Agentforce 360: Review : Strengths and Weakness скачать в хорошем качестве

Agentforce 360: Review : Strengths and Weakness

AgentForce 360 Review Agentforce 360 was announced at Dreamforce 2025. Here is my review 👇 🔥 Strengths: • One Unified platform for Agentic AI lifecycle • RAG + vector DB support • Bring-your-own-model flexibility • Built-in governance, security, and observability • Adoption of open-standards like MCP etc. ⚙️ Weakness and My Recommendations: • Latency: Inference too slow. Single prompt executions are slow. If you have an agent with multiple actions making external API calls, it becomes even slower. I think Salesforce is still running on CPUs and hence the slowness. Salesforce needs to run inference on GPUs. • RAG Accuracy: Improve RAG accuracy. • Fine-Tuning: Provide capabilities to allow customers to fine-tune models on their own CRM data within. This will improve accuracy and adoption. Meta is going in this direction. • Chat UI: (See video for screenshot) Provide a conversational interface like ChatGPT as a first-class citizen. It should work out-of-box on customer CRM data. Right now, AgentForce sits on the right side in a small rectangle. • Model Choice: Add small open-source models (free tiers). • Setup Complexity: Simplify. • Terminology: Use standard AI terms (functions, tools, reasoning plan) to ease adoption. #Salesforce #AgentForce360 #Dreamforce2025 #GenAI #AgenticAI