Agentforce 360 Deep Dive: Observability, Grid, Connected Agents and Agentforce Sales in ChatGPT
Wow what a packed house today! We explore some of the newest features in the Agentforce 360 Platform and draw some conclusions as to what they bring to the AI table.
We could probably write an article a week about Salesforce Agentforce and never repeat ourselves. For the sake of concluding this series soon, I’ve opted to introduce you to a few of the newest platform features currently rolling out in pilot and beta releases.
Agentforce is maturing in operational depth as it expands in capability. The newest features focus less on “what can an agent do?” and more on “how do we scale, orchestrate, and govern agents across the enterprise?”
Let’s take a shallow dive into Agentforce….
Can we see what’s happening?
Agentforce Observability
Monitoring and optimizing agents is a challenge, but now a single mission control for unified visibility is available! Observability gives leaders insight into what agents are doing, when they act, and how outcomes are generated. Tracking usage is one thing, but Observability exists to understand quality, escalation patterns, and potential failure points before they scale. If agents are going to operate like a workforce, they need workforce-level oversight.
Why Observability in the Agentforce 360 Platform is exciting:
Because scaling AI without visibility is reckless. Observability makes AI measurable, auditable, and accountable.
In simple terms, you describe what you want built and Salesforce handles how it gets built.
Taken at Agentforce World Tour in Houston, January 2026 – Observability, now in GA.
Complete AI Observability is fully GA as of February 2026. This includes Session Tracing Data Model, Agent Analytics, Agent Optimization, and Agent Health Monitoring.
Can we test and scale intelligently?
Agentforce Grid
AI workflows often break in ways that are hard to predict. Grid introduces a structured, spreadsheet-like environment where teams can combine Salesforce data, prompts, actions, and agents in one place. Instead of testing one scenario at a time, you can run batch simulations across hundreds of rows, chain prompts together, and evaluate quality at scale. Grid is in Beta, as of this writing.
Why Grid is exciting:
Because experimentation becomes systematic. Grid turns AI testing from trial-and-error into something repeatable and engineered.
Taken at Agentforce World Tour in Houston, January 2026 – Grid, in Beta as of this writing.
Can we orchestrate complexity?
Connected Agents
As organizations deploy more agents, fragmentation becomes a real risk. Connected Agents allow enterprises to present one primary “brand agent” to the user while coordinating multiple specialized agents behind the scenes. Complex requests can be broken down and routed to the right expertise without exposing that complexity to the customer.
Why Connected Agents are exciting:
Because more agents shouldn’t mean more confusion. Orchestration ensures the experience stays unified even as intelligence becomes distributed.
Can we extend agents into where people already work?
Agentforce Sales in ChatGPT
Instead of asking sales teams to toggle between systems, Agentforce Sales in ChatGPT brings CRM context directly into conversational workflows. Teams can access relevant Salesforce data, trigger actions, and maintain data integrity without leaving their existing workspace.
Why it’s exciting:
Because AI becomes embedded in the way people already operate, not layered on top as another tool to manage.
Agentforce 360 – The Platform Goes on, But this Series Wraps Up
When we started this series, we framed the Agentic Divide (the gap between experimenting with AI and operationalizing it across the enterprise). Since then, we’ve walked through the layers Salesforce has introduced to close that gap.
While anyone can build helpful agents, constructing an AI workforce that can be tested, measured and governed is the real differentiator.
The Agentforce 360 Platform helps enterprises embed agents into everyday workflows. It may not be the flashiest demo or the coolest AI thing you’ve ever seen, but it will be designed intentionally, monitored carefully, and scaled responsibly.
Remember, an agentic enterprise elevates human work with agents that operate inside trusted systems. And that’s how you win.
Consulting Partner
