Accelerating Scalable & Secure Agentforce Solutions

Enterprise leaders know they need to move quickly on AI, particularly as Salesforce continues to rapidly advance Agentforce capabilities. But amid the hype, urgency, and new announcements, one challenge continues to slow down innovation: figuring out where to start.

During Dreamforce 2025 at the HIKE2 Lodge, Garry Polmateer, CEO of Red Argyle, explored the realities of building scalable and secure Agentforce solutions in the enterprise. His message was clear: winning with AI is not about boiling the ocean. It is about identifying the right early use cases, understanding the security implications, and creating a sustainable model for iterative improvement.

This session distilled lessons from enterprise implementations, Salesforce partnerships, and security-driven AI programs so organizations can accelerate responsibly and confidently.

Key Insights: Turning AI Momentum into Measurable Value

AI success begins with starting small

Polmateer acknowledged the familiar question many companies face: “This sounds great. Where do I start?”

While AI maturity is accelerating, the first step still often stalls organizations. The answer is to start with targeted, high-value use cases rather than attempting broad transformation from the outset.

Pick use cases that are specific, measurable, and relatable

Based on real customer work, Polmateer shared a structured six-question exercise to evaluate Agentforce potential:

• What task needs improved?
• Who is the persona?
• What is the input?
• Is the workflow deterministic or probabilistic?
• What knowledge does the agent need?
• What is the expected output?

This forced-focus approach prevents “AI for AI’s sake” and identifies achievable first wins.

AI agents need ongoing care and optimization

Early implementations revealed that treating agents like traditional projects leads to failures. The real work begins after deployment.

Agents need to be monitored, refined, and optimized to improve accuracy and control costs. Think lifecycle management, not one-time build.

Agentforce fits where flexibility, security, and rapid iteration matter

Agentforce excels in low-to-mid volume, highly iterative business processes with changing needs, especially where data must stay within Salesforce’s trust boundary.

For large-scale or fixed-structure workloads, hybrid or custom models may be a better fit. It is not either/or; it is about matching architecture to context.

Security and trust remain central

For regulated industries, Agentforce’s security posture offers meaningful benefits, but understanding your regulatory environment and attack surface is critical. Core considerations include:

• What data does the agent access?
• What leaves the system?
• What does the LLM retain?
• How does tokenization protect sensitive information?

Enterprises must pair innovation with rigorous security evaluation.

Real-World Example: Replacing a Legacy Chatbot with Agentforce

Polmateer shared a case where a client needed to replace a three-year-old chatbot with expiring licenses. Instead of replicating every feature, the team mapped real value paths:

• Lead creation and qualification
• Case creation with richer context
• Direct access to key knowledge-base content

The team eliminated unnecessary complexity and built the new solution in a matter of weeks. The lesson: modern AI allows us to rethink experience and efficiency rather than replicate legacy components.

Tactical Takeaways for Starting Strong

Focus on:

• One use case at a time
• Internal or partner audiences first to reduce risk
• Measurable outcomes before broad rollout
• Lightweight pilots that expand over time

Avoid:

• Trying to rebuild entire workflows on day one
• Assuming every scenario needs AI
• Launching without a plan for maintenance and monitoring

As Polmateer emphasized: Once an agent goes live, the real work begins.

Looking Ahead: A Practical Path to Responsible Innovation

The Agentforce landscape is evolving quickly, with new capabilities emerging every release cycle. Yet speed does not replace fundamentals.

Successful organizations will:

• Thoughtfully evaluate use cases
• Pair experimentation with security rigor
• Grow AI maturity through iterative delivery
• Build internal confidence and buy-in through practical wins

Innovation is not about being first in everything. It is about moving deliberately, learning fast, and scaling what works.

This conversation at the HIKE2 Lodge reinforced a simple truth: momentum builds from taking the first step. When teams start small with clear value and strong guardrails, they accelerate faster than those who wait for perfect clarity.

The future belongs to organizations who are willing to begin, learn, and continually improve.