Over the last year, I’ve worked hands-on with Salesforce’s Agentforce across a range of client projects at Credera. As a solution architect, I can honestly say it’s the most fun I’ve had in my Salesforce career. For our clients, the most important question is: How can Agentforce actually deliver value? Not just as a shiny new artificial intelligence (AI) feature, but as a practical tool that solves real problems.
In this blog post, I want to share what I’ve learned “from the front line”: the real-world challenges, the wins, the surprises, and what it takes to go from a pilot to something that makes a difference. My goal is to give you a practical view of Agentforce—what works, what to watch for, and how to set yourself up for success.
Wait, but where do you start?
The AI hype is everywhere. Most organisations are excited by the possibilities it brings, and when it comes to agentic AI like Agentforce, the first question is often simply, “Where do we start?”
Do you pick customer service? Sales? Internal processes? How do you avoid wasting time and money on something that won’t deliver value?
When we worked with a major transport provider, their vision was simple but powerful: Let passengers ask about arrival times on WhatsApp and get instant, accurate answers. The technical bits were important, but what really mattered was mapping Agentforce to genuine customer pain points and iterating quickly based on feedback. Every 15 minutes, we updated the agent prompts, demoed it back to the client and watched their excitement grow as we fixed bugs and the solution improved in real time.
Bridging the skills gap
One of the biggest challenges was making sure the client team could truly take ownership of Agentforce. It’s not enough to launch an Agentforce minimum viable product (MVP)—you need your team to feel confident owning and evolving your Agentforce solution. Otherwise, it just sits on the shelf or you never move past the MVP stage.
On all our projects, our best results came when client teams were genuinely hands-on: editing prompts, tweaking instructions, and testing features for themselves. Salesforce’s professional services team provided a great Agentforce foundation. Our role as a partner was to help the client team “get their hands dirty” and build real capability in-house.
Here’s what’s worked for us:
Pair configuration and prompt design: We don’t just do it for you—we do it with you, so your team builds confidence from day one.
We create reusable documentation and templates while you’re figuring things out: This not only helps with future projects, it also demystifies things like pricing, governance, and scaling.
Phase the handover: We make sure your team is ready, with clear frameworks to measure success (like agent accuracy) before rolling out to a wider audience.
Scaling beyond MVP: Building for the future
The magic happens when you move beyond the initial use case and start to scale. That’s where new challenges (and opportunities) emerge: integrating APIs, supporting more complex journeys or combining service and sales agents in creative ways.
Here’s how it went for us with the transport provider. After the initial WhatsApp passenger companion agent, the customer team was eager to integrate journey planning APIs for multi-leg trips and to roll out across new channels.
Here are my three practical tips for how to move forward, with future success in mind:
Design your solution to be modular: Build agent instructions and integrations so you can add new features and channels over time.
Balance excitement with risk management: Pilot new features in phases, test thoroughly, and prove return on investment (ROI) before scaling up.
Stay connected to the Agentforce ecosystem: Salesforce is evolving Agentforce fast, so keep learning, share your own best practices, and don’t be afraid to borrow ideas from the community (starting with Credera’s latest Salesforce Release Webinar with our guest host Keir Bowden that you can now access on demand here).
The bottom line
Agentforce, and agentic AI in general, is a new way of working that can genuinely transform how your teams and customers interact. But to do it right, you need the right approach, the right partner, and a willingness to learn as you go. Just remember:
Start small, but bring your team in from the start.
Invest in enablement and documentation so you can truly own the solution.
Build for the future, but don’t lose sight of solving real business problems.
At Credera, we’re committed to making the tech work for you—not just as consultants, but as partners invested in your long-term success. We’re proud to be passionate about AI and Agentforce, and honest about what works (and what doesn’t).
If you’d like to talk about where Agentforce could take your business or you want to learn from our experience, please reach out to our team today.
Contact Us
Let's talk!
We're ready to help turn your biggest challenges into your biggest advantages.
Searching for a new career?
View job openings
