On the Agentforce 360 Platform, what starts as a request for a “simple form” can quickly lead to a series of architectural decisions. While the technical execution of a form may be simple, making the right decisions for its long-term success is a significant challenge. You must account for data integrity, security, and a fragmented landscape of devices and platforms. 

Without a clear strategy, teams often default to familiar tools rather than tools that best fit the requirements. The result is over-engineered custom code for routine internal tasks or standard interfaces that do not meet the needs of high-stakes customer experiences. Bridging the gap between business vision and technical reality starts with defining the human at the other end of the screen.

The new Building Forms Decision Guide provides the technical framework for navigating these choices. Applying it well requires two things: understanding your audience and knowing your surface. With these requirements in mind, this post covers four ways to align your technical choices with human audience’s needs while at the same time preparing for an AI-driven future.

1. Start by defining your audience and surface

Every form decision follows from two questions: 

1. Who is interacting with this data? 
2. Where does that interaction happen?

An internal employee processing high volumes of claims in a Lightning record page has different needs than a first-time external customer applying for a mortgage on a branded portal. Ergonomics, trust model, performance requirements, and acceptable cost of ownership all vary. If you get the audience wrong early, you make technical choices that are hard to reverse.

Once you define the audience, identify the surface. Today that could mean a browser on desktop, the Salesforce mobile app, a Slack channel, an Agentforce agent chat, or a custom web application built in React or Vue.js. Each surface has different constraints, and not every tool works on every surface.

The Building Forms Decision Guide structures its decision points around four factors: 

1. The runtime use case, including scope, navigation, location, logic, validation, interaction design, styling, and layout
2. The audience
3. The surface
4. Nonfunctional considerations, including security, testing, and deployment

Before selecting a tool, you need to lock in at least the first two.

2. Navigate the internal landscape

For internal users, the goal is to keep the total cost of ownership (TCO) low by maximizing out-of-the-box (OOB) capabilities. Start with metadata-driven tools and only add custom code when the requirements truly demand it.

Dynamic Forms in Lightning App Builder remains the right starting point for single-object record-centric updates. This approach keeps logic in metadata, supports field-level visibility rules, and gives admins control without developer involvement. For multistep processes or any guided interaction that requires state management, screen flows are the primary engine. Together, these two tools cover the majority of internal use cases and should be your first choice before reaching for anything else.

The internal surface is expanding, though, and this matters architecturally. Slack Record Layouts define how Salesforce records preview and unfurl directly into the flow of work in Slack. Screen flows for Slack handle structured data entry without pulling users out of the channel. If you’re building for Agentforce, Lightning type overrides let you create rich, functional components that agents can render directly within a chat interaction. For Industries customers, Omnistudio is the right path for high-density, declarative LWC-based forms that handle complex hierarchical data models.

Across all of these, the architectural principle is the same: build for the surface where work happens, not the surface where your admin is most comfortable.

3. Build for the external user’s journey

External surfaces require more freedom and more rigor at the same time. You need brand consistency, low-friction UX, and long-term maintainability. These are often in tension with each other.

The surface dictates your starting point. Here’s how the main external surfaces map to your options:

• Experience Cloud: The primary surface for authenticated portals and help centers. Dynamic Forms are not supported here, so use page layouts for record pages, screen flows for guided processes, and custom LWC for data-heavy or bespoke interfaces.
• Enhanced Messaging: For conversational surfaces like Service Agents, use Custom LWC in Enhanced Chat to override standard responses with LWC-built interactive elements, such as scheduling carousels or product pickers, rendered directly in the chat window.
• Mobile: Mobile Publisher lets you surface an Experience Cloud site as a branded app. For more complex native requirements, the Salesforce Mobile SDK gives you access to Salesforce data when building native iOS or Android applications with LWC components.
• Custom web applications: Lightning Out 2.0 lets you embed specific Salesforce LWCs within a React or Vue.js front end. Or treat Salesforce as a pure API layer and build a completely custom front end. That approach gives maximum UI freedom but removes the guardrails that reduce long-term maintenance overhead. Know what you are trading when you go that route.

For external users, start with the surface and select the tool that fits. This is the inverse of the internal approach, where the tool usually leads.

4. Prepare your architecture for agentic UI

The Building Forms Decision Guide is about UI designed for humans. That scope holds regardless of what triggers the interface. When an agent surfaces a component based on a conversational prompt rather than a user navigating to a page, the underlying requirement is the same: a performant, secure, user-friendly form that captures the right data. Your form doesn’t just need to work when a human opens it. It needs to work when an agent decides it’s the right thing to show.

In a Headless 360 scenario the dynamic shifts. Agents don’t need UI. They need metadata. Field definitions, validation rules, visibility rules: these are what an agent reads to understand the data it’s working with and the constraints it’s operating inside. An agent navigating a record doesn’t render a page layout. It reads the metadata layer behind it.

Salesforce is responding to both of these realities at once. The Agentforce Experience Layer, previewed at TDX 2026, separates what an agent does from how it appears. Governed business logic is defined once and rendered as rich, interactive components across Slack, mobile, ChatGPT, Teams, or any MCP-capable client. Alongside it, Salesforce Multi-Framework lets developers build on the platform using React, with Salesforce security and governance intact. The front end is becoming genuinely separable from the platform.

Taken together, this is a significant moment for Salesforce UI. The surface landscape is expanding, the rendering model is opening up, and AI is beginning to participate in both the triggering and the assembly of interfaces. The architects who navigate that well are the ones building governed, metadata-rich foundations now, because that foundation serves humans today and agents tomorrow.

Apply the decision guide to your next form

Apply the Building Forms Decision Guide from the start of your next form project. Define your audience and surface before you select a tool, start with out-of-the-box options, and treat customization as a deliberate trade-off rather than a default. 

The Building Forms Decision Guide focuses on UI for humans, and that is the right place to start. Prioritize the metadata layer, use LWC where the requirements demand it, and understand where each tool sits on the spectrum from admin-configurable to developer-owned. The decisions you make today for human-facing forms are the same ones that position your architecture for the agentic innovations already in motion. Use the guide to make those decisions well.