How Merlin AI outranked Procore and Autodesk in AI Search with Verseodin
Merlin AI is a construction ERP startup founded in 2024 in California. It is pre-seed, has roughly 15 people, and serves modular and prefabricated construction.
Key results
- 21.88% AI citation share on ChatGPT.
- #1 average trust position in AI answers.
- 175 trust mentions.
- 13.08% GEO citation share.
- 199 brand mentions.
- 8.29% share of all mentions in the category.
- 1.53M Google organic impressions over 15 months.
- 14.9k total Google clicks over 15 months.
- Average position: 6.3.
Client
- Merlin AI (merlinai.co)
- Industry: Construction technology, AI-powered ERP, modular construction
- Engagement: Technical AEO audit, structured data, content strategy, prompt universe tracking
About Merlin AI
Merlin AI is an AI-powered ERP built for construction. It combines estimating, procurement, scheduling, project management, finance, and CRM in one system. The product replaces eight to seventeen separate tools that contractors typically use.
The founding team has over a decade in construction supply chain and lean manufacturing, early engineering experience at Rippling, and a previous startup exit.
Challenge
Merlin AI was competing against Procore, Autodesk, Oracle, Sage, ServiceTitan, CMIC, Jonas, and Viewpoint. These incumbents had far more content, backlinks, and brand history.
Its site was not structured for AI retrieval. Blog posts buried answers in long paragraphs. Structured data was thin. Heading hierarchy was inconsistent. FAQ content was absent. There was also no prompt-level measurement for AI search visibility.
What Verseodin did
Phase 1: Technical audit of the retrieval surface
- Reviewed title tags and meta descriptions.
- Checked heading hierarchy and H1 integrity.
- Checked robots.txt and sitemap.xml accessibility.
- Reviewed canonical handling and structured data.
- Implemented JSON-LD site-wide.
- Added Article schema to every blog post.
- Added FAQPage schema to key product and category pages.
- Supported migration from Webflow to Next.js with Sanity CMS.
Phase 2: Content strategy built for AI retrieval
- Researched 743 keywords across the construction ERP category.
- Built content for extractability, not just ranking.
- Used question-based titles.
- Front-loaded direct answers in the first paragraph.
- Added supporting evidence, use cases, and differentiation below the lead.
- Included FAQ sections with five standalone Q&A pairs per page.
- Shipped FAQPage and Article JSON-LD with the content.
Phase 3: Prompt universe measurement
Verseodin created six prompt universes and tracked them daily across ChatGPT, Gemini, Grok, and Claude.
- MerlinAI: Competitive prompts against Procore, Autodesk, Buildertrend, ServiceTitan, Oracle, Sage, CMIC, Jonas, and Viewpoint.
- MerlinAI 2: Expanded unbranded category prompts.
- Merlin EOS: Operating system and workflow prompts.
- Merlin PI: Project intelligence and analytics prompts.
- Merlin Merchant: Procurement and supply chain prompts.
- Offsite and Modular: Modular construction, prefabrication, and offsite building prompts.
Phase 4: Measurement-and-respond loop
- Weekly review of Answers and Citations views.
- Tracked which prompts moved and which competitors gained or lost share.
- Identified new citations by URL.
- Used blindspot prompts to drive the content queue.
- Refreshed pages based on brand mention patterns.
AI Search results
AI Trust leaderboard on ChatGPT
| Brand |
Coverage |
Trust mentions |
| merlinai.co |
21.88% |
175 |
| procore.com |
7.13% |
57 |
| construction.autodesk.com |
3.88% |
31 |
| buildertrend.com |
0.75% |
6 |
| servicetitan.com |
0.5% |
4 |
| cmicglobal.com |
0% |
0 |
| jonasconstruction.com |
0% |
0 |
| oracle.com |
0% |
0 |
| sage.com |
0% |
0 |
| viewpoint.com |
0% |
0 |
Merlin AI held 21.88% citation share at an average trust position of #1. That was more than three times Procore and more than five times Autodesk Construction Cloud.
GEO leaderboard
| Brand |
Coverage |
Prompts citing domain |
| merlinai.co |
22% |
176 |
| procore.com |
11.5% |
92 |
| cmicglobal.com |
5.13% |
41 |
| construction.autodesk.com |
5% |
40 |
| buildertrend.com |
1.13% |
9 |
| servicetitan.com |
1% |
8 |
| oracle.com |
0.63% |
5 |
| sage.com |
0.25% |
2 |
| viewpoint.com |
0.25% |
2 |
GEO citation share was 13.08% with an average citation rank of #1. Merlin AI was cited across 176 prompts.
AEO brand mentions
- Mention share: 8.29%
- Brand mentions: 199
- Total mentions in the category: 2,494
- Average mention rank: #2
Organic search
- Time period: May 2025 to July 2026
- Total clicks: 14.9k
- Total impressions: 1.53M
- Average CTR: 1%
- Average position: 6.3
- Monthly click volume: 800 to 1,400
- Monthly impressions: 60k to 180k
Before and after
Before Verseodin
- Content was written for human readers, not retrieval engines.
- Structured data was thin.
- Heading hierarchy was inconsistent.
- FAQ content was absent.
- No measurement layer existed for AI search.
After Verseodin
- Pages front-load direct answers.
- Article and FAQPage JSON-LD are in place.
- Six prompt universes run daily.
- The dashboard records mentions, citations, exact URLs, and trust position.
- Blindspot prompts feed into the content queue.
Why it matters
The case shows that structured content can beat legacy authority in AI Search. Merlin AI is a pre-seed startup competing against much larger incumbents, yet it reached the top trust position in AI answers.
The advantage came from pages built for extraction, schema from the start, and weekly iteration based on prompt-level data.
Next tracks
- Close emerging category gaps in modular construction, digital twins, AI-driven safety monitoring, and autonomous site logistics.
- Expand into offsite surfaces such as review sites, Reddit, industry publications, and YouTube.
- Deepen the weekly measurement-and-respond loop.