LandingAgent: A Reference-Annotated Dataset and Agentic Generation Framework for Landing Pages

Injun Baek; HyeongSeok Lee; Yearim Kim; Junhoo Lee; Nojun Kwak

cs.CLAug 28, 2026

LandingAgent: A Reference-Annotated Dataset and Agentic Generation Framework for Landing Pages

Authors: Injun Baek, HyeongSeok Lee, Yearim Kim, Junhoo Lee, Nojun Kwak

Organizations: Seoul National University · Samsung Electronics

Abstract

Landing pages are goal-oriented web interfaces that must communicate a target-specific value proposition while organizing information flow, visual hierarchy, and calls to action (CTA). Although large language models can generate plausible webpage code from natural-language prompts, direct generation often yields generic templates and unsupported persuasive claims. We study target-grounded, reference-guided landing-page generation, where a system must create an executable page for a new target by adapting reusable patterns from real pages without copying them. We introduce LandingBench, a reference-profile dataset that abstracts real landing pages into section sequences, layout patterns, tone descriptors, visual emphasis, and CTA structure. Building on LandingBench, we propose LandingAgent, a three-phase agentic framework that profiles the target, constructs a reference-guided wireframe, and refines the page through critique-guided polishing. We evaluate LandingAgent against direct prompting on faithfulness, conciseness, readability, aesthetics, and structural diversity. Experiments show improved target grounding, presentation quality, and layout diversity. Code is available at https://github.com/IAURAI/LandingAgent.

Figures & tables

Figure 2: Overview of the LandingBench collection and filtering pipeline. Candidate pages sampled from Common Crawl are processed through rule-based filtering, LLM-based scoring and human verification.

Level	Reference-profile fields
Page-level axes	category, target, conversion goal, tone, information density, scroll depth, CTA, asset dependency
Section structure	section order, section role, layout pattern, visual emphasis
Visual descriptors	palette descriptor, typography descriptor, imagery & iconography descriptor
Retrieval fields	page-structure axes tuple, reference screenshot, abstract metadata

Table 1: Core reference-profile schema of LandingBench.

Figure 3: Overview of the LandingAgent framework. Given a user-provided target specification and optional image assets, LandingAgent first performs a Profiling Phase that transforms the inputs into a Structured Page Brief. In the Wireframing Phase, a Retriever selects a structurally relevant reference from LandingBench, and a WireframeBuilder constructs a skeletal page draft. A WireframeCritic either approves the draft or triggers a Restart Loop that reselects a reference and regenerates the wireframe. In the Polishing Phase, a Polisher iteratively refines the draft with feedback from three parallel critics: AestheticCritic, MessageCritic, and FunctionalCritic.

Phase	Agent	Role
Profiling	Profiler	Transforms the target specification and image assets into a Structured Page Brief.
Wireframing	Retriever	Selects a reference profile from LandingBench that matches the Structured Page Brief along structural and visual axes. When a restart occurs, it uses the diagnostic hint to avoid repeating the same structural failure.
	WireframeBuilder	Constructs a skeletal page draft guided by the selected reference profile at the level of section roles, information flow, and layout organization, without using raw HTML or verbatim source-page copy.
	WireframeCritic	Evaluates whether the draft is globally aligned with the Structured Page Brief and the retrieved reference profile. It either approves the draft for polishing or triggers a restart with a diagnostic hint.
Polishing	Polisher	Refines the approved wireframe into an executable landing page by adding target-grounded copy, visual styling, asset placement, and implementation details while preserving the global structure.
	AestheticCritic	Assesses visual presentation quality, including typography hierarchy, color harmony, whitespace, alignment, visual readability, modernness, brand expression and reference fidelity.

Table 2: Functional role of each agent, grouped by phase.

Method	Generator	Critic	Diversity	Content		Presentation
Method	Generator	Critic		Faithfulness	Conciseness	Readability	Aesthetics
Direct Prompting	gemini-2.5-pro	—	0.26	3.83	4.40	3.90	3.67
Direct Prompting	gpt-5	—	0.28	4.60	4.13	3.77	3.63
One-shot Prompting	gemini-2.5-pro	—	0.29	4.05	4.67	3.95	3.67
Direct Prompting + Polisher	gemini-2.5-pro	claude-sonnet-4-6	0.25	4.33	4.71	3.95	3.62
Landing Agent	gemini-2.5-pro	claude-sonnet-4-6	0.36	4.30	4.83	4.03	4.00

Table 3: Main evaluation results with structural diversity, content quality ( Faithfulness , Conciseness ), and presentation quality ( Readability , Aesthetics ). Scores are rounded to two decimal places.

Figure 4: Qualitative comparison of structural diversity. For the same target specifications, Direct Prompting repeats hero layouts, visual assets, and section order, whereas LandingAgent produces diverse yet target-faithful layouts, visual grounding, and section flows.

Figure 5: Qualitative comparison of Direct Prompting and LandingAgent on two target specifications: Faithfulness (top) and Aesthetics (bottom). Red boxes mark failure regions; green boxes mark successful regions.

Figure 6: Pairwise user study results: Ours (LandingAgent) vs. Baseline preferences across the five criteria.

Method	Div.	Faith.	Conc.	Read.	Aesth.
Hard filter only	0.29	4.00	4.43	4.00	3.86
w/o LandingBench	0.25	4.24	4.48	4.00	3.86
w/o RAG	0.29	4.10	4.38	4.00	3.86
w/o Profiling	0.32	4.24	4.52	4.00	3.95
w/o Polishing	0.24	1.14	1.76	3.38	2.24
Ours	0.38	4.33	4.71	4.05	4.00

Table 4: Ablation results for reference filtering, LandingBench availability, reference retrieval, and pipeline phases. Scores are rounded to two decimal places.

Appendix figures & tables18 assets

Supplementary material from the paper’s appendix.

Appendix

Stage	Output	Filtered
Initial crawl	1,000	—
After Hard Filter	938	62
After Soft Filter	636	302
After Human Filter	438	198

Appendix

Table 5: Candidate counts at each stage of the LandingBench filtering pipeline. The final 438 pages are those that passed all three filtering stages and have valid extracted input images and annotation files.

Category	Criteria
Adult or unsafe content	Slug-level keywords ( porn , sex , xxx , nude , adult , etc.); multilingual visible-text keywords (English, Arabic, Hindi); gambling and spam signals
Inaccessibility	HTTP 4xx/5xx fetch status; screenshot file size below 30 KB; visible-text keywords indicating error pages, login walls, bot verification, geographic restrictions, or JavaScript requirements
Insufficient content	Visible-text word count below 80; absence of all structural signals (H1, CTA, hero) simultaneously
Page-type mismatch	URL path patterns matching /blog/ , /news/ , /docs/ , /jobs/ , /wiki/ , /help/ , /p/\d

Appendix

Table 6: Rule-based exclusion criteria applied in Hard Filter Stage 1.

Dimension	Description	Score
Information Completeness	Product/service information and the value proposition are sufficiently presented	1–5
Brand-Style Consistency	Tone and visual style are coherent across the page	1–5
Layout Readability	Section boundaries and information flow are clear	1–5
Visual Presentation Quality	Overall visual completeness and polish are sufficient	1–5

Appendix

Table 7: Soft filtering rubric for estimating reference suitability.

Verification criteria
Verification item	Description
Inputs to reviewers	636 original webpages; full-page screenshots; complete visible text
Annotators	2 independent reviewers
Rejection rule	Rejected if either annotator assigns Reject
Landing-page validity	Is the page a genuine landing page? Articles, portals, blogs, and documentation are rejected.
Annotability	Can the page be reliably segmented into sections with clearly identifiable boundaries?

Appendix

Table 8: Human verification protocol for final LandingBench candidate selection.

Exclusion type	Count	%
News or editorial	77	38.9
Ad-tech or infrastructure	27	13.6
E-commerce	16	8.1
Other exclusion categories	78	39.4
Total	198	100.0

Appendix

Table 9: Exclusion-type distribution of the 198 candidates rejected during human verification.

Page structure
Component	Description
Rendered reference screenshot	Full-page screenshot used as visual context during generation
Structured section sequence	Ordered list of sections, each annotated with a role label ( kind ), purpose, headline, key copy excerpts, visual description, and component list
Section-role labels	Per-section label from a 13-class soft enum: features, other, footer, hero, cta band, testimonials, product tour, metrics, logos strip, use cases, pricing, faq, integrations
CTA structure	Primary and secondary CTA copy with placement-pattern description
Page-structure axes tuple	Structured retrieval axes: message length, information density, scroll depth, CTA structure, asset dependency

Appendix

Table 10: Reference profile components extracted for each LandingBench page, organized by category.

Market category	Count	%
DevTools / Infra	82	18.7
Media / Content	80	18.3
Security	61	13.9
B2B SaaS	49	11.2
Marketing / Analytics	29	6.6
Other	27	6.2

Appendix

Table 11: Market-category distribution of LandingBench pages.

Axis	Category	Count	%
Brand archetype	Sage	107	24.4
	Ruler	61	13.9
	Creator	59	13.5
	Caregiver	50	11.4
	Hero	42	9.6
	Magician	27	6.2

Appendix

Table 12: Distributions of brand archetype, company stage, brand polish level, and primary color hue across the 438 LandingBench pages. Brand metadata is extracted using Claude Sonnet 4.6 from each page’s screenshot and visible text. Archetype and stage categories are normalized from free-text model output. Color hue is derived from the primary palette hex value.

Agent	Phase	Inputs	Output
Profiler	Profiling	Target specification, optional target images	Structured Page Brief (JSON): value prop, audience, tone, must-have sections, 11 page-structure axes, 5 visual axes
Retriever	Wireframing	Structured Page Brief, LandingBench profiles, optional restart hint	Selected reference profile (id, rationale)
WireframeBuilder	Wireframing	Structured Page Brief, reference profile	Page draft (HTML)
WireframeCritic	Wireframing	Rendered draft, page draft, Structured Page Brief, reference profile	Proceed / Restart decision with diagnostic hint
Polisher	Polishing	Page draft, Structured Page Brief, style guide, critic feedback	Refined page implementation
AestheticCritic	Polishing	Rendered page, reference profile, Structured Page Brief	Score (1–5), approval decision, visual feedback, optional patch

Appendix

Table 13: Agent prompt summary. Each agent’s prompt follows a common Role / Task / Input / Output schema structure.

Metric	Definition
Diversity	Measures whether outputs generated for the same landing target differ in section composition, information flow, and layout pattern rather than repeating the same template.
Faithfulness	Measures whether the generated page accurately reflects the target specification, including product or service description, value proposition, intended audience, and conversion goal.
Conciseness	Measures whether the page communicates essential information without unnecessary repetition, excessive explanation, or vague promotional copy.
Readability	Measures whether the page has clear section flow, legible text, identifiable CTAs, and an interpretable layout.
Aesthetics	Measures the overall visual completeness of the page, including color use, typography, spacing, alignment, visual balance, and perceived polish.

Appendix

Table 14: Evaluation metric definitions.

Item	Setting
Judge model	Claude Sonnet 4.6
Evaluation format	Pointwise scoring
Score scale	1–5
Input to judge	Target specification and generated landing page.
Aggregation	Mean score per metric and method.

Appendix

Table 15: LLM-as-a-Judge evaluation settings.

Dimension	Score	Rubric
Faithfulness	1	Misrepresents the brand or omits the value proposition; fabricated claims.
	2	Major gaps — several sections the brief calls for missing, audience/tone clearly off, or only a generic brand mood with no concrete explanation of the offering.
	3	Broadly on-brief but with notable omissions, vague substance (purpose/value not concretely conveyed), missing brief-supplied pricing/product detail, or one fabricated element.
	4	Faithfully covers the brief; minor omissions only.
	5	Fully faithful — value prop, audience, tone, the concrete substance of the offering, brief-supplied pricing/product detail, and the sections the brief calls for all accurately represented.
Conciseness	1	Severe walls of text, or near-empty of real content — either way the message fails to land.

Appendix

Table 16: LLM-as-a-Judge scoring rubric.

Category	Count	%
DevTools / Infra	3	10.0
Media / Content	7	23.3
Security	1	3.3
B2B SaaS	4	13.3
Marketing / Analytics	3	10.0
Other	1	3.3

Appendix

Table 17: Industry-category distribution of the 30 evaluation target specifications. Percentages are rounded to one decimal place.

Setting	Value
Temperature	0.7 where supported by the model API
Generation seed	No fixed seed specified
Prompt-suite version	1.0
Maximum wireframe restarts	$R_{\max}=3$
Maximum polishing rounds	$T_{\max}=3$

Appendix

Table 18: Generation and reproducibility settings.

Count per run	0	1	2	3
Wireframe restarts	71	9	4	6
Polishing-review retries	25	17	9	39

Appendix

Table 19: Observed restart and retry distributions across 90 full-pipeline runs.

Variant	Description
Hard filter only	Uses a retrieval pool constructed from pages that pass only hard filtering, without soft filtering or human verification.
w/o LandingBench	Uses 438 samples randomly selected from Common Crawl as the RAG reference pool, preserving the retrieval mechanism and reference-pool size while replacing LandingBench.
w/o RAG	Removes LandingBench-based retrieval from the generation process.
w/o Profiling	Omits the Profiling phase from the generation process.
w/o Polishing	Omits the final Polishing phase and uses the generated page draft as the final output.
Ours	Uses the full LandingAgent pipeline with LandingBench-based retrieval and full filtering.

Appendix

Table 20: Ablation variants.

Figure 7: Additional qualitative comparisons between Direct Prompting and LandingAgent. From top to bottom: Gallery — the baseline omits required content; Concord — the baseline uses text whose color is too close to the background, hurting readability; View — the attached image overflows into the right margin and copy stretches vertically, reducing readability; Library — the baseline omits the CTA and renders an empty black region.

Figure 8: Example of brand de-identification for reporting. Left: Original screenshot, with identifiable text blurred to obscure recognizable commercial brand identifiers. Right: De-identified version in which all identifiable elements — brand-specific names, UI marks, and other recognizable content — are replaced with fictional counterparts (e.g., Discord

\rightarrow

Concord).

Web Application Generation

Web application generation uses AI to turn natural-language, visual, or multimedia specifications into functional, interactive websites and applications.

See all 41 papers

RefCraft: Benchmarking Visual-Reference-Conditioned Agentic Generation through Web Development
Oct 19, 2026Yuchi Wang, Rongyao Fang, Weihao Yu +7AI Agent Benchmarks Web Agents
MMLab, CUHK · Alibaba Token Foundry · The Chinese Univerity of Hong Kong +2
MM-WebAgent: A Hierarchical Multimodal Web Agent for Webpage Generation
Apr 16, 2026Yan Li, Zezi Zeng, Yifan Yang +12Web Application Generation Web Agent Benchmarks
Shanghai Jiao Tong University · Xi’an Jiaotong University · Microsoft Corporation +1
Structured Distillation of Web Agent Capabilities Enables Generalization
Apr 9, 2026Xing Han Lù, Siva ReddyLLM Agent Training Web Agents
Mila – Quebec AI Institute McGill University

Oct 19, 2026J/K move · Enter open · S save

RefCraft: Benchmarking Visual-Reference-Conditioned Agentic Generation through Web Development

Yuchi Wang, Rongyao Fang, Weihao Yu, Xilin Wei+6

MMLab, CUHK · Alibaba Token Foundry · The Chinese Univerity of Hong Kong+2

AI Agent Benchmarks Web Agents

Open paper AskPDF arXiv

Beyond basic tool use and coding skills, effectively incorporating visual references into agentic generation is equally essential for applying AI agents to real-world productivity tasks. We study this crucial capability through webpage development, a practical and readily evaluable setting in which existing benchmarks remain largely idealized: they typically focus on reconstructing visual targets or generating webpages from textual instructions alone, overlooking the more common workflow of developing from visual examples. Specifically, we introduce RefCraft, a benchmark comprising 300 realistic multimodal webpage-development tasks that combine natural-language instructions with screenshots from one to five reference webpages. Agents must selectively interpret and utilize these references while rendering, visually inspecting, and iteratively refining their outputs. We evaluate the resulting open-ended artifacts using independently verifiable keypoints and complementary programmatic and VLM judges. Across nine frontier (M)LLMs, the strongest system achieves an overall score of only 52.12 out of 100, revealing that visual-reference-conditioned agentic generation remains a substantial challenge for current models in practical webpage-development settings.

Abstract

Web Application Generation

Explore similar work

RefCraft: Benchmarking Visual-Reference-Conditioned Agentic Generation through Web Development

MM-WebAgent: A Hierarchical Multimodal Web Agent for Webpage Generation

Structured Distillation of Web Agent Capabilities Enables Generalization