Episode 5: AI Readiness for Salesforce Podcast

In this fifth episode of the AI Readiness for Salesforce podcast, Luca Benini talks to Francis Pindar, Founder & CEO of NetStrongHold and AdminToArchitect, as well as co-founder of the biggest Salesforce community event in Europe, London’s Calling. Together, they discuss the three pillars of AI readiness (mindset, data, and governance), how Agentforce can transform businesses when grounded in the right data, real-life use cases of AI agents and a discussion about the shift to consumption-based licensing.

Video transcript

Hello everybody, and welcome to another new episode of our AI readiness for Salesforce podcast today with a very special guest, Francis. Thank you, Francis, for being with us. Do you want to introduce yourself? Not that you know nobody. Well, I know you, but you know this. People don’t know me well, thanks for having me on the podcast.

Yeah. I’m Francis Pinder, and I’ve been in the Salesforce ecosystem for, I don’t know, about 15 years now. I’ve trained over 170,000 people in 160 countries in Salesforce. I’ve been an architect. I’ve been an architect for many years, worked for big consultancies as well as end users, and run a consultancy now, myself, as well as my training company adminarchitect.com.

And yeah, I’m kind of now in the world of AI and see what that has you know how that works. And trying and testing and, and seeing where we can maximize, you know, investment in, in, this agent world. Oh, well, you know, we’re definitely going to go into that. But before that, I want to unpack a couple of things that you mentioned.

Well, first is, and just to make sure that you didn’t actually pull a number just out of thin air, but you mentioned that 170,000 people trained using Salesforce, right? Yeah, yeah, 170,000, which is just mind boggling. And 160 countries as well, which is just like, yeah, talking to those students in these vast different countries… is amazing

the impact that Salesforce has had, like across the globe, which yeah, is is quite amazing as well as, you know, London’s calling the event we do in London that side job of yours of organizing the largest community event in the world for Salesforce ecosystem. It’s bonkers. Yeah. And it’s only like, so was it 6 or 7 weeks away?

Something like that. Maybe two months or something like that. Yeah. June the 6th it’s just around the corner. Exactly, exactly. So let’s actually start with that. So this year London’s Calling is going to be, I guess, you know, a wild guess here. It’s going to be about AI, AI, Agent Force. Right. So today the podcast and the whole theme is about AI readiness.

So what’s the readiness level of the ecosystem and how will they walk into London’s Calling to get more ready, if that’s a word, readier? Yeah. For for for Agentforce. I, I think, I think it’s all about, you know, being AI ready right? And it’s not about having, like, the latest tools. I think it’s more about having the right mindset, data and governance.

So you can kind of turn those agents into kind of value for your organization and into results. And I think it’s almost like pulling back a little bit. We’ve kind of dived into that technology of AI, Agentforce. But now it’s kind of about, right, let’s pull back and see where are we going to get the value from it,

what are those use cases and what do we need to ground that AI in to actually prove it in an organization? And take tiny steps to get there, right. Try it, test it, move on and do that until you kind of got good ROI from it. Because you’re going to pick five options. One of them will be good, four of them won’t maybe.

And you need to find those kind of those, those scenarios that are going to work for you. So I like the, I actually took a note, as you were saying it, you know, mindset, data and governance, like the three key areas to have and achieve, I guess, for AI readiness. Can you help unpack it? Start with mindset

right? So that has to do with what, culture you think, it’s leadership, it’s mindset in the people? Yeah, I think it’s got to start from the top right? C-level kind of being able to kind of take that jump into trying an AI experiment and seeing if it works right? But also having that mindset shift on, it’s not a functional project, you know, it’s not my classic example is somebody had built Agentforce and if you looked at it, it looked like a chat bot.

It was giving no value out. It was giving, you know, it wasn’t what I would expect an agent to do right. And so I think it’s that kind of mindset shift of going, you’re not kind of programmatically really deciding how this thing is going to work. You’re giving it the actions that you and the controls and that trust guard, you know, guardrails on how it’s going to work and then really focus down on the customer and seeing how they’re going to interact with that.

But that’s all grounded on your data. Yeah. And so the mindset shift is kind of don’t look at it as a technical solution, shift up to kind of think about how this is going to add value and how people are going to interact with it to get those best results. And then the yeah, the thinking around data is, you know, you can ground this on data and it may be in Salesforce, it may be outside, it may be an agent to agent integration to get that data.

But you need at some point to ground it on data. And initially that isn’t going to be everything. You’re going to need to target it on a use case, that you can test to see if it’s going to add value. So with that mindset shift, focusing small, and then finding the data that backs up that particular use case you’ve identified and then putting the governance around that to decide, is it successful?

Has it failed? Let’s move on. So that you can kind of turn that kind of the technology implementation into results. And you’re designing the majority of the time, right, and not implementing. That makes sense. And I want to actually come back to the chat bot example because I think we, we, you know, we obviously you explain the mindset and you explain the grounding in the data as an important, you know, component of AI success or AI readiness in our case.

Now I want to unpack data, as in, because we often hear about, oh yeah, you need to have your data. What you know, the chatbot example, what’s data and what type of data level of depth, horizontal vertical depth, you know, completeness, contextualization. What is that data that will make an agent ready and therefore successful? Because I kind of think of it, yeah, as the depth of the data you want, but also the breadth of it, but also thinking around well actually you’ve got structured data, you have unstructured data and you’ve got the model’s data as well

that is kind of underpinning everything and how you can leverage each of those to ground that, that agent and those prompts in the best data to solve that specific use case rather than going, hey, we’ve got everything let’s chuck it all in. It’s kind of I think of it is finding that kind of narrow data source across structured, unstructured, and the model data and how the prompts and the AI can leverage each of them to solve that.

So through that use case or use cases, you want to test to see if you get the ROI, and then and then get it back. But I think a lot I think there’s been quite a focus on structured data, which is fair enough. But for me, I think where the value comes is, yeah, you’ve got the key structured data within your Salesforce.

Org, but the value I think comes with all the data in the LLM, right. How can you harness that information to provide it back based on the customer personas or your user personas that are accessing the agent to give a much richer experience right? And that could be, yeah, at the LLM’s data. Or it could be your own corporate data, you know, all the process documentation or whatever it may be that you’re doing in that particular use case, without, you know, chucking it all in.

But I think people miss and yeah, they focus on the data within the org. Yeah. Then the next one is they’re focusing on, oh, the knowledge base articles or the static, the unstructured internal data, but then they kind of miss out on the value you’re getting from the, from the LLM, and the data you could potentially use or ask to search things back, to give a much valuable, more valuable rich response than you would necessarily with just your customer data.

And, yeah, unstructured data within your organization. Very, very interesting. Now I’m going to challenge you to give me a practical example of everything you said. Right. So we talked about a chatbot. What’s a what’s a practical application that you’ve seen of a chatbot that actually leverages the internal org data or, you know, structured knowledge, base data or external land based knowledge that is pre built and existing in the in the model.

And then use that to answer questions, I guess, or two. Yeah. I think drive the conversation. I think a really good example would be one company where they’d, they’d built their agent and and it is around, real estate or estate agencies. And they basically developed it as if it was a chatbot rather than an agent.

So it asked the typical things, where do you want your house, buy a house? How many you know, bedrooms do you want it, blah, blah, blah. And it was kind of going, this thinking of everything very functionally and data driven as you would kind of in a Salesforce org database. Yes, I this is how I’m going to search through data and respond.

And here are the full houses we have that match your requirements. Exactly. So that’s the old way. Yeah exactly. And then and then it’s like and they built it in Agentforce like this. And it was like where’s the rich data? The you know the information that I want to get back. And also what do you what’s the problem you’re trying to solve?

And in, in the estate agent world, basically. Oh, the majority of people think they know what they want when they buy a house, but actually don’t, right? So, and they might just fall in love with, oh, the river outside. And actually, all the requirements of the house, it’s got nothing to do with what their original requirements were, right?

And actually, as an estate agent, what you really want to find out is, you know, their current circumstances, they’re, you know, single are they, you know, looking to build a nest, you know, have they got a big family, you know, and based on that information, you’re kind of pulling out. What persona do they kind of fit into?

And then what is the value we can give back to them to help them trust us more that they want to reach out to us and use this as an agent. So, for example, if I’m asking them, oh, you’re looking for a family home in this area, I could say, well, actually, if you look, 200 yards down the road or a mile down the road, you’re actually in the catchment area of three top schools.

Now, all that information is potentially in the LLM realm, right? So you can bring that information back to them and go, oh, we’ve got some great houses. You know, a mile down the road. It isn’t in your specific area, but it is next to these three great schools. We’ve got this here, that here, blah, blah, blah, blah.

And then ask them questions around their personal kind of experience, what they want to get out of the house in their current, circumstances to better inform, you know, local information and, that can really go, wow, this is really cool. Oh, actually, I never even thought about the local schools. Oh, yeah, we’re thinking of having a family.

Oh, yeah. Primary schools suddenly jump to the top of the list. It’s not the number of bedrooms, right. And being able to inform people with that. And that’s where I think is, is, you know, it’s a killer, right? Between different, you know, if you get that experience compared with somebody who hasn’t got that, who are you going to reach out to?

You know, it’s going to be that estate agent, right? I like the, you know, the way you positioned it, which is which is the ultimate goal. Right? So going beyond just helping you in the immediacy of, you know, this is my listing or, you know, these are the listings that are matching whatever criteria you enter. But really thinking about a very smart estate agent that has the ultimate goal of creating trust.

And, you know, in sales, you would call that a good discovery call, right? Where you’re doing a lot of listening and you’re allowing the conversation to go off in directions that really follow the pain point or the ultimate goal, rather than whatever a client that it’s talking to you. And I think every day they think they want, like you said, right.

So sometimes they have a list of features that they want to double check that you actually meet. And that would be, you know, applying in the real estate, you know, five bedrooms and, you know, two, doors opening on the backyard or whatever, whatever that is. You know, that would be a mansion, actually, that I just. Yeah, that drives the whole AI.

Right. So if you’ve got that unstruct data. So, for example, that unstructured data of the call right of oh yeah that, that this type of person blah blah blah blah, that drives that whole experience there going forwards. Yeah, maybe the emails are slightly tailored than that to that or, you know, the documents I create, tailored to, you know, their circumstances rather than just giving them, here’s the property profile, right.

That doesn’t mention schools. Yeah, it kind of cascades all the way through, to get there. And I would imagine that if you brought in that, you know, again, following the same example, there might be news in the area that are relevant, you know, hey, there’s a new the road closure or rebuild the park. And, you know, they’re updating the, you know, playground down the road, the, you know, the park on the other side of the street.

And, you know, all of a sudden it’s relevant information that wouldn’t have nothing to do. But an agent will be able to, if instructed in the right. Exactly. Yeah. If it’s. Yeah, use that information in the same way that if I’m a single person buying a property, then it’s going to go, hey, look, you know, there’s a Michelin star restaurant just down the road from this property, you know?

Oh, cool. So, coming back to the kind of AI readiness. So how how ready are clients to have this type of conversation, to trust an agent, to have these type of conversations with their customers? And how ready is Salesforce to cater with Agentforce, to these type of scenarios?

I think it’s it’s definitely built on the trust of the information you get out of it. Right. Especially when you’re looking at AI, adoption and how you use that. So this is why I kind of think that you’re starting small on a specific use case, right, where you’ve got people in a specific department that already get AI, or they get the value of that before you go to the people that are a little bit more hesitant and don’t want to be involved with it and that you kind of slowly do that kind of incremental progress on it.

And I kind of say AI rather than Agentforce because I think I it does need to be a kind of looking at your AI strategy as a whole, where it adds value rather than looking at the technical enablers for an if Salesforce is always and ready. I think this there is a big challenge in the Salesforce ecosystem for me is around the shift from kind of user based licensing to, consumption based licensing, which I don’t think Salesforce has really explained brilliantly

well, and I think this has had a negative impact on Agentforce. For me, I think it’s great because the more you know, you can prove the ROI based on, you know, the, the how you’re consuming it, you can then know what the scaling up is going to cost you. And you’re going more traditional I and, you know, traditional I.T of AWS, which is all consumption based.

Right. But I think a lot of people in the Salesforce ecosystem don’t know how to architect or build solutions using this consumption based model. Right. But I think, you know, I’ve heard, you know, there’s our safe harbor, you know, there’s there’s, AI calculators for… Salesforce is great at calculators for working out all these credits based on the volume of data and the transformations, everything within Data Cloud that you need.

So you can kind of more accurately work out your ROI. But this is where I think, you know, incremental progress is so important. Right? I’m really looking at it holistically across the organization and see where you can add value. And that may not even be Agentforce. It could be another AI tool that you’re using to get the trust of AI, into the organization.

And then to be honest, the users won’t know the difference between if it’s Agentforce or if it’s some other tool right. As if it’s getting the results they need, then. Awesome. Yeah. And they can get to get quicker then that’s brilliant. And then you’re kind of essentially where I started right at the beginning, that mindset shift.

Yeah. It’s getting people into that mindset of what AI is and what it can do for you. Right. And how it can improve the way you work. And it’s not going to take away your job. So, I think we’re still quite a way from… there are certain industries, I think yeah, obviously a little bit more dodgy than others, but,

yeah, I think there’s still a way to go.

Very interesting. You mentioned the consumption model, and you know how, Salesforce didn’t necessarily do a great job in communicating about it and creating some uncertainty around, you know, the only certainty that I get is it’s expensive. And then, you know, from there I think it was. Right. So so that’s a interestingly, I believe that Salesforce has built its success on results, you know, I, you know, you and I have been to and you organize, you know, as well events, Salesforce events where the first thing that you hear and you see is always happy customers mentioning some form of stats of

the results that they’re getting. Yeah, 34% more sales is the, good old, impact. Right. And and based on that then immediately justifies the investment in the technology that they’re proposing. So historically they’ve been selling like that. I think what you’re saying is, is is correct. They lost a bit that narrative, in the ability to ground the success into without talking about the cost.

I think they were forced down the road of talking about the cost without being able yet to explain the ROI impact. Expected. Right. So we all know it cost $2 per conversation. Some of them. It’s a cheap right. And I think and I think this is the this is the ultimate thing, right. Is you know, AI you know, AI in Salesforce isn’t about replacing people.

It’s about empowering them to do more faster and smarter. Right. And so if you put that hat on, it’s like saying, well, okay, it costs me $7 to push a ticket to third line support. Yeah. So how much deflection am I getting from pushing it to third line using AI, am I having right? And what is the ROI you know, have I got a good ROI on that?

And so I think that is the kind of stats and trying to really do the working out at the beginning of what good looks like or what success looks like when you’re implementing AI and you do that design up front, such that you are ready to implement it into whatever tool you you would like, hopefully. Does that answer your question?

I got I well, I absolutely I think it does because it’s I think your premise is, quite right for someone in your position dealing with data in a way at the level that you do. But your assumption is that most companies have worked out with that level of detail, what what is the cost for them to manage different parts of their processes, and therefore it they can immediately cross compare.

What would it cost if I were to do it with. And I think that’s where the small steps approach. And then you mentioned earlier will will go a long way because, you know, I start dissecting parts of your process if you know exactly what that is costing you, then you will be able to see, you know, does it actually make sense for me to replace it with, with an AI, will it accelerate my success, my path.

Exactly. And it’s going to be it’s augmenting those kind of human capabilities to kind of free up the teams to do high value work and do more of that customer relationship building based on, you know, that grounded information of knowing, it’s a family that needs to house or whatever it may be, right? So you can make more meaningful conversations, which will hopefully convert more sales.

And away we go. Right. That’s that’s absolutely right. So building on that, what’s what’s next? What do you see happening. You know, 2025, you know, after around and after London’s Calling. So kind of I mean, the second part of 2025, what’s, what’s we I, I think, I think what we’re in at the moment has been a bit of,

Oh, my word. It costs lots. We don’t really know, understand it. I think it’s it’s kind of it’s that understanding of how AI is going to benefit an organization and where it can benefit. And I think once we’ve hit that realization, then you will then have a big disruption happening. I think within the Salesforce ecosystem that could be one of them, could be just around the AppExchange, right where, I was talking to somebody a couple of weeks ago, and literally this app’s been around for a very long time and it’s not going to exist.

You know, in a couple of years time, I don’t think. Right. But it’s a very, you know, fundamental app. Basic app. Right. But it’s going to get replaced by, by an AI agent. Right. And I think the but the ones that are going to be really valuable, that are going to really kind of start pushing is the ones that are collecting data.

Right? That’s it’s people will suddenly realize the value of grounding their AI, structured or unstructured and suddenly realize, hey, there’s a lot of unstructured data, more data that we can start getting to really ground them. And actually a CRM hasn’t got great data in it, but we can ground it on lots of other things around the customer.

It could be integrations to, I don’t know, other, you know, other data providers to bring it in based on the location of the person or the company information or whatever to get much better AI journey, and so I think there’s going to be a bit of flip. Yeah. AI realization and then going right to get real value out of this, we need to get much better grounding on data both inside and outside of our org.

And where that comes from and how that’s brought into Salesforce is another thing, right? Maybe it doesn’t at all. Maybe you’re just doing the actual, reality is going to be you’re going to be doing an AI, you know, agent agent integration, where you can be asking Agentforce to talk to another agent to get the relevant information back, and then bring that into the prompt with an Agentforce to then answer the questions.

So rather than pulling all the data into Data, Data Cloud, who knows. Right. I think we’re still and this is I think the beauty of it where we are right now is people are just kind of trying to find their feet, and all the, and working through all these different ways of doing AI so that we can really kind of learn from each other and find out what the best approach should be based on different scenarios.

Yeah. And to paraphrase what you just said, I think people are and companies are getting ready to embrace AI and, ultimately bringing it into their reality, a reality of mindset, reality of data as you just were explaining and reality of governance and then unleash its power and, and enjoy the, the benefits, hopefully. Yeah. In the short term and the, the, the future of, you know, Salesforce AI are the ones that are going to be ready.

Right? You know, where strategy isn’t just a roadmap, it’s, you know, it’s that ticket to your success with AI, right? So yeah, it’s exciting times. Definitely, definitely exciting times. Look, this was, incredibly interesting. Thank you for, sharing all your knowledge and, you know, kind of bringing it down, I think, to very, real examples, real, tangible, practical, examples.

Well, I’ll see you at London’s Calling, I guess. Any anything that you want to say to, you know, our, our audience regarding London’ Calling what we expect. Yeah. I look for apart from a Vinton Demo Kam that. Oh, yeah. Oh, yeah. Absolutely. I do love a good Demo Jam, and it’s always one of my favorites. But the yeah is our 10th anniversary, so we are having a I don’t think I should say there’s something happening at the end of the day, which is quite cool.

Okay. So let’s stick around. Yeah, yeah, I’ll be there. Yeah, it’s quite cool. And, yeah, we’ve we’re literally maxing out the venue again, even more so than previous years with I think we got like 8 or 9 tracks of talks this year. I think it’s bonkers. But if you do want to, if you haven’t got a ticket yet, you can grab one in person

ticket from londonscalling.net, and it’s on the 6th of June, hopefully Friday the 6th of June. I’ve got the date. Right. And yeah, and we also will be launching the online tickets as well. So if you’re anywhere else in the world and you can’t get to be here in person, then you can sign up for a, online ticket which gives you access to all the sessions during the day and the hubbub, then voting in the demo jam and all kinds of stuff.

So, yeah, should be a lot of fun. Yeah. I’m really looking forward to it. I’m looking forward to learning more about, you know, AI readiness examples and, you know, practical ways of bringing Agentforce and I guess in general, AI and productivity and success into every Salesforce implementation, every business that uses Salesforce, I guess. Yeah, it’s going to be good.

It’s going to be good. Well, thank you very much again. It was a pleasure having you, Francis. Thank you for sharing all your knowledge with us. And, well, I’ll see you. I’ll see you at the brewery around the corner. Thanks for stopping by. Super. Thank you. Bye bye.