What is Agentforce Labs? Salesforce’s Experimental Hub for AI Agents

Artificial Intelligence is rapidly changing how businesses operate, and Salesforce is positioning itself at the center of this transformation. While many organizations are still exploring AI-powered chatbots and automation tools, Salesforce is already investing in a future where autonomous AI agents can perform meaningful business tasks with minimal human intervention.

One of the most interesting developments in this space is Agentforce Labs.

Although Salesforce Agentforce has gained significant attention as the company’s AI agent platform, Agentforce Labs represents a different and equally important initiative. It serves as an innovation hub where Salesforce experiments with new AI agent capabilities, explores emerging use cases, and develops the next generation of intelligent automation technologies.

For businesses, developers, and Salesforce professionals, understanding Agentforce Labs provides valuable insight into where AI-powered CRM and enterprise automation are heading.

This guide covers everything you need to know — what Agentforce Labs is, why it was built, what experiments it currently contains, how to access it, and what it signals about where Salesforce is heading in the agentic AI era.

What is Agentforce Labs?

Agentforce Labs is Salesforce’s experimental hub for early-stage agentic AI tools — where Agentforce Product and Engineering ships cutting-edge experiments, open-source tools, and AI research for developers to try and build with before they’re available anywhere else.
The simplest way to understand it: Labs is to Agentforce what a research preview is to a software product. Everything in Labs is experimental and pre-release. You’ll find features and tools that aren’t ready for the full product yet but are real, working, and ready for developer feedback.

This distinction matters. Agentforce Labs is not a documentation hub, a training environment, or a marketing preview. It is a live collection of working experiments that real developers can pick up, test against their actual use cases, and provide feedback on — feedback that directly influences which experiments graduate into the core Agentforce platform.

Read: The Ultimate Guide to AgentForce – Features, Benefits and Industry Use Cases

Agentforce Labs vs Agentforce Product: What’s the Difference?

Agentforce Labs is a separate destination from the Agentforce product. Understanding this separation is important for developers evaluating what to use where.

The core Agentforce product — including Agentforce 360, which reached General Availability on February 23, 2026, as part of the Spring ’26 release — is the production platform. Agentforce Labs is where you influence what eventually makes it into Agentforce.

Why Salesforce Built Agentforce Labs

To understand why Labs exists, it helps to understand the problem it is solving for the developer ecosystem.

At TDX 2026, Salesforce focused on a question they kept hearing from developers: why does it take so long to go from an agent idea to something running in production? The gap between a working prototype and a deployed agent is real — provisioning, authoring, testing, deployment, observability. Each step has its own friction.

Agentforce Labs attacks the very first step of that gap: access. Before Labs, a developer who wanted to experiment with Agentforce needed to provision a Salesforce org, configure the environment, navigate the Setup screens, and complete multiple steps before they could test a single agent behavior. That friction was a meaningful barrier for rapid prototyping and early-stage exploration.

Agentforce Labs serves two main purposes. First, it gives developers and technically curious admins a way to experiment with emerging AI capabilities. Each tool or experiment is designed to be usable, not just theoretical — allowing users to test what’s possible by taking these tools, trying them out, and seeing how they might apply to real use cases they are working on today. Second, it acts as a feedback loop for Salesforce. By making these tools available early, Salesforce can gather real-world input on what works, what doesn’t, and what should evolve before becoming part of the core Agentforce experience.

The result is a developer community that is not just a consumer of Salesforce’s roadmap — but an active participant in shaping it.

Also read: Is Agentforce Designed to Slowly Replace Einstein?

How to Access Agentforce Labs

Signing up is simple. When you go to labs.agentforce.com, log in with your existing Trailblazer account or create a new account. Agentforce Labs is free to access.

Access Requirements

• A Trailblazer account (free to create at trailhead.salesforce.com)
• No Salesforce license or org required for Labs access
• No sandbox provisioning required

What You Get on Access

Once inside Labs, developers find a curated collection of working experiments — each with documentation, feedback mechanisms, and clear context about what stage of development it represents. No waiting for the roadmap — pick an experiment, start building, and provide the raw feedback that shapes how Agentforce evolves.

What’s Currently in Agentforce Labs: The Experiments

Experiment 1: Agentforce Labs Quickstart — IDE-Native Agent Development

The flagship experiment at launch. The first experiment is Agentforce Labs Quickstart: instant access to Agentforce from the IDE of your choice. No org provisioning, no setup screens. Developers who want to kick the tires of Agentforce can connect and start building in Claude, Codex, or Agentforce Vibes.

This is a fundamental change to the developer experience. Instead of starting from the Salesforce UI, developers connect their preferred coding agent — Claude, Codex, or Agentforce Code — and begin building agents directly from their IDE. The Salesforce environment is accessed through the coding agent, not the Setup tab.

Agentforce Labs brings instant, sandbox-free prototyping so you can spin up and test agents in minutes from your own IDE. It is available in every Developer Edition org for free with MCP servers.

What this enables practically:

• Test Agentforce concepts in minutes rather than hours
• Build directly in the environment where developers already work
• Connect to Salesforce data and business logic without UI navigation
• Iterate rapidly without the overhead of traditional org provisioning

Experiment 2: Observer — Teaching Agents by Watching

One of the most innovative experiments in Labs is Observer: a tool that teaches an agent how to complete workflows just by observing you.

Rather than manually coding agent behaviors or scripting every step of a workflow, Observer watches how a human user completes a task in Salesforce and learns from that demonstration — generating agent instructions from observed behavior rather than explicit definition.

This represents a significant advance in how enterprise agents are trained. The traditional approach requires a developer to anticipate every step of a workflow, encode it, and test against edge cases. Observer inverts this model: the agent watches a domain expert do the work, captures the pattern, and translates it into replicable agent logic.

Experiment 3: Open-Source Agentforce ADLC Skills

Build enterprise agents faster with open-source Agentforce skills and best practices — this is the third core experiment in Labs, and one with significant implications for the broader developer ecosystem.

ADLC (Agent Development Life Cycle) skills are open-source tools for building, testing, deploying, and observing agents from sandbox to production. The repository is available at github.com/almandsky/agentforce-adlc with installation instructions and full documentation, and is also available in Agentforce Code and Agentforce Studio.

The cycle is continuous: build, test, deploy, observe, improve. The skills work in any environment that supports the skills.md standard.

What ADLC skills cover:

• Building agents from defined skill templates
• Testing agent behavior against defined criteria before deployment
• Deploying agents from sandbox to production with governance guardrails
• Observing agent decisions in production for continuous improvement

The open-source nature of ADLC skills is significant: it invites the broader developer community to extend, improve, and contribute back — accelerating the development of best practices that no single team inside Salesforce could produce alone.

Check out: How to Build AI-Powered Workflows in Salesforce?

Understanding the Relationship Between Agentforce and Agentforce Labs

Many people confuse Agentforce with Agentforce Labs.

Agentforce is Salesforce’s AI agent platform designed for enterprise use. It enables organizations to create autonomous AI agents that can support employees, customers, sales teams, service teams, and operations.

Agentforce Labs serves as the innovation engine behind future Agentforce capabilities.

A simple way to understand the relationship is:

Agentforce = Production Platform

Agentforce Labs = Innovation and Experimentation Environment

Ideas developed inside Agentforce Labs may eventually become standard Agentforce features after extensive testing and validation.

How Agentforce Labs Supports AI Innovation

Agentforce Labs focuses on several critical areas of AI innovation.

Autonomous Decision-Making

Traditional automation follows rules.

AI agents can evaluate context and make decisions.

Agentforce Labs explores how agents can:

• Analyze business situations
• Recommend actions
• Execute workflows
• Adapt to changing conditions

This moves automation beyond simple task execution into intelligent decision support.
Natural Language Interactions

Modern AI systems must communicate naturally with users.

Agentforce Labs experiments with ways AI agents can:

• Understand business requests
• Interpret complex instructions
• Generate accurate responses
• Maintain conversational context

This enables more intuitive interactions between humans and AI systems.

Multi-Step Task Execution

Many business processes involve multiple actions across different systems.

Agentforce Labs investigates how AI agents can:

• Gather information
• Analyze data
• Trigger workflows
• Update records
• Generate reports

all within a single coordinated process.

Industry-Specific Intelligence

Different industries have unique requirements.

A healthcare organization operates differently from a financial institution or an e-commerce business.

Agentforce Labs explores specialized AI agent capabilities for:

• Healthcare
• Financial Services
• Manufacturing
• Retail
• Technology
• Professional Services

This helps Salesforce create industry-relevant AI solutions.

Also check: Salesforce Integration v/s. Migration – Which Strategy Works Best for Your Business

Potential Use Cases Emerging from Agentforce Labs

The innovations being explored within Agentforce Labs could reshape how businesses operate.

AI-Powered Sales Agents

Future sales agents may:

• Qualify leads automatically
• Analyze buying intent
• Recommend next-best actions
• Generate personalized outreach
• Update CRM records

This could significantly improve sales productivity.

Intelligent Customer Service Agents

Support teams often spend significant time handling repetitive requests.

AI agents developed through Agentforce Labs could:

• Resolve common issues
• Route complex cases
• Generate knowledge articles
• Provide personalized recommendations

while maintaining high service quality.

Marketing Automation Agents

Marketing teams face growing complexity.

AI agents could help by:

• Building campaigns
• Segmenting audiences
• Optimizing content
• Analyzing performance
• Recommending improvements

This allows marketers to focus on strategy rather than execution.

Operations and Workflow Agents

Business operations involve countless repetitive processes.

AI agents may eventually automate:

• Data validation
• Approval workflows
• Resource allocation
• Process monitoring
• Compliance checks

creating significant efficiency gains.

Why Agentforce Labs Matters for Salesforce Customers

Many organizations view AI as a future investment.

In reality, AI-driven business operations are becoming a present-day reality.

Agentforce Labs matters because it provides a glimpse into the future capabilities businesses may soon have access to.

Organizations using Salesforce can prepare for:

• Greater automation
• Faster decision-making
• Improved customer experiences
• Increased productivity
• More intelligent workflows

Businesses that understand these trends early are often better positioned to gain competitive advantages.

The Importance of Trust and Governance

One reason Salesforce is investing heavily in controlled experimentation is trust.

Enterprise AI must be:

• Secure
• Transparent
• Governed
• Compliant

Businesses cannot simply deploy autonomous agents without oversight.

Agentforce Labs helps Salesforce test governance frameworks that ensure AI agents operate responsibly.

This includes:

• Permission controls
• Data security measures
• Audit trails
• Human review processes
• Compliance safeguards

These capabilities are essential for enterprise adoption.

How Agentforce Labs Aligns with the Future of AI

The future of AI is moving beyond simple chatbots.

Industry experts increasingly believe the next phase of AI will involve autonomous agents capable of completing meaningful business tasks.

This shift is sometimes called the transition from:

Generative AI → Agentic AI

Generative AI creates content.

Agentic AI takes action.

Agentforce Labs sits directly at the center of this transition.

Its mission is not simply to generate answers but to explore how AI can actively participate in business operations while remaining trustworthy and controllable.

Check: Salesforce Health Check – Why Your CRM Might Be Underperforming

The Broader TDX 2026 Context: What Labs Fits Into

Agentforce Labs did not arrive in isolation. It was announced as part of a comprehensive developer platform release at TDX 2026 that addressed the end-to-end gap between agent idea and production agent.

Agent Script opens authoring. Labs opens access. The last layer is everything between building an agent and running it in production — testing, safety reviews, deployment, and observability.

The Full Agentforce Developer Platform Announced at TDX 2026

Agent Script (Open Source)

Agent Script is an agent definition language that lets developers specify when agents should use LLM reasoning and when they should follow deterministic logic. Now fully open source, it gives developers precise control over the balance between AI reasoning and rule-based behavior — essential for enterprise-grade reliability.

Agentforce Code (formerly Agentforce Vibes 2)

An integrated AI development environment that understands enterprise metadata through Salesforce Unified Catalog, supports multiple models (Claude, ChatGPT), and generates code for any framework (React, etc.).

Starting at TDX 2026, every Developer Edition org includes access to Agentforce Vibes IDE, Agentforce Vibes with Claude Sonnet 4.5 as the default coding model, and Salesforce Hosted MCP Servers, all at no cost.

Agentforce Experience Layer (AXL)

Define interactive components once and embed them into conversational flows across surfaces (web, mobile, Slack, voice) without rebuilding for each platform.

Expanded AgentExchange

AgentExchange is a unified marketplace consolidating 10,000 Salesforce apps, 2,600+ Slack apps, and 1,000+ Agentforce agents, tools, and MCP servers into a single AI-searchable, one-click-activated destination.

MCP Integration (60+ New Tools)

Over 60 new MCP tools and 30+ preconfigured coding skills give your coding agent complete, live access to your entire Salesforce org data, workflows, and business logic directly from Claude Code, Cursor, Codex, Windsurf, and VS Code — no Setup tab, no clicking.

Testing Center

Pre-launch gap analysis, custom scoring evaluations, and post-launch session tracing. Testing Center goes GA in May 2026.

Agentforce Labs sits at the very beginning of this stack — the point where a developer first makes contact with a new idea. From Labs prototype to Agent Script definition to Agentforce Code implementation to Testing Center validation to production deployment: this is the complete developer journey Salesforce is building.

What Agentforce Labs Means for Salesforce Admins

Agentforce Labs is primarily positioned as a developer tool. But technically curious admins have a meaningful role here too.

Agentforce Labs gives developers and technically curious admins a way to experiment with emerging AI capabilities. Each tool or experiment is designed to be usable, not just theoretical.

For admins, Labs offers:

Early visibility into what’s coming

Experiments in Labs are pre-release features that will eventually graduate into the core product. An admin who has been experimenting with a Labs tool for six months before it goes GA is in a significantly better position to deploy and govern it than one encountering it for the first time in a release note.

Influence over platform direction

The feedback loop is real. Salesforce’s engineering team is reading and acting on feedback from Labs participants. Admins who surface practical pain points, edge cases, and workflow gaps from their specific org contexts are contributing data that shapes product decisions.

A safe space to explore without org risk

This is not the final version of Agentforce. This is a space where new ideas can be explored in real time. Labs provides experimental access without the risk of affecting a production org — making it an appropriate starting point for admins who want to understand agentic AI before they are asked to govern it.

What Agentforce Labs Means for the Enterprise

The existence of Agentforce Labs reflects something important about where enterprise AI is in 2026.

Salesforce Agentforce 360 reached General Availability on February 23, 2026. Enterprises across 124 countries are now deploying AI agents that resolve 85% of customer queries without any human involvement, with escalation rates as low as 5%. In Q4 FY2026, Salesforce closed 22,000+ Agentforce deals and processed 771 million Agentic Work Units — a 57% quarter-over-quarter increase.

Enterprise AI agents are no longer experimental in the core product. They are in production, at scale, delivering measurable outcomes. The question Salesforce is now addressing through Labs is what comes next — what the next generation of agent capabilities looks like before it is ready to be embedded in a platform that 124 countries depend on.

2026 is the year businesses are forced to scale AI responsibly. Salesforce is now building toward an agentic future: a world where AI doesn’t just assist people with drafts and recommendations, but actively executes work, orchestrates workflows, and interacts across systems like a digital colleague.

Agentforce Labs is the R&D layer of that future — visible, participatory, and open to the developers who will ultimately determine whether the experiments become enterprise infrastructure or go back to the drawing board.

Read: Customizing and Branding Salesforce for a Better Customer Experience

How to Get Started with Agentforce Labs

Step 1: Create or Log Into Your Trailblazer Account

Visit trailhead.salesforce.com and create a free account if you don’t already have one.

Step 2: Access Labs

Navigate to labs.agentforce.com and log in with your Trailblazer credentials. No additional licensing or org setup required.

Step 3: Explore the Current Experiments

Review the available experiments. Each one includes documentation, context about what stage of development it represents, and clear instructions for getting started.

Step 4: Try the Quickstart

Connect your coding agent of choice and start building agents directly in your IDE. Claude, Codex, and Agentforce Code are all supported. This is the fastest path from account creation to working experiment.

Step 5: Provide Feedback

The feedback you provide in Labs is the product. Salesforce’s engineering team is reading it, debating it, and acting on it. Be specific: what worked, what didn’t, what your actual use case was, and what you needed that the experiment didn’t provide.

Step 6: Follow labs.agentforce.com for New Experiments

More experiments will follow at labs.agentforce.com as Salesforce identifies and eliminates the next friction points. The Labs catalog will grow — checking back regularly is how you stay ahead of the roadmap.

Who Should Be Using Agentforce Labs Right Now

Salesforce Developers — especially those building agentic applications, custom components, or integrations that involve AI agents. The IDE-native Quickstart is directly relevant to your workflow.

ISV Partners — building on the Agentforce platform for distribution through AgentExchange. Early access to Labs experiments gives you a head start on capabilities your customers will eventually expect.

Technically Adventurous Admins — who want to understand what’s coming before it arrives in a release, and who want to have a voice in how it is shaped.

Salesforce Architects — evaluating how agentic AI will fit into enterprise architecture, governance frameworks, and integration patterns. Labs provides the earliest signal of what those patterns will look like.

Organizations Planning Agentforce Deployments — who want their technical teams experimenting with the frontier of the platform, not just the currently-shipped version.

Check out: How Salesforce Helps SaaS Companies Scale Faster

Agentforce Labs FAQs

Q: What is Agentforce Labs?

Agentforce Labs is Salesforce’s experimental hub for early-stage agentic AI tools — where Agentforce Product and Engineering ships cutting-edge experiments, open-source tools, and AI research for developers to try and build with before they’re available anywhere else.

Q: Is Agentforce Labs free?

Yes. Agentforce Labs is free to access. All you need is a Trailblazer account.

Q: Where do I access Agentforce Labs?

At labs.agentforce.com. Log in with your existing Trailblazer account or create one free at trailhead.salesforce.com.

Q: Is Agentforce Labs the same as the Agentforce product?

No. Agentforce Labs is a separate destination from the Agentforce product. Everything in Labs is experimental and pre-release — real, working tools that are not ready for the full product yet but are ready for developer feedback.

Q: Can I use Agentforce Labs experiments in production?

No. Everything in Agentforce Labs is experimental and pre-release. Labs is for experimentation, prototyping, and feedback — not for production deployment.

Q: What experiments are currently in Agentforce Labs?

At launch from TDX 2026, Labs includes the Agentforce Labs Quickstart (IDE-native agent development with no org provisioning), Observer (teaching agents by watching you work), and open-source ADLC skills (for the full agent development lifecycle from build to production). More experiments will be added at labs.agentforce.com.

Q: When was Agentforce Labs launched?

Agentforce Labs was announced and launched at Salesforce TrailblazerDX 2026, held April 15–16, 2026 in San Francisco.

Q: What coding agents work with Agentforce Labs Quickstart?

Developers can connect and start building in Claude, Codex, or Agentforce Code (formerly Agentforce Vibes).

Q: How does feedback in Agentforce Labs influence the Salesforce roadmap?

By making these tools available early, Salesforce can gather real-world input on what works, what doesn’t, and what should evolve before becoming part of the core Agentforce experience. Agentforce Labs is where you influence what eventually makes it into Agentforce.

Q: Do I need a Salesforce org to use Agentforce Labs?

Agentforce Labs brings instant, sandbox-free prototyping so you can spin up and test agents in minutes from your own IDE. It is available in every Developer Edition org for free with MCP servers. A Trailblazer account is all that is required for Labs access.

Conclusion: The Roadmap Is Now Participatory

Agentforce Labs represents a meaningful shift in how Salesforce develops its platform — and a significant opportunity for the developers, admins, and partners who engage with it.

The traditional product development model keeps experiments internal until they are polished enough to ship. Salesforce is inverting that model: shipping real, working experiments to the developer community early, collecting feedback that shapes the final product, and building a generation of developers who understand the platform’s direction before it is a finished decision.

No waiting for the roadmap — pick an experiment, start building, and provide the raw feedback that shapes how Agentforce evolves.

In a market where Salesforce processed 771 million Agentic Work Units in a single quarter — a 57% quarter-over-quarter increase — the pace of agentic AI development inside the platform is not slowing down. The developers who are experimenting in Labs today are the ones who will be building production-ready agents with tomorrow’s capabilities.

labs.agentforce.com is free. Your Trailblazer account is all you need. The experiments are live and waiting.

The roadmap has never been more participatory. The question is whether you want to be part of what it becomes.