Semji vs Scrunch AI
Scrunch AI is an agent-experience platform. It measures how AI engines answer about you, audits how agents read your site, and — with AXP — intercepts those agents at the CDN to serve them a token-light copy of the pages you already have, without touching the site or the CMS. Semji fixes the same problem at the source: brief, write, optimise and publish new content into your CMS, alongside your SEO, in one platform.
Last verified 1 August 2026
The short answer
Scrunch changes what AI agents are served. Semji changes what is on the page. Which one you need depends on whether your bottleneck is the site you have or the content you do not.
- Your site is the problem, not your content: agents cannot parse it, and a CMS change is a quarterly release. AXP fixes that at the CDN, with version history and side-by-side diffs, and no ticket.
- You need evidence of crawler behaviour — which bot hit which page, and whether it came to train, to index or to retrieve.
- Sentiment, share of voice, product-level shopping visibility, or coverage of Google AI Mode, Microsoft Copilot, Grok and Meta AI are on your requirements list.
- You want to prove value before opening a procurement cycle, or you are an agency: a published $250/mo entry price, a 30-second no-login audit, a self-serve trial and a $500/mo agency tier.
- You are measured on visibility that improves, not visibility that is reported.
- The bottleneck is producing and shipping the content itself — at volume, with a calendar, assignments and approvals.
- You want the fix to live in your CMS, on pages you own and rank, rather than in a second rendering served at the edge.
- SEO still drives most of your traffic and you refuse to run two disconnected tools.
- You want AI agents that draft, optimise, localise and publish, governed at workspace level.
- Procurement will ask for SOC 2 Type II, SAML SSO and AI consumption reported per workspace and per user.
Feature by feature
Every Scrunch AI cell links to the page we read it on, with the date. Six rows go to Scrunch; they say so.
From insight to published page
Both products turn a gap into a brief. They part company over where the fix ends up — in your CMS, or in a copy served to agents at the edge.
| Capability | Semji | Scrunch AI |
|---|---|---|
| Detected gap becomes a content brief | SemjiYes. The gap opens as a brief in Content Hub, with Search Intents, Questions and Related Terms already populated for the editor | Scrunch AIYes. Content Gaps — “Turn any gap into a content brief in seconds—competitive context baked in.” The brief carries the format, the questions to cover and who is cited today, then offers direct hand-off links to ChatGPT and ClaudeSource for Detected gap becomes a content brief, checked 1 August 2026 |
| Approved brand knowledge applied at generation | SemjiYes. Knowledge Base and Brand Voice are workspace assets, on by default as agent tools, applied while the draft is being written | Scrunch AIPartial. Knowledge Studio builds Knowledge Documents from Notion, SharePoint and file connectors, with drift detection when a source changes. It is “available in beta free of charge for all customers on Scrunch's enterprise plans”, with pricing to come at general releaseSource for Approved brand knowledge applied at generation, checked 1 August 2026 |
| Editorial planning, assignments and approvals | SemjiYes. Planning calendar, custom draft statuses (Draft / Writing / In Review / Ready / Published), threaded comments with assignment and resolve, version history, bulk actions | Scrunch AIPartial. A real review workflow — comment, edit and approve, with agents triaging the queue — but it governs brand Knowledge Documents, not article drafts. No editorial calendar and no draft-status pipeline is publishedSource for Editorial planning, assignments and approvals, checked 1 August 2026 |
| Optimisation score on the thing you are writing | SemjiYes. Content Score live in the editor across seven families — Title tag, Length, Search Intents, Questions, Related Terms, Incoming links, Outgoing Links | Scrunch AIPartial. Site Maps grades each mapped URL with an Audit Score out of 100 (the demo shows “Good 87%”) beside that page's agent traffic by engine and its citation count. It grades a published page for AI retrievability, not a draftSource for Optimisation score on the thing you are writing, checked 1 August 2026 |
| Publishing to your own CMS | SemjiYes. WordPress including Yoast meta descriptions, Drupal, Contentful, Strapi, a custom MCP server and a send-to-CMS webhook; the CMS Publisher agent holds create, read, list and update scopes | Scrunch AIPartial. Scrunch's published route is an MCP recipe — “Supported integrations: Sanity, WordPress, HubSpot, and any CMS with an MCP connector.” Its own tool table lists the CMS MCP as “Recommended”, not shipped: you supply the connector, your AI client drives it, Scrunch holds no CMS credentials, and with none connected the fallback is a “Generate only” tab you paste fromSource for Publishing to your own CMS, checked 1 August 2026 |
| Serving AI agents an optimised copy at the CDN | SemjiNo. No edge layer, by design — Semji changes the page itself, so agents and humans read the same content | Scrunch AIYes. AXP intercepts AI traffic at the CDN layer (Akamai, Cloudflare, Vercel) and serves real-time retrieval bots structured, JavaScript-free pages, while “Google and Bing indexing crawlers continue to see your normal site”. No code changes, no CMS releaseSource for Serving AI agents an optimised copy at the CDN, checked 1 August 2026 |
AI visibility measurement
Scrunch's home ground, and mostly parity. Two rows go to Scrunch outright, and they are the two metrics Semji has never shipped.
| Capability | Semji | Scrunch AI |
|---|---|---|
| Prompt tracking, mention and citation rates | SemjiYes. AI Visibility Score out of 100, mention rate, citation rate and average mention position, on prompts carrying topic, persona, funnel stage and real monthly search volume | Scrunch AIYes. Brand presence, competitive presence, answer share and position, per prompt and per platform, with prompt-level response inspectionSource for Prompt tracking, mention and citation rates, checked 1 August 2026 |
| Cited-source analysis | SemjiYes. GEO Sources aggregates cited sources by AI platform, source type, page type and top root domain, with a prompts-citing-a-source column and export | Scrunch AIYes. Top Domains Cited with share, Citations by Owner split across your brand, competitors, third party and social, breakdown by content type, citation trends and an Influence ScoreSource for Cited-source analysis, checked 1 August 2026 |
| Recommendations that tell you what to do next | SemjiYes. GEO recommendations carry the prompts they would move and go to Planning in one click, next to competitor ranking within the topic | Scrunch AIYes. Insights scores page diagnostics across Access Controls, Content Delivery and Content Quality, proposes brands you are not tracking (“we've identified 17 potential competitive brands”), and runs a Greedy Coverage Curve that trims your prompt set while preserving coverageSource for Recommendations that tell you what to do next, checked 1 August 2026 |
| Sentiment analysis | SemjiNo. Not available | Scrunch AIYes. Sentiment in answers, trended over time, broken down by persona and benchmarked against an industry average — the product shot shows 85% against a 72% market average. Ticked on Core, not gated to EnterpriseSource for Sentiment analysis, checked 1 August 2026 |
| Share of voice as a named metric | SemjiNo. Semji reports mention rate, citation rate and competitor ranking within a topic, but publishes no share-of-voice figure | Scrunch AIYes. Share of voice is one of the metrics Scrunch says traditional SEO tools cannot measure, defined on its own FAQ as “How often do responses mention a user's brand versus a competitor's?” and reported per prompt and per platformSource for Share of voice as a named metric, checked 1 August 2026 |
| Engines included in your plan | SemjiChatGPT, Google AI Overviews, Gemini, Claude, Perplexity and DeepSeek — six, on every plan, with no per-model add-on. DeepSeek is API-only, with no search and no citations | Scrunch AINine on Enterprise: ChatGPT, Claude, Perplexity, Gemini, Meta AI, Google AI Mode, Google AIO, Copilot and Grok. Core at $250/mo gets four — ChatGPT, Perplexity, Google AIO and Copilot. No per-engine add-on price is published; the rest arrive as “Expanded model coverage” with an Enterprise quoteSource for Engines included in your plan, checked 1 August 2026 |
Search, traffic and the site itself
AI answers have not replaced Google. Scrunch says plainly that it complements an SEO tool rather than replacing one — and it reads your server logs in a way Semji cannot.
| Capability | Semji | Scrunch AI |
|---|---|---|
| Rank tracking in Google search | SemjiYes. Daily rank tracking, Focus Keyword, keyword difficulty, Top 1–3 / 4–10 / 11–100, branded versus non-branded traffic, search intents, cannibalisation detection and Competition duels | Scrunch AINo. “SEO tools manage Google rankings; Scrunch manages AI visibility.” Scrunch is explicit that teams run both in parallel, and leaves keyword rankings and backlinks to the other toolSource for Rank tracking in Google search, checked 1 August 2026 |
| Real search demand behind a prompt | SemjiYes. Actual monthly search volume on every tracked prompt, from SEO data providers, next to its topic, persona and funnel stage | Scrunch AIPartial. The Keyword Volume Estimator gives estimated weekly US AI-assistant volume, a demand score out of 100, a commercial-intent score out of 100 and a 20-week trend — in beta with access granted by a CSM, US-only with history from mid-2025, and explicitly excluding Google AI Overviews. Scrunch calls it “a calibrated estimate, not a census”Source for Real search demand behind a prompt, checked 1 August 2026 |
| Traffic, conversions and revenue from AI answers | SemjiPartial. Search Console Clicks, Impressions and Avg. position sit on each page's Performance tab, and GA4, Matomo, Adobe Analytics and Piwik Pro connect — but Semji does not isolate revenue from AI referrals | Scrunch AIYes. AI Referrals reads a connected GA4 property and reports AI-referred sessions and page views, which platform sent them, top landing pages, conversion metrics and purchase revenue from AI traffic. Adobe Analytics feeds the same dashboard through the documented AI Referrals push API, so on analytics connectors this is close to parity — Semji's remaining edge is Search Console, Matomo and Piwik ProSource for Traffic, conversions and revenue from AI answers, checked 1 August 2026 |
| AI bot and crawler log analytics | SemjiNo. Not available — Semji ingests no server or edge logs | Scrunch AIYes. Agent Traffic reads real bot hits through native Akamai, AWS CloudFront, Cloudflare, Fastly, Vercel and WordPress integrations, plus a Cloudflare-proxy route for HubSpot CMS, BigCommerce, Shopify, Squarespace, Uberflip, Webflow and Wix. Each visit is classified training, indexer or retrieval, broken down by bot and by page. Ticked on CoreSource for AI bot and crawler log analytics, checked 1 August 2026 |
AI agents and automation
Both automate. The question is what comes out of the pipeline — a finished draft in your CMS, or a new version of a page at the edge.
| Capability | Semji | Scrunch AI |
|---|---|---|
| Prebuilt agents that produce work | SemjiYes. About 24 templates — Content Master, GEO Expert, E-E-A-T Auditor, Content Opportunities, Top 3 Hunter, Legal Compliance, Content Localizer, Social Repurposer, CMS Publisher, Product Page Specialist — all on Claude Sonnet 5, in the editor or across Planning in bulk | Scrunch AIPartial. Three review agents — Compliance, Fact Check and Conflict — that check company knowledge for factual errors, conflicts and sensitive material. They sit inside Knowledge Studio, which is beta and Enterprise-plan only; there is no catalogue of task-specific content agentsSource for Prebuilt agents that produce work, checked 1 August 2026 |
| Agents that draft, optimise and publish end to end | SemjiYes. Bulk mode across Planning with an Activities log and relaunch on failure; agent tools reach the Knowledge Base, Brand Voice, GEO and SEO recommendations, Google search and page fetch | Scrunch AIPartial. The Optimize and Deploy pipeline chains Site Audit, a Content Optimizer that “Rewrites page content to improve AI visibility for target prompts and personas”, and a deploy stage that “Publishes optimized content to your AXP site as a new version” — tracked in Version History with original and optimised side by side. It is “currently in early access and available on request”, and the publish target is AXP rather than your CMSSource for Agents that draft, optimise and publish end to end, checked 1 August 2026 |
| MCP server | SemjiYes. mcp.semji.com — OAuth Authorization Code with PKCE, a permission system, audited tool-call logs and the full draft lifecycle; listed in the Claude connectors directory | Scrunch AIYes. OAuth through normal Scrunch sign-in with no API key, across Claude, ChatGPT in Developer Mode, Microsoft Copilot Studio, Cursor, VS Code and Windsurf. It reads metrics and responses and writes brands, competitors, personas and prompts, with a published library of 35+ prompts and multi-tool workflows. Documented limits: no cited-URL detail, no Sitemap or Page Audit data, 1,000 records per requestSource for MCP server, checked 1 August 2026 |
| Agents inside Slack and Microsoft Teams | SemjiYes. A Microsoft Teams bot on AppSource with a proactive Content Opportunities agent — opportunity cards, in-thread interaction, push to Planning — plus Slack | Scrunch AINo. Slack appears on Scrunch's plan matrix only as an Enterprise support channel (“Support Channels: Email + Slack”). Neither Slack nor Teams is a product row on it; posting to a channel is a workflow recipe that needs your own Slack MCP alongside Scrunch'sSource for Agents inside Slack and Microsoft Teams, checked 1 August 2026 |
Enterprise readiness
The section procurement reads. Scrunch clears the security bar and beats Semji on the one thing procurement never gets to see: a trial before the paperwork.
| Capability | Semji | Scrunch AI |
|---|---|---|
| SOC 2 Type II | SemjiYes. Certified | Scrunch AIYes. Certified, with a public Trust Center carrying the badge, an access-gated SOC 2 Type 2 report, eight named policies and Vanta-monitored controls. Its three published subprocessors — Google Cloud Platform, Clerk and Neon — are all listed in the US, and no EU data-residency option is publishedSource for SOC 2 Type II, checked 1 August 2026 |
| SSO / SAML | SemjiYes. SAML with automated certificate rotation | Scrunch AIPartial. SAML 2.0 and OIDC with Okta, Microsoft Entra ID and Google Workspace named, plus just-in-time provisioning and domain enforcement — but “SSO is available on the Scrunch Enterprise plan”. Core gets Google SSO only, and activation runs through support rather than self-serveSource for SSO / SAML, checked 1 August 2026 |
| Evaluate it before you talk to sales | SemjiNo. No free trial and no free tier — evaluation runs through a scoped demo on your own prompts | Scrunch AIYes. A 7-day self-serve trial of Core with all 125 prompts, no sales call. Scrunch's own FAQ is clear about the terms: “Scrunch requires a credit card to start the 7-day free trial. Users will automatically be upgraded to a paid subscription after the end of the trial unless they cancel.”Source for Evaluate it before you talk to sales, checked 1 August 2026 |
| How you buy | SemjiQuote-based, with no public price — scope follows workspaces, markets and AI consumption, and every plan includes the whole platform | Scrunch AICore is $250/mo: 125 prompts, 5 licences, 1 brand workspace, 4 of 9 engines, 1 country, 25 mapped pages and 1 page optimisation a month. AXP, Query and Advanced API access, Looker Studio, SAML SSO, Product Observability and a dedicated account team are all dashes on it. Enterprise is custom-quotedSource for How you buy, checked 1 August 2026 |
See where your brand actually stands
Run a free GEO audit on your own domain. No signup, results in about two minutes.
- Free
- No signup
- Results in 2 minutes
Three things a delivery layer cannot do
Not feature gaps — structural ones. AXP is a good answer to a real problem. It is an answer about rendering, and these three are about the page.
The fix lives on the page you own
- A prompt shows a competitor cited and your brand absent
- Semji opens a brief, or the existing page in the editor
- An agent drafts in your brand voice, grounded in your knowledge base
- Scored, reviewed, and published to WordPress, Drupal, Contentful or Strapi
SEO and GEO on one set of pages
- One topic, persona and competitor model across search and AI answers
- Daily rank tracking and AI visibility on the same page record
- Real monthly search volume on the prompt, not an estimated score
Production your legal team will sign off
- Brand Voice and Knowledge Base applied at generation, not after
- Legal Compliance and E-E-A-T Auditor agents on the same draft
- Custom statuses, threaded comments and version history before anything ships
- SOC 2 Type II, SAML SSO, and AI consumption reported per workspace and user
Frequently asked questions
Can I run Semji and Scrunch AI together?
Better than most pairs on this list, because the overlap is narrow. Scrunch owns the delivery and diagnostic layer — edge-served pages for agents, bot logs, site audits — and Semji owns the editorial pipeline that ends in your CMS: calendar, statuses, approvals, in-editor scoring and one-click publishing. The genuine duplication is prompt monitoring and cited-source analysis, which both do well, and the GA4 and Adobe Analytics plumbing, which both read. Teams that keep both usually consolidate the monitoring eventually, but there is no technical reason to rush it.
What does Scrunch AI do better?
A lot, and it is worth being specific. AXP has no Semji equivalent: it fixes AI readability at the CDN without a CMS ticket, with version history and side-by-side diffs. Agent Traffic reads real edge logs and tells you which bot crawled which page and why — Semji cannot answer that at all. It scores sentiment and share of voice, which Semji does not report. It tracks Google AI Mode, Microsoft Copilot, Grok and Meta AI, which Semji does not. It runs an AI-retrievability crawler and site audit; Semji has no crawler. It tracks product-level shopping visibility; Semji has nothing at SKU level. Its MCP has a far deeper published recipe library than ours, and Signals and Explorer give it automated anomaly detection and natural-language analytics that Semji lacks. Commercially, it publishes a price, a 30-second no-login audit, a self-serve trial and a $500/month agency tier with a partner programme; Semji publishes none of those.
Which AI engines does each tool cover?
Semji tracks six on every plan with no per-model add-on: ChatGPT, Google AI Overviews, Gemini, Claude, Perplexity and DeepSeek — DeepSeek via API only, with no search and no citations. Scrunch lists nine on Enterprise: ChatGPT, Claude, Perplexity, Gemini, Meta AI, Google AI Mode, Google AI Overviews, Microsoft Copilot and Grok. The published $250/month Core plan includes four of them — ChatGPT, Perplexity, Google AI Overviews and Copilot — and no per-engine add-on price exists, so Claude, Gemini, Google AI Mode, Grok and Meta AI all sit behind a custom quote. Scrunch's advantage on Google AI Mode, Copilot, Grok and Meta AI is real; it is an Enterprise-tier advantage.
What does the Sitecore acquisition mean for Scrunch AI?
Scrunch announced on 3 June 2026 that it had been acquired by Sitecore, and now presents itself as a Sitecore company. Its own post says it will “continue to operate as a standalone platform, support the customers and partners we serve today, and accelerate our product roadmap.” We report that flatly and take no view on how it plays out. The point for a buyer is that the thing being evaluated is now a component of a large DXP vendor's portfolio rather than an independent tool, and the standalone commitment is a statement of intent rather than a contract — so whose roadmap the product follows, and how it is priced inside a suite, are fair questions to put on the table before signing a multi-year deal.
How does pricing compare?
Scrunch publishes Core at $250 per month: 125 unique prompts, 5 user licences, 1 brand workspace, 4 of 9 engines, 1 country, 2 languages, 25 mapped pages and 1 page optimisation a month — positioned by Scrunch itself as the tier on which to evaluate and benchmark before scaling. AXP, API access, Looker Studio, SAML SSO and product-level shopping visibility are all Enterprise, which is custom-quoted, so an enterprise comparison is quote against quote. Agency Core is $500 per month, published in an FAQ rather than on the pricing page. The 7-day trial is genuinely self-serve, but it takes a credit card and converts to a paid subscription unless you cancel. Semji is quote-based because scope varies with workspaces, markets and AI usage, and because the platform covers production as well as measurement — the demo is the fastest route to a number.
Serve a copy, or publish the page
Book a demo and we will run your own prompts through both halves of the platform — the visibility report, and the content Semji ships to close the gap.
Scrunch and Scrunch AI are trademarks of Excelsa Labs, Inc., dba Scrunch AI. Semji is not affiliated with Scrunch AI or Sitecore. This comparison is based on publicly available information as of 1 August 2026; products change, so check both vendors' current documentation before deciding.