What do you like best about Salesforce Sales Cloud?
What I find most compelling about Agentforce Sales (formerly Salesforce Sales Cloud) is its evolution from a transactional database into an active, autonomous revenue engine.
Here is what stands out to me:
From Passive Record-Keeping to Autonomous Action: Traditional CRMs require constant manual data entry from reps. Agentforce shifts this paradigm by deploying autonomous agents that can independently qualify inbound leads, answer product queries, manage follow-ups, and update CRM records. It moves reps away from administrative toil and lets them focus entirely on high-value human relationships.
Declarative Control with Flow & Agent Builder: You don’t need to rewrite your entire backend to build intelligent automation. Salesforce allows teams to define clear guardrails, custom logic, and trigger points using low-code tools like Flow and Agent Builder, giving organizations fine-grained control over how AI operates within their business rules.
Contextual Grounding via Data Cloud: AI is only as good as the data feeding it. Because Agentforce operates natively on top of Data Cloud, agents act on real-time, multi-source customer data rather than isolated snapshots. This keeps outreach accurate, context-aware, and aligned with actual customer behavior.
Unified Ecosystem Architecture: The seamless connection across Sales, Service, and external integrations means that lead routing, opportunity tracking, pipeline forecasting, and post-sale handoffs happen within a single ecosystem without fragmented data silos.
Ultimately, it changes the role of the CRM from a tool where reps report work to a platform where work actually gets executed. Review collected by and hosted on G2.com.
What do you dislike about Salesforce Sales Cloud?
What I find most challenging about Agentforce Sales isn't its overall capability, but rather the gap between its high-level promise and the reality of implementation.
Here is what I find frustrating:
Predictability of Cost and Credit Consumption: Consumption-based pricing model (Flex Credits) for automated actions can make budget forecasting stressful. High-volume testing in sandboxes, unexpected automated loops, or sudden spikes in inbound lead processing can burn through credits faster than anticipated, adding cost anxiety on top of base licensing.
The "Data Debt" Penalty: Autonomous agents amplify the quality of your underlying metadata—for better or worse. If a org has duplicate records, legacy field usage, or inconsistent object relationships, the agent will act on those bad inputs. Resolving data hygiene and structural tech debt ends up being a mandatory, resource-heavy prerequisite before seeing real value.
Instruction Tuning and Edge-Case Governance: Designing prompt instructions, topic assignments, and guardrails in Agent Builder sounds simple in theory, but fine-tuning them to handle nuanced edge cases without hallucinations requires deep admin expertise. Tracing why an agent chose a specific action path during debugging can sometimes feel like a black box.
Ecosystem Friction and Admin Overhead: Moving legacy logic into modern Flow and Agent frameworks while maintaining existing Apex, validation rules, and integration endpoints builds up architectural overhead. For teams operating with limited admin resources, managing this layer of complexity alongside day-to-day maintenance is a heavy burden.
In short, while the technology is powerful, the path to running it smoothly demands far more architectural discipline, data cleanup, and cost governance than the marketing implies. Review collected by and hosted on G2.com.
