Uniform blog/Scout, Uniform Code, and the dual-audience

Scout, Uniform Code, and the dual-audience

Q: What is Uniform Code and how does it differ from standalone code generation tools?

Uniform Code is a new code generation environment that currently generates components from various inputs, including text prompts, Figma files, hand-drawn sketches, or URLs. The critical difference from standalone code generation tools is architectural integration: everything generated is natively integrated with the Uniform platform from the start. After generation, it takes the output directly into the Visual Workspace, where every component follows Uniform best practices and inherits the orchestration layer, content model, and personalization infrastructure already configured. This allows marketing teams to edit visually while engineering teams extend it in existing codebases, both working from the same source of truth. Additionally, Uniform Code can connect to legacy codebases built before AI tooling existed, adding new components without requiring a rebuild, enabling incremental adoption rather than replacement. Uniform Code accelerates the creation of experiences serving both human visitors and AI agents simultaneously.

UniformPosted on Jun 8, 2026

8 min read

Scout, Uniform Code, and the dual-audience

TL;DR

Scout and Uniform Code help teams create and manage content for both human audiences and AI agents from the same content model. Answer engines, crawlers, and agentic interfaces now influence how digital experiences are discovered, structured, and delivered. Key takeaway: Uniform enables a practical, phased approach: use Scout to scale governed content operations and Uniform Code to generate platform-aware experiences without forcing an all-at-once rebuild.

In a recent webinar, Uniform CEO and Co-founder Lars Petersen demonstrated significant updates to Scout since its 2024 launch. He then previewed Uniform Code, a new code generation environment built directly into the platform. The session centered on a single operational question: how do content teams build for both human visitors and AI agents without doubling the workload?

What changed in Scout

Scout's latest release represents the largest update since its 2024 launch. The new capabilities change how teams interact with the agent at scale, and Petersen walked through each one live using a fictional travel company called EcoQuest.

Custom skills let teams encode brand voice, content standards, and design system rules directly into Scout. In the demo, Petersen showed a three-tier guidance system: brand-level rules (tone, personality, do's and don'ts), component-level purpose (what a Flex Hero is for), and field-level instructions (what makes a good eyebrow line). He also demonstrated persona-based writing skills by asking Scout to create a blog post "written like David" and by producing detailed travel content that mimicked a specific storytelling style. Agent-generated output follows the same guardrails a human editor would enforce, producing consistent results whether the task is a single-field edit or a bulk operation across hundreds of entries.

Semantic search replaces keyword-based content lookups with meaning-based queries across the entire project. A prompt like "find every entry about Nordic destinations and add structured FAQ fields" executes in a single operation, regardless of how the content was originally tagged or categorized. In the demo, this extended to asset management: Scout automatically classified uploaded images with titles, descriptions, and labels, then matched the best assets to new compositions based on context.

MCP server connectivity works in both directions. Externally, Uniform and Scout are available as an MCP (Model Context Protocol) server, meaning any tool that supports the protocol can interact with Uniform content without opening the platform UI. Petersen demonstrated this from Slack, issuing commands to update blog entries and manage locales without leaving the messaging interface. Internally, Scout connects to external MCP servers, bringing data from CRMs, commerce platforms, and project management tools directly into the content workflow. In the demo, Scout queried a HubSpot account, retrieved company data, and generated a targeted landing page complete with messaging tailored to that specific account, all from a single prompt.

Petersen also emphasized Scout's two operating modes: autonomous for teams ready to let the agent execute at speed, and review for teams that need every change approved before it goes live. The choice can be toggled by query. Organizations can begin in review mode and expand autonomy as trust builds.

The AEO and GEO workflow

One of the most practical demonstrations was the answer engine optimization workflow. Petersen outlined a four-step process:

Determine which agents to optimize for (Perplexity, Gemini, Claude, each with different schema expectations).
Ask Scout to add the necessary fields across the content model at scale.
Have Scout populate those fields with optimized content.
Human review before anything ships.

The same pattern applies to translation, localization, accessibility audits, and personalization setup. Petersen listed eight use cases for the updated Scout: creating and refining content; generating testing and personalization ideas with execution; brainstorming content models; bulk operations; enriching content for AEO and GEO; running audits (accessibility, CX, GEO); translation and localization at scale; and external MCP server connectivity.

Uniform Code: from prompt to platform-aware component

The second major announcement was Uniform Code, currently in preview. The environment works similarly to standalone code generation tools, but with one architectural difference: everything generated is natively integrated with the Uniform platform from the start.

Petersen demonstrated several input methods: a text prompt describing the desired layout, a Figma file, a hand-drawn napkin sketch, or a URL. In one example, he generated an entire marketing homepage for a fictional company from a single prompt. The output included navigation, hero sections, feature cards, and FAQ components, all using reusable component definitions rather than monolithic code.

The distinction from standalone code generation is what happens next. Clicking "Open in Canvas" takes the generated output directly into the visual workspace, ready for editing. Every component follows Uniform best practices and inherits the orchestration layer, content model, and personalization infrastructure already configured. Marketing teams edit the output visually. Engineering teams extend it in an existing codebase. Both work from the same source of truth.

Petersen also showed Uniform Code connecting to an existing codebase, one built years before any AI tooling existed. The environment can add new components to legacy projects without requiring a rebuild, which means teams can adopt it incrementally rather than as a replacement for existing infrastructure.

The throughline: one content model, two audiences

Petersen kept returning to a specific framing throughout the session. For 25 years, digital experience teams built for one audience: humans. The tools evolved, but the target never changed. Now content teams have a second audience: AI agents, crawlers, answer engines, and agentic interfaces that discover, evaluate, and transact on behalf of human users.

The division of labor Petersen articulated maps directly to the tools demonstrated: Scout handles the agent-led work defined by scale, repetition, and speed, while Uniform Code accelerates the creation of experiences that serve both audiences. Composable orchestration connects the two, rendering the same source material as a visual experience for human visitors and as structured schema for AI crawlers from a single content layer.

Agentic at your pace

Petersen's closing message was deliberate: none of this requires organizations to move all at once.

His suggested sequence was specific. Short-term, use code generation for a greenfield project. Next, connect an existing stack with Uniform for orchestration. Then evolve gradually, expanding agent autonomy as governance and guardrails mature.

The framing was "agentic at your pace," not agentic all at once. Organizations that are not ready to run Scout in autonomous mode can start in review mode, where every change requires approval. The architecture supports both postures, and teams can shift between them as confidence builds.

Uniform Code is available in preview. Scout's updated capabilities are live now.

The dual-audience shift raises deeper architectural questions about how content models, design systems, and delivery infrastructure need to evolve. That is the subject of a companion post: The audience changed. The architecture has to follow.

FAQs

The solution lies in a two-pronged approach using Scout and Uniform Code. Scout handles agent-led work defined by scale, repetition, and speed, while Uniform Code accelerates the creation of experiences that serve both audiences. The key is composable orchestration, which renders the same source material as a visual experience for human visitors and as structured schema for AI crawlers from a single content layer. This eliminates the need to create separate content versions, allowing teams to maintain one content model that serves multiple audiences simultaneously. Additionally, organizations can start in review mode where every change requires approval, then expand agent autonomy gradually as trust builds, making the transition manageable rather than disruptive.