Description
In part two of this series, we explore Salesforce’s five AI readiness factors and what it takes to move beyond early Agentforce experiments. With practical examples and lessons learned, you’ll see how aligning strategy, data, governance, platform, and culture helps teams build agentic solutions they can trust and scale.
Transcript
Welcome. My name is Jack McCurdy. I'm DevOps advocate here at Gearset. Today for part two of this webinar series, it is a pleasure once again to be joined by Dave Rant, the Agentforce Engineering Lead here at Gearset, and also to be joined by Ben Coleman, is the founder and CEO of Performa.
Welcome to you both. It's a pleasure to have you here. For the attendees that were not here as part one or as a recap if you were here as part one, we discussed in the first webinar a few key points. We talked about CIOs being under pressure to innovate with AI, but ninety five percent of organisations are getting zero returns.
So there's a gap between the ambition and the execution in this space right now. We talked about the agentic maturity model that maps five levels from rule based chatbots all the way to multi agent orchestration. It's a journey, not a jump across this model. We talked about secret escapes and their progress from ten percent auto with Einstein bots to a forty five percent deflection rate with Agentforce.
What took six months with bots took two weeks with agents. We talked about agentic maturity and how it starts with DevOps. Teams with those strong DevOps foundations adopt AgentForce faster and more confidently. And that Gearset's capabilities span the full DevOps lifecycle for AgentForce, org intelligence, sandbox seeding, data masking, code reviews, automated testing, data archiving and observability, a whole suite in the Gearset product that can help with your Agentforce journey.
We also talked about the four questions to answer before building your first agent. What's the business goal? What's your risk appetite? And what data does the agent need?
And how people are going to use that agent?
We also gave some homework at the end because everybody loves a little bit of homework, but we asked you to do a bit of an ideation exercise between the different levels of the Agentforce journey across level one, level two, level three, and of course level four. And if you've had time to do that, I hope it was insightful and useful for you and set some context about where you are in your journey and what you might need to get there. But to kick things off in release readiness, I'm gonna throw it over to Dave to take us through. So Dave, floor is yours.
Thanks very much, Jack. Thanks for coming on for part two folks. Pleasure to see you all again. So in part one, we covered the maturity levels and what that journey looks like, but none of that matters unless you can figure out where you are today and what's getting in the way of you making progress. So today, I want to introduce you to Salesforce's concept of AI readiness factors. These are five things that keep cropping up when you look at the organizations that are making real progress with Agentforce compared to the ones that are still stuck in pilot.
The organization seeing meaningful impact from AI are the ones that are moving on all five of these at once, and we're seeing this firsthand at Gearset. We've got over three hundred and seventy customers who are deploying Agentforce and Data Cloud today, and sixty percent of them already have agents in production. And to us, the pattern is really clear. The teams with strong DevOps foundations have already got the visibility, the automation, and the guardrails in place to move confidently.
Not starting from scratch, and they're ready to adopt Agentforce.
So here's the full picture. We've got five factors down the left and five maturity levels across the top.
Each one of these factors breaks down into three core drivers. So for example, strategy isn't just one thing. It involves your AI strategy, your business alignment, and your execution.
And that matters because most organizations won't be evenly mature across a single one of these factors. You might have a decent AI strategy, but no ability to execute on it. And knowing that will change where you focus your time.
Now, these five factors don't work in isolation. Your strategy is only as good as your data. Your platform doesn't matter if nobody knows how to use it, and your governance falls apart without the culture to back it up.
And it's worth acknowledging that where most organisations actually sit today. Deloitte found that while thirty percent of organisations are exploring agentic options and thirty eight percent are piloting them, only fourteen percent of those organizations are ready to deploy and just eleven percent of them have agents actually in production.
So if you're looking at this thinking, we're mostly in the first couple of columns, then don't worry, you're in good company.
But the point isn't to jump straight to level five, it's to figure out which factor is your bottleneck, because that's the one dragging down everything else. And often it's not the obvious one. Teams might have a decent strategy and access to an agentic platform, but they can't deploy reliably, they can't test properly, and they don't have the visibility into what's happening in production. And that's the gap that we keep seeing. So let's dive into each factor in a little bit more detail.
There's a statistic from McKinsey that I always come back to for strategy.
Organizations with a formal strategy hit an eighty percent success rate for their AI projects, and without one it's only thirty seven percent and that's a big gap.
Most organisations aren't starting from a formal strategy though. They're starting from someone in service who stood up an Agentforce employee agent and someone in sales who's experimenting with an SDR agent, And those two efforts have never been in the same room together. Nobody's connecting them to business outcomes. There's no dedicated budget and there's no clear ownership. So the first meaningful shift is just to formalize it. Get executive sponsorship, set a budget, and attach KPIs. That's where you start seeing real numbers.
With AgentForce deployed to production, Gearset customers like Reddit have achieved a forty six percent case deflection rate, Vionic saw a twenty five percent drop in average handle times, and Zotus saw a forty percent increase in lead capture.
But the really interesting shift happens further along when your strategy stops being how do we use AI to do things faster and becomes, what could we do now that we couldn't do before?
The handful of organizations right at the top are treating Agenetic AI as a way to deliver a wider range of value to their customers alongside their people.
They're not just deploying agents for their current workloads, they're expanding into new markets and redefining their product market fit in the process.
And on expertise and culture, Deloitte surveyed over three thousand leaders at the start of the year and found that the skills gap came back as the single biggest barrier to AI integration.
Last year, Salesforce found only twenty eight percent of employees knew how to use the AI tools available to them. There's a real disconnect between what organizations are rolling out and what their people are actively consuming.
So what stands out at the top end though is when expertise stops looking like a training program and starts to define how the company operates.
AI first companies are going to keep seeing greater productivity gains over time, and eventually, that will become the difference between keeping pace and falling behind.
Now according to Salesforce's AI trust survey, eighty percent of organizations have already run into risky agent behaviors, things like improper data exposure and unauthorized access. But only one in five have got mature governance for those autonomous systems.
The good news here is that the Salesforce platform gives you solid foundations through the Einstein Trust layer. Every AI interaction goes through a pipeline of secure data retrieval, grounding to reduce hallucinations, PII masking, prompt defense, toxicity scanning, and so on. You don't have to build that yourself. It's there out of the box for you.
But platform level guardrails aren't enough on their own. You need your own governance on top, like acceptable use policies and risk classification for your AI systems.
And this is becoming a regulatory requirement too, with the AI Act's high risk system rules coming into effect in August later this year.
For the top organizations here, governance shifts from manual to automated. For example, policy as code, real time alerting, and role based access control. The most progressive companies are even starting to adopt Guardian agents where AI is being used to monitor itself, which does kind of make sense when you think about the scale of what we're heading towards.
And the technology platform is where Salesforce's investments have really come through.
Agentforce now has processed over three point two trillion tokens, and it's built on a composable architecture of the Atlas reasoning engine, agent builder, data cloud, and the Einstein Trust layer, which are all designed to grow with you.
The important mindset shift early on is that agents aren't deterministic. You can't just test the correct outputs like traditional software. You're testing for emergent behaviors. And that's why the DevOps life cycle and shifting left matters so much.
You need to be running your agent through hundreds of parallel conversations within your test suites just so you can validate how the agent actually behaves in the real world.
The real unlock though for Agentic platforms is interoperability.
And by giving agents access to the tools they need and the ability to communicate with each other across systems, you'll be giving your systems the autonomy to solve complex real world problems.
And lastly, your data foundation underpins everything else. Your strategy, your team, your governance, your platform won't particularly matter if your agents can't get to the right data.
Most organizations are starting from a challenging place, and according to MuleSoft, only twenty seven percent of enterprise applications are interconnected, and agent knowledge is broadly limited to whatever is sitting in your CRM.
Even the data you do have might not be in great shape. Duplicates, stale records, inconsistent formatting, That matters more with agents because they'll act on whatever they find, and bad data at scale means bad decisions at scale.
So we've walked through each readiness factor quickly and what maturity looks like at each level.
The question is, what actually helps you move up? What practical steps can you take to unblock the next level? So I'm gonna run through each factor and show you where Gearset fits, and you can do this as a take home exercise by scanning the QR code at the top of this slide.
Figure out where you are today as an organization and where you need to be.
The biggest blocker to strategy execution is that teams don't actually know what they've got. You can't plan an agent if you don't understand your org, what metadata exists, what business processes are in play, what data your agent needs to touch. Gearset gives you that blueprint. It scans your complete org architecture and lets you ask ask questions about it. So you skip the weeks of discovery and start building with a clear picture.
And as you progress and need to start working with multiple agents across multiple domains, you'll need the ability to deploy and iterate fast enough to keep pace with your strategy. And that's why having proper CICD pipelines, version control, and repeatable deployment processes becomes the difference between a strategy you can deliver on and a strategy that stalls.
With expertise and culture, the skills gap is the blocker early on.
People don't know how to build agents, and they definitely don't know how to test them. But with Gearset, admins, QAs, and developers can all deploy agents and run tests without writing code. And that brings more people into the process, and it builds confidence across the team.
Knowledge discovery helps here too. And instead of needing that deep tribal knowledge of an org to understand how it works, anyone can just ask the Gearset agent and get answers.
When you're starting out on security and guardrails, you're unlikely to have much formal governance with agents interacting with data in uncontrolled environments. Two things can help here immediately. Firstly, Gearset can scan every code change against frameworks like OWASP and the well architected framework, catching vulnerabilities during development rather than in your production environments. It can also protect sensitive records in your sandboxes, either masking them as you seed them or obfuscating the data that's already there, so you're never building agents against real customer PII.
As you scale up and governance becomes more automated, these become part of your pipeline rather than manual checks. Every deployment gets scanned, every sandbox gets masked. It's consistent and enforced, not dependent on someone remembering to do it.
And to build confidence in the agents you're running on your Authentic platform, Salesforce provides testing center for validation out of the box. But the reality is your agents don't run-in isolation.
They're calling flows, they're running Apex, and they're hitting other integrations, and that layer needs testing too.
Gearset's automated testing is different from traditional robotic testing tools. It's AI driven, so it understands the semantics of what it's interacting with, meaning your tests will stay resilient as your UI evolves.
But the bigger challenge comes once your agents are alive. They're dynamic and their behavior shifts as data and prompts change. You need to know when something's gone wrong before your users do.
And with Gearset, you can surface every flow and apex failure across your orgs, track your org limits, and put it all into one place. You can see the problems early, you can spot the patterns, and then you can respond quickly. So while the Agentforce platform gives you the power, Gearset is giving you the confidence to keep it running.
And at the lower maturity levels for your data foundation, your agents are limited by whatever data they can access, and it's often messy. Gearset keeps your orgs lean by clearing out old and unused records, so agents are only working with relevant high quality data. Then there's the challenge of testing. Agents are data hungry, and if you're testing agents against a handful of records, you won't catch the edge cases that break things in production.
So use Gearset to populate your lower environments with realistic production like data so you can test properly.
So those are the fibrogenic readiness factors. I want to leave you with this thought though.
We get the maturity model and readiness factors right, where do we actually end up as agentic organisations? Well, here are some of the possible behaviours you'll start to see emerging in your workplace.
Your people get to focus on work that energizes them with creative and strategic challenges.
Agents will take on the mundane and repetitive tasks.
Humans and agents will collaborate regularly with humans maintaining oversight and becoming managers of agent workloads. And lastly, you and your colleagues will confidently delegate routine work to intelligent systems rather than being inundated by it.
So using the maturity model and the readiness factors, the foundations you put in place today are what will make this all possible in the future for your organisation.
Thanks for listening, Back over to you, Jack.
Amazing. Dave, thank you so much for introducing us to those readiness factors and the things that the folks can do to prepare themselves for the next stage or the start of their journey. Really, really helpful.
Now's the time to throw it over to Ben. Ben, I want to bring you into the webinar and discuss some of the real world things that you are seeing at Performa when introducing agent development and Agentforce development on the Salesforce platform. It's a pleasure to have you Ben.
Thanks, Jack.
Okay. So I'm gonna start by, as Jack says, just just basically talking around very briefly, because I know we've covered it before, the maturity model, but just a few of the things that we're seeing around that and our thoughts on it. And then I'm going to go into the readiness factors and how people are tackling it and what we're seeing.
So obviously, we build anything sorry, previous slide.
Before we build anything, we need to assess where we are.
And I think lot of what we're seeing is that companies are around that kind of level one, two at the moment. So they're performing simple kind of information retrieval or sort of siloed tasks.
You know, it might be using like leveraging knowledge, for example.
And our goal at Performa is really to help you on your journey to get to level four, so the multi agent orchestration.
And I think, you know, as we're talking about here, it is critical to have a really robust DevOps process and tooling because otherwise you're going to really struggle to get to that kind of maturity level.
You know, it's really difficult to orchestrate complex AI if your underlying Salesforce data and metadata are in chaos.
Next slide, please, Jack.
Thank you.
So what we've got here is just an example of what we might do at level one. So this is where we see agents assisting humans by retrieving policy information. So that could be from knowledge or data cloud or now data three sixty, as I call it, is a great start. But even here, readiness is key. If your knowledge base and your data cloud isn't properly structured, then the agent's information retrieval becomes misinformation retrieval.
In terms of our partnership with Gearset, we ensure that the strategy is right and Gearset ensures that the deployment of these data streams is reliable. Next slide, please.
So at level two, this is really where agents start doing. So in this example here, Ellie is arranging a collection by triggering an API into MuleSoft to talk to an order management system.
This is where readiness becomes technical challenge. So now we're managing integrations and logic across different domains. There's no longer a simple prompt. It's a functional piece of software.
And that's kind of the difference here. So you can write one prompt to do something quite basic, as we saw in level one, but level two is really multiple prompts and it starts to become a piece of software. Critically, that requires DevOps discipline to make sure it functions. Next slide, please.
So level three, this is like really where the magic happens, but it's also where the risks increase we mentioned just earlier. So now the agent is leveraging real time data graphs, in this example, to personalise offers, and it's calling inventory agents to call to check stock.
To be ready for this kind of level three, your Salesforce environment really does need to be a well oiled machine.
And you need a way to track the behavior of these interacting agents before they ever touch a real customer.
And obviously, it goes without saying, in order to have that well oiled machine, you know, you need your DevOps processes working really well. Next slide, please.
So why DevOps matters in AI?
So as we're all being told time and time again, for good reason, AI is only as good as the processes and the controls around it, as well as having good quality data for it to learn from. You can have the latest tech, and in our case, that's Agentforce with Service Cloud Voice. But if you can't reliably build, test, and deploy systems around it, it can soon become a very expensive hole in your pocket.
For us, this is where DevOps and Gearset step in. Devs DevOps isn't just about speed anymore. It's about trust.
Trust that when you release your latest AI driven customer service agent, it won't accidentally start writing poems about your competitor's product.
This does happen. AI can be very democratic if you're not careful.
With AI baked into Salesforce from predictive recommendations to conversational agents, the stakes are now quite a lot higher.
These aren't just features. They're frontline brand representatives. And if your AI agent behaves badly, it doesn't just break a workflow, but it can break trust and by proxy your customer loyalty.
Next slide, please.
So building an agent is not a kind of one and done. So I know most of you are probably working in continual evolution of your platforms. But I think, you know, kind of in the past, people have thought of, you know, they develop a feature or a function on Salesforce and it's kind of job done. You know, we never come back to it. But building agents isn't as a cycle as we can see here. And Gearset allows us to treat these configurations with the same rigor that we treat code.
And that is crucial to ensuring that every deploy is a known quantity.
Next slide, please.
Readiness challenges that customers are tackling at the moment.
So these are also challenges which we're tackling alongside our customers.
So the first one here is architectural alignment. So this is moving from technical specs to a defined brand persona and also the ethical guard framework.
The solution here is strategy before build, as we said earlier.
So you need to establish the tone, the boundaries, personality requirements, all of those things before a single line of logic is written.
The second thing is version continuity.
So this is around preventing Frankenstein deployments caused by an automated updates during developments.
The solution here is lifecycle discipline. So this is establishing rigid version controls to manage the rapid and constant evolution of agent iterations. That's something that we see a lot.
And I'm sure those of you who are building agents at the moment will see that, you know, it really is a kind of a very quick, rapid, iterational process.
The final thing is integrated governance. So this is kind of managing the blurry line where prompt engineering and DevOps meet.
So the solution here is really around unified DevOps.
So you need to ensure that your Salesforce ecosystem is prepared to manage the prompts, the data and the guardrails as code.
Next slide, please.
And just a few things here.
These are things which we think people may not necessarily be thinking about just now.
So the first thing is data drift and governance.
Data changes, customer behavior shifts, and product names evolve. Without governance, your AI can start drifting like an unmoored boat, but instead of peacefully across the Mediterranean, it could be heading into disaster on the rocks. So DevOps needs to account for monitoring, retraining and versioning of data itself, not just metadata.
The second thing is ethics and compliance as part of the pipeline. So it's not enough to check that your pipeline deploys now. You need to ask more. So things like, does this comply with GDPR? Does it avoid bias? Does it meet internal ethical guidelines?
So we're moving towards pipelines that don't just fail on a unit test, they could actually fail on an ethical test. And we think that's a good thing.
The third thing is cross functional collaboration. So AI agents don't just live in a vacuum. They span customer service, sales, marketing, legal, IT, which means your DevOps process has to support collaboration across teams who may not usually work together.
Think of it like organizing a Christmas family dinner. If you don't plan it properly, disaster looms. And then finally, observability. So once deployed, AI doesn't just need logging, it needs observability.
You need to watch for hallucinations, incorrect responses, customer sentiment, and really this is new territory. So your pipeline needs to bring telemetry from the behaviour of the agent, not just the agent's uptime.
Next slide, please.
So how are we addressing these challenges?
And really, you know, kind of what can developers and DevOps leads in the trenches do to actually tackle these?
So the first thing, treat prompts as code.
I'm sure it's quite a simple, know, sort of first step, but really, you know, we need to keep things version controlled, peer reviewed and tied to the user stories that are in your backlog.
This way you can not only follow traditional version control processes, but you can also shift your testing left by identifying the small prompt changes that can cause ripple effects.
Second thing is automate scenario testing. So preemptively create simple libraries of scripted conversations to catch glaring issues early. Testing center is great for this initially, particularly for understanding the reasoning to trigger actions, But your flows and APEX, as we've said earlier, they still need the same testing approach.
Third thing is monitoring behavior, not just uptime. So we need to build lightweight logs of agent responses and review them for drift.
Consistent monitoring and continuous improvement are key to making an agent accurate, something that builds over time.
These logs can also help inform your testing scenarios and your version control prompts.
The fourth thing is baking in governance manually from day one. So adding checklists for compliance and bias reviews to your release process until automated gates arrive. Salesforce manages a lot of this for you with a trust layer, but building in a pull request government governance for AI using Gearset can massively help ensure your governance and the build behind your agent hits all the areas you need.
And then finally, collaborate widely. Bring customer support, legal, marketing, all the different departments that are going to be involved, bring them early into test cycles so that trust doesn't become a problem.
We've seen projects ourselves stalled where other departments have not signed off.
For example, one client, we had quite a lot of delays through IT security.
We've had ones where it's been legal, and it's really because they haven't brought those teams in early on. So ensuring everyone who could need to be involved brought into the vision, literally from day one, really helps keep the project on track for deadlines, and it's going to ensure that you get the best results from your agent.
I feel like these small pragmatic steps that we've just talked around, they should help ensure that you're building good habits now. So when you're ready to start your agentic journey, you can integrate these efforts directly into your existing DevOps discipline.
Next thing is an iteration model. So we feel that readiness is about starting small and then scaling sustainably.
So we look at establishing the pain points and we find the happy path. So by happy path, we mean where everything kind of works and flows and there's not really any exceptions. And then we use every single exception or escalation point as the milestone to expand the agent's reach or capabilities over time. And this allows for parallel work on new use cases without breaking existing ones.
Next slide, please.
So we mentioned this earlier.
We are really proud of Performa to have been behind the first Agentforce solution, which went live in EMEA and is actually was actually, you know, consuming credit. So that was over a year ago now, which is amazing how time flies. So if you're familiar with this success story, you'll know that the team placed a huge emphasis on readiness, starting small with the right mindset and the right strategy, building a great foundation to scale their use cases and continue to enhance and iterate. We've mentioned the fantastic results that they've had, so I'm not going to go over those again. Next slide, please.
So here at Performa, we've worked on a number of different agents now. So everything from the SDR to internal agents on field service lightning, Secret Escapes, the service agent, and we're also just finishing off a really exciting project with Reconomy, which is a broker agent, which is going to save them a huge amount of time. So, some really interesting stuff going on.
Next slide, please.
So, I just wanted to talk about really quickly, you know, where we help you start with this agentic journey?
We would really recommend taking advantage of one of our workshops. They are complementary, there's no charge for it.
And this is where we use a trusted framework to take plans from ideas to tangible defined solutions that can be managed, tracked, implemented and measured.
These workshops are made up of three sort of key pillars. So first thing is looking at the foundation and the tech stack. So it's understanding your existing Salesforce ecosystem.
Second thing is data integrity and AI readiness. So this is ensuring your brain or data in your system is ready for the AI agent. As part of that, we do a five point data health check. We look at systems integrations, and we also look at sort of friction audit as well.
The final thing is strategic goals and ideation. So we mentioned ideation before. I know that was part of your homework, so we can help people with this.
So really the goal here is defining the high impact use cases for Agentforce. So, things like looking at, you know, what are the top five business priorities? What are some of the really painful friction points?
And what does success look like? You know, defining success.
These workshops, they're a sixty minute session, and then after that, we will basically prepare and deliver to you your own bespoke Agentforce implementation roadmap with recommendations on what we think your next steps could be.
So if you want to scan the QR code on the slide, please go ahead, and that will take you to a page where you can basically have a look at some of our demos, webinars that we've done in the past, and you can also book one of these sessions.
So I just really wanted to thank you very much for listening to me today.
We are, as Performer, a boutique SI Salesforce partner with Summit Status. We've been around for over fourteen years now. And crucially, sort of part of this presentation today, we have been a Gearset Custom Ramp partner for a number of years now. And, you know, we couldn't do what we do without Gearset.
So, yeah, if you want to talk, see if we can help you on your agentic journey, please reach out.
Dave, thank you so much for sharing your wisdom and your insight.
Something that's really encouraging to me is we've run through this webinar series and we have learned more about AgentForce and what it takes to be ready for AgentForce. It isn't just that it's the art of the possible, it's that there's customers out there like Secret Escapes and like the customers that you work with Ben that are seeing success today from the capabilities of Agentforce and the art of the possibility is already here for a lot of folks and I think we should all be encouraged by that. Yeah.
Thank you so much.
Folks, I hope you have enjoyed this webinar. We have a little bit over but I'm going leave a little bit of time for Q and A here as well. Ben, if you're happy to stick around and Dave, you're happy to stick around as well. Yeah, of course. Just for a little bit of insight as if anybody has questions for them you can pop them in the chat. But I'd be really interested in both of your perspectives on if there was one thing that you think would be helpful for people to know probably at the start of their Agentforce or early on in the start of their Agentforce journey that you wish that you knew at the beginning, what would that be? Ben, I'll throw that to you first.
Okay. I would say you need to bear in mind that the product itself is continually evolving and the release schedule from Salesforce of new features and enhancements, but also bug fixes is really fast paced.
So I think you just have to bear in mind that there will be, you know, a number of different iterations and there may be some things which change over time and you kind of need to tweak it. So it goes back to that kind of it's not just a one and done. You don't just kinda go, there's the agent. Leave it to it.
Never come back to it. Because the kind of platform capabilities are changing over time, and, you know, the way that that that certain things are handled within the platform can change. So we've seen that quite a bit where, you know, we'll the agent behaves in one way initially. Salesforce makes a change, and it's and it's, you know, kinda changes it.
It's not really how exactly how our client wants it. So we just need to do some tweaks to just kind of, you know, get it, you know, to continue to work in the way that that is expected.
Yeah. For sure. Continuous iteration, which is why DevOps marries so nicely with this kind of topic. And Dave, what what about yourself?
I think there can be a real temptation when you start your Agentforce journey to try and go from naught to a hundred as quickly as possible. And with AI, there's so much learning to be done across the ecosystem. I think it's really important to learn to walk before you can run.
So start start simpler than than you think you'll need to with your first use case.
So I think the temptation to over engineer your first agent is is absolutely real.
And actually, I think we found that the teams that really succeed with making progress in their Agentforce journeys, they're picking a really focused level one use case to begin with. They're getting that into production, and then they're learning from real user interactions how to iterate and improve on that. And one of the things that we've seen with a lot of those teams that are deploying their level one agents and then moving fairly quickly to level two, and we'll see a lot of level three agents probably starting to take shape this year, is that getting that DevOps pipeline in place while you're building your agents is really important. Just have have a really reliable route to production that that you guys can be confident with.
Yeah. For sure. A hundred percent. And that's where those That homework that we set everybody, that's why the homework is useful so you can start to think about that walking before you're running and understanding the art of the possible with a well defined problem space and something that you're familiar with and gradually build confidence over. Ben, we do have a question in the chat here as well directed to you from Ankit. I'm new to Agentforce. From where can I get the topics to start with?
Well, as ever, I would always recommend having a look at Trailhead Salesforce, you know, because this is their trajectory now in terms of, you know, their sort of key technology offering.
In my experience, it's been very good. I've done the AI consultant certification, and that's fantastic. It literally will take you through from start to finish of how to create a basic agent.
So I'd recommend starting with those trailheads personally.
Perfect. Thank you, Ben. And one last question from my side, selfishly, as I like to focus a lot on team cultures and slightly humans in the loop, which we've discussed a little bit here.
What are the some of the traits that you think, Ben and Dave, that you think are gonna make Salesforce teams excel when they're taking on an Agentforce project? What is the kind of key elements to the culture of those Salesforce teams that is essential to to these projects?
I think so if we look at, for example, secret escapes, all of those things that I spoke about just earlier in terms of getting people on board and involved in the project right from the start across different departments, I think that's really crucial. I think one of the things that, you know, we noted was that some of the agents in the call centers at Secret Escapes were initially kind of really worried like, is this technology that's going to take my job away and what I'm going to do? So bringing them in right from the start. So we had a lot of different agents from the call centre who were involved in the project, they knew exactly what the agent was going to do, what it wouldn't do, how it would work, how it would transition to them. And I think that really kind of took the fear away and then got them on board, and what they could see was actually my job's going get much more interesting and I'm going to do more valuable useful things.
The crucial thing as well with that was we then had them all on board for the testing phases as well, and we made it fun, know, we designed some competitions, we had little prizes and things, you know, can you break Agentforce?
Nobody actually got a prize, because we couldn't actually break it, we couldn't get it to do things that it wasn't meant to do. But I think just in terms of the culture of that business, and the way they got all of the departments on board and involved and really good communication throughout the project, that that that really helps.
Yeah. Super resonant for sure, Dave.
I think one of the things that we've seen in in kind of echoing Ben's comments is that to execute on your AI projects, you've really gotta have a strategy that you can execute on, and executing on that strategy takes a village. You've gotta bring people with you to do that.
As part of doing that, you've got to really focus in on business problems that have meaningful ROI. So it's not just an engineering team that are picking up tickets from Jira. You're talking about business transformation.
So you've really got to start small and prove value quickly, and then try and connect it to governance as well. So show your leadership how the changes that you're making will be tested, deployed, and tracked, and that will really start to get buy in for for your projects and your programs. And that will that will tip maybe cautious stakeholders over from that's interesting to, hey. Let's do it. Let's let's let's actually, get this into production.
Great. Great advice, Dave. Thank you so much. And we have quite a nice closing question coincidentally coming in the chat here, also again directed at you Ben.
Ankit, there is no such thing as a silly question. You said it might be a silly question, there are no such things. What does the future of AgentForce look like and is it secure and worthwhile platform to move into?
I well, obviously, I'm gonna be biased because we are a Salesforce partner. But I I do genuinely think that, you know, if you compare, Salesforce to a lot of the other platforms that are out there, I do think it is kind of really one of the front runners at the moment. And I think it's that trust framework that the technology embedded within that trust layer that's a real key differentiator. We've personally seen some customers of ours where they've done their own in house AI builds. It's been really costly, not just from a sort of financial perspective, but they have had instances where because, you know, they're sort of taking off the shelf LLMs and coding their own solutions, they didn't really have a trust layer. And then, you know, the agents were actually doing things and saying things to customers that they shouldn't have done.
So, yeah, I think I think that that's a key differentiator. Sorry. Can you just repeat the question again?
Yeah, what does the future of AgentForce look like? Is it a secure and worthwhile platform to move into? Which you've somewhat addressed. The future, I guess, Ben. What's the future look like?
I think Get your crystal ball out.
Yeah, well, I think personally, I think we're gonna see a lot more of Salesforce configuration and administration is going to be, you know, completely managed, you know, using an agent, you know, embedded into the actual platform. So I think as administrators and consultants and developers, we're gonna be, you know, kind of talking to Salesforce. Obviously, Voice, you know, is is sort of fairly new, but I think we're going to see, yeah, the sort of landscape of how we actually develop Salesforce, I think that's going to change. I think we're going to leverage some of that technology before. Obviously, you can see Salesforce are embedding it across all of their different clouds.
So I think over time, we're gonna see a lot more kind of, you know, sort of really popular use cases. I think Salesforce will start to plug some of those into the platform, you know, and sort of build on what's already there.
Great. Exciting times indeed ahead for for all of us that are developing on the Salesforce platform. Ben, Dave, absolute pleasure to have you both on this webinar today. Really appreciate, really value your time and your insights into everything that we have discussed from the maturity model to the release readiness factors and those folks out there that are seeing success with the AgentForce platform. It's been a pleasure to learn from you. Folks, if you have enjoyed webinar, you will receive a recording so you can go back and re digest everything that we've learned here today. Look out for that in your inbox over the next day or two.
For any questions for Ben, LinkedIn the best place for you Ben, alternatively?
Yep, absolutely shoot me a message on LinkedIn.
Perfect. Dave, what about yourself?
Yeah. Same LinkedIn. Or if if we connect with you on Slack Connect, Slack's a great place too.
Oh, yeah. Perfect. We're on Slack as well.
Perfect. Thank you very much both for your time. Thank you for attending and tuning in to this webinar. I hope you enjoyed it and we will catch you again on another webinar sometime soon.
