilities and more modernization. So the upstream environment has been a great success and a great shining star, if you will, for what we've been able to do.

Leveraging the power of a single customer view has been particularly beneficial, he concluded: 

The ability to take disparate data sets and consolidate and centralize to understand what our customer is, to have that singular view, to be able to inform, better recruit, better go out and and really enable the workforce to understand or [have] NASA understand how we're touching people, what our stakeholders are doing, what the engagements are, where we're making inroads and campaigns and activities. From a Federal sector standpoint, the ability to recruit is a big deal. We have certain restrictions with salary and other things that sometimes doesn't compete with the commercial sector, but at the same time, bringing these recruitment tools in allows us to better leverage and understand what the populace needs, so that we can then return back to our own budgets or Congress.

Generative

Salesforce pioneered AI for CRM, argued Lynne Zaledonis, EVP for Cloud and Industry Product Marketing, and now has AI in every cloud offering in its portfolio, with 215 billion Einstein-generated predictions every day. Next up is generative AI, with Zaledonis pointing to the business potential of such tech, as well as some concerns: 

The possibility of increasing productivity and changing the way you engage with your customers. Imagine a public sector organization can take a very complex, multi-step process and simplify it for their constituents by using a chatbot. But generative AI is also raising a lot of questions. People are concerned about the data - where it comes from? Does it know me? Does it know what I want? Can I trust it? People are worried about following it blindly.

But, she went on, what could happen if generative AI could learn from trusted customer data, data about prospects, partners etc? This has led to Salesforce's Einstein GPT, positioned as “the world’s first generative AI for CRM”. Zaledonis explained: 

This is AI trained on your customer data, your data, your models, your experiences, all within the trust, you know to expect from Salesforce. Einstein GPT is going to let you create messages for sales reps and service reps automatically. For your marketers, you'll be able to get a landing page with just a blink of an eye. And for developers, you'll now be able to get an AI helper to help you create Apex code even faster.

She added: 

It starts with your customer data, that's your CRM data, your Slack data, telemetry data, all within the trusted boundary. Then we train the AI model on your data. Because it's an open framework, you can use our Large Language Model, you can build your own, you can even use one that you might already be using from the [Salesforce] ecosystem - it doesn't matter. But here's the important step - there's always a human in the process, somebody who reviews and approves the information and the outcomes before it's sent out, because AI is here to help us, not replace us.

My take

The generative AI aspect of this week’s event in Washington is something that we’ll be hearing a lot about as the World Tour program rolls out around the world, climaxing inevitably with Dreamforce in September. The NASA use case was a compelling one. The disappointing track record around ROI in tech investment is not restricted to the US Federal sector. Governments around the world have carried the burden of soaring budgets and low value in return for decades. If automation and AI can be used to get more bang for the taxpayer’s dollar/pound/euro, then so much the better.