The claim
The world's first engine that runs your brand's entire presence, from the map to the model, autonomously, under your approval.
First on 18 September 2026, and first is not a moving target. As at that date, no other product did all of it in one place.
No instructions needed. It finds the work across Maps, Search, AI Search, Social and Reviews, ranks it and runs it day by day. Give it a goal and it plans to it. Everything lands in one approval queue.
The engine runs. You decide.
One engine runs your entire presence: Maps, Search, AI Search, Social and Reviews. It plans. It executes. It publishes. You approve. As at 18 September 2026, no other product did all of it in one place. This page sets out the tests a platform must pass to match the claim, then shows the evidence, vendor by vendor, with a source and a date for every line.
What the words mean
Every word in the claim, defined
Engine
One product that finds the work, plans it, then executes and publishes it across the five surfaces, with a goal from you or without one. Not a dashboard that hands you a list. Not a builder you drive one task at a time.
Entire presence
The five surfaces a buyer uses. Maps is Local. Search is SEO. AI Search is GEO. Social is Social. Reviews is Perception. Your website is not a sixth surface. It is the foundation, the crawl layer that Search and AI Search write into.
Autonomously
CLEO plans, executes and publishes across the five surfaces without a person starting each task, on every platform that permits automated publishing. Left to itself, it finds its own opportunities, ranks them and works through them day by day, so the scores move without instruction. Give it a goal and it writes a dated plan, runs it on a cadence over weeks and re-plans from what the results show. Ask it for one thing, by email, and it does that too. Where a platform does not permit automated publishing, the engine prepares the post and a person places it. No one can automate what a platform forbids.
Under your approval
Approval depth is set per surface: approve every action, approve by category, or approve once. Every action is logged with its before and after. Applied fixes reverse in one click. Named Approvers hold the yes.
From the map to the model
From the map that finds you to the model that answers for you.
A claim you can test is a claim worth making.
The tests
The four tests
A platform matches the claim only if it passes all four, in one product. Tick a test to filter the evidence table below. The second column is the check you can run yourself, without asking us.
Five surfaces in one product: Maps, Search, AI Search, Social, Reviews.
Open the vendor's product menu. Count the five surfaces.
It executes on each surface, and it does not wait to be told. It finds the opportunities itself, ranks them and works through them day by day. Given a goal, it produces a dated plan and re-plans from measured results. It does not only recommend, and it does not only run one task at a time.
Ask whether it publishes, or hands you a list to publish. Then connect one site, give it no instructions for 30 days, approve what it proposes, and compare the scores. Then give it one goal, such as "rank for X and be cited for Y within 90 days", and ask to see the dated plan and how it changed after the first month.
It writes site fixes server-side into your own CMS or code repository, by default and across stacks, not as an optional export. The fixes persist in your own source and survive removal of the vendor. Not a script tag. Not an overlay. Not a hosted page. Not a ticket.
Turn the tool off, then run curl -s https://yoursite.com | grep canonical, and again with grep application/ld+json for the schema block. If it is gone, it was never in your source.
One approval queue. A full action log. One-click reversal.
Ask to see one queue containing an SEO fix, a social post and a review reply, with a before and after on each.
Nothing ticked yet. Tick a test to filter the evidence table below to the platforms that do not meet it.
Four tests. One product passes.
Why test 3 decides it
AI crawlers read the page a server sends. Most do not run the JavaScript that arrives with it. Analysing more than 500 million GPTBot requests, Vercel and MERJ found no evidence of JavaScript execution; the same report finds the crawlers behind Claude, Perplexity, Meta and ByteDance also do not render (The Rise of the AI Crawler, December 2024). Google's Gemini and Apple's AppleBot are the exceptions: both render, because both ride on existing search infrastructure. So a fix delivered by a script tag or an overlay is invisible to most of the engines now answering your buyers, and it disappears the day the tool is removed. A fix written into your own CMS or code is seen by every crawler and stays.
If the machine cannot read it, it did not happen.
Where others lead
What other platforms do better
Yext reaches more than 200 listing endpoints. CLEO's Local is Google only. Profound's builder and prompt data are more mature workflow tooling than we expose. Semrush and Ahrefs hold deeper keyword and backlink data. Search Atlas executes across more sites. We say so plainly, because a first that hides the field is not a first worth trusting.
HubSpot, Salesforce, Adobe and Braze lead on campaign and journey orchestration across email, ads, SMS and CRM. They take a goal and build a plan, and they re-plan from results, at a depth and with an enterprise data foundation we do not claim to match on those channels. What none of them does is run that plan across Maps, server-side site SEO, AI search and third-party reviews together, written into your own source. Search Atlas on SEO, SOCi on local and Profound on AI search each find and act on their own opportunities without being asked. Each does it on its own surface.
Honest about the field. Certain about the seam.
The evidence
Every platform reviewed, test by test
Each row says what the platform does well, then whether it finds and plans the work itself, then where it stops, then what you still have the day after you cancel. Checked on 18 September 2026, with the rows re-checked on 20 September 2026 marked in the Source column. "Tests not met" refers to the four tests above, and a platform matches the claim only when that column is empty.
A fix that disappears when you cancel was never yours. This column says what you keep.
| Platform | What it does well | Finds and plans the work | Surfaces, execution and how site changes are made | What survives if you cancel | Tests not met | Source |
|---|---|---|---|---|---|---|
| AI-VISIBILITY SPECIALISTS | ||||||
| Profound | AI search analytics across up to nine engines. A no-code builder for "Agents", with an approval step before anything publishes. Publishes content to WordPress, Sanity, Contentful, Webflow, Framer and Adobe Experience Manager. | Aim (July 2026) watches AI search signals, opens a project with tasks and routes them to an execution agent. AI Search only. | Centred on AI Search. Page-level health on tracked pages rather than a site-wide technical SEO audit and fix suite. No Google Business Profile, third-party review management or multi-platform social listening shown publicly. | Published content remains in your CMS. Analytics and "Agents" configuration do not. | 1, 3 | tryprofound.com (product, Agents and blog pages) |
| AthenaHQ, Peec AI, Otterly, Goodie, Evertune, Bluefish | AI answer monitoring, competitor benchmarking, recommendations. Some draft content. | Monitoring and recommendations. No goal-to-plan execution. | One surface. Monitoring and recommendation. None starts at the crawl layer. | Any drafts you published remain where you published them. Monitoring stops, and nothing else is applied to your site. | 1, 2, 3 | Vendor sites |
| Scrunch | Monitoring, plus delivery of an AI-readable version of a site to AI crawlers. | None. A delivery layer, not a planner. | One surface. A delivery layer in front of the site rather than a write into the customer's own source. | Nothing. The AI-readable version is delivered in front of your site. | 1, 3 | scrunch.com |
| LOCAL AND REPUTATION PLATFORMS | ||||||
| Yext | Listings pushed to 200+ publisher endpoints, far beyond CLEO's Google-only Local. Reviews, Social, Pages, and Scout for AI search monitoring. Executes across listings, reviews and social. | Scout surfaces actions and the platform can trigger listing, review and social updates. No plan that spans site SEO or AI search execution. | Broad. Does not write site-wide technical SEO fixes into the customer's own CMS or code; it manages its own Pages product and publisher network. | Yext stops sending data to the publishers. Listings stay live but are no longer updated, and drift as each publisher falls back on other sources. Pages are Yext's own product and go with the subscription. | 3 | yext.com (platform and Scout pages); help.yext.com, "What happens when a Listings subscription is cancelled", read 20 September 2026 |
| Birdeye | Reviews, listings on Google, Apple, Facebook, Yelp and Bing, social, AI search visibility. Automated responses and outreach. | BirdAI agents (September 2025) complete local marketing jobs end to end against an outcome, with supervised autonomy. Listings, reviews, social and messaging only. | Broad on reputation. No server-side technical fixes on the customer's own site. | Published review responses remain on the review platforms. Birdeye stops sending listing data. | 3 | birdeye.com |
| GoHighLevel | CRM, conversations, reviews, social, funnels. Reviews connected to Google Business Profile. | AI Employee and workflows trigger calls, messages and content from rules. CRM-led; no presence plan. | Assembled modules. No site-wide technical SEO fix suite. No cross-engine AI answer monitoring. | Published posts remain. Funnels, pages and conversation history live in GoHighLevel's own modules. | 1, 3 | gohighlevel.com |
| Vendasta | Listings, reviews, social, SEO and ads for agencies to resell. Automated social and WordPress blog publishing with mobile approval. | AI Employees handle lead capture, conversations and review responses for agencies to resell. Job by job, not a cross-surface plan. | A partner and fulfilment platform. Publishes posts, not site-wide server-side technical fixes across stacks. | Published blog posts and social posts remain. | 3 | vendasta.com |
| SOCi, Uberall, Reputation, Podium, Rio SEO, BrightLocal | Depth in multi-location listings, reviews and local social. | SOCi Genius agents detect, decide and act across listings, reviews and Google Posts per location. Local only. | No server-side technical SEO fixes on the customer's own site across the five surfaces. | Published responses remain. Listing management stops. | 3 | Vendor sites |
| AUTONOMOUS SEO TOOLS | ||||||
| Search Atlas (OTTO) | Executes technical SEO, content, schema, internal links, Google Business Profile tasks and paid search across a very large installed base, with an approval mode and rollback. | OTTO finds and deploys SEO fixes on its own, and Atlas Agent is described as running SEO, ads, content and site health without prompting, with approval and rollback. The closest single vendor on self-directed execution; see test 3. | Changes are deployed through a JavaScript pixel, a Cloudflare Worker or a WordPress plugin. Search Atlas documents that OTTO uses dynamic client-side rendering, applied in the user's browser and not in the CMS or static HTML; its WordPress plugin can optionally save a subset of optimisations (meta titles and descriptions, alt text, headings, internal links) into native WordPress fields. Persistence is therefore opt-in and limited to one CMS. Social publishing covers Facebook, Instagram and X. Reviews are Google Business Profile only, not third-party review platforms. On AI Search, its "LLM Visibility" product measures brand mentions, sentiment, share of voice and ranking inside AI-powered search and chatbot answers, naming ChatGPT, Gemini, Perplexity, Claude, SearchGPT, Grok and Copilot; it monitors rather than executes, with a separate "Coworker" feature that surfaces drops and ships fixes. | Google Business Profile updates remain, and so does anything the WordPress plugin saved into native WordPress fields. Pixel-deployed and Cloudflare Worker fixes stop rendering. | 1, 3 | searchatlas.com; help.searchatlas.com (OTTO setup and deployment), read 20 September 2026; wordpress.org/plugins/metasync, read 20 September 2026; searchatlas.com/llm-visibility, read 20 September 2026; PR Newswire, 16 October 2024 |
| Alli AI, BrightEdge Autopilot, Conductor, Surfer, AirOps, Gradial | SEO and content execution or optimisation. | Sequenced SEO and content workflows. SEO and content only. | SEO and content only. Script injection or content publishing rather than five-surface execution. | Published content remains. Script-injected changes stop rendering. | 1 | Vendor sites |
| Adobe LLM Optimizer | Serves an AI-optimised version of pages at the edge, without JavaScript. | None. Edge delivery. | One surface. Edge delivery, oriented to Adobe Experience Manager. | Nothing. Optimised pages are served at the edge. | 1, 3 | business.adobe.com |
| MARKETING SUITES | ||||||
| Semrush, Ahrefs | Keyword and backlink data far deeper than CLEO's. Local, social and AI visibility toolkits (Semrush). | None. Insight and recommendation. | Insight and recommendation. No execution into the customer's own stack. | Not applicable. Nothing is applied. | 2, 3 | semrush.com; ahrefs.com |
| MARKETING SUITES WITH GOAL-DRIVEN CAMPAIGNS | ||||||
| HubSpot (Breeze) | The Campaign Agent takes one goal and builds a full multi-channel campaign plan with content and automation, and each campaign informs the next. A genuine goal-to-plan loop. | One goal to a full campaign plan, and each campaign informs the next. Email, CRM, social and content only. | Email, CRM, social and content. Publishes only to HubSpot's own CMS. No Google Business Profile, no third-party reviews, no server-side SEO into other stacks. | Content remains only if you keep HubSpot CMS. | 1, 3 | hubspot.com/spotlight; hubspot.com/products/artificial-intelligence |
| Salesforce (Agentforce for Marketing) | Marketers define goals, budgets and guardrails; the Marketing Goals Agent chooses content, audience, channel and timing and re-plans across journeys. In pilot at the date checked. | Goal, budget and guardrails in; content, audience, channel and timing out, re-planned across journeys. Email, SMS, ads and CRM only. | Email, SMS, journeys, ads and CRM. No Maps, no server-side SEO, no third-party reviews. | Journeys and content inside Salesforce. | 1, 3 | salesforce.com (Agentforce Marketing announcement) |
| Adobe (Agent Orchestrator, GenStudio) | Coordinates content, audience, journey and experimentation agents. Generally available since September 2025. | Agent Orchestrator coordinates the four agents against a brief. Enterprise content and journeys only. | Enterprise content and journeys. No Maps, no reviews, no server-side SEO across other stacks. | Content and journeys inside Adobe. | 1, 3 | news.adobe.com; business.adobe.com |
| Braze | Agent Console builds agents that plan, test and optimise campaigns, with a closed loop inside the journey. | Agent Console plans, tests and optimises inside the journey. Messaging channels only. | Messaging channels: email, push, SMS, in-app. | Campaigns inside Braze. | 1, 3 | braze.com |
| Jasper | Agents and Canvas plan and run content campaigns, with delivery through Braze. | Plans content campaigns, delivered through Braze. Content and campaign orchestration only. | Content and campaign orchestration. No Maps, no server-side SEO, no reviews. | Published content remains. | 1, 3 | jasper.ai |
| ActiveCampaign | Goal-based automation that orchestrates cross-channel campaigns and optimises for outcomes. | Goal-based automation, optimised for the outcome set. Email and SMS led. | Email and SMS led. No Maps, no server-side SEO, no reviews. | Automations inside ActiveCampaign. | 1, 3 | businesswire.com (ActiveCampaign announcements, 2025 and 2026) |
| SMALL-BUSINESS "DONE FOR YOU" PRODUCTS | ||||||
| Wix, GoDaddy Airo, Squarespace, Shopify Magic | SEO, Google Business Profile and social tools inside their own site builders. | GoDaddy Airo.ai (beta, November 2025) infers a goal and acts across GoDaddy products. All work only on their own hosting. | Work only on sites hosted with them. No cross-engine AI answer monitoring with citation-gap content. | Everything, as long as you stay on their hosting. | 1, 3 | Vendor sites |
| EARLIER "WORLD'S FIRST AUTONOMOUS" CLAIMS | ||||||
| Albert.ai | Described at launch as "the first autonomous artificial intelligence platform for digital marketers". Runs paid media across Google, Meta, YouTube, Bing and programmatic. | Goal-based optimisation of paid media only. | Paid media only. | Campaigns run in your own ad accounts on Google, Meta, YouTube and Bing, and stay there. | 1, 3 | albert.ai |
| Search Atlas (OTTO), October 2024 | Announced as "the world's first AI SEO agent to optimise websites autonomously". | OTTO finds and deploys SEO fixes on its own. SEO only, and the October 2024 claim is the one this row records. | SEO only. See the row above. | See the row above. | 1 | PR Newswire, 16 October 2024 |
And CLEO, in the same terms
Finds and plans the work
Self-directed across all five surfaces: finds, ranks and executes its own improvements daily. Given a goal, runs a dated plan re-planned from rankings, AI citation share and the CLEO Presence Index. Takes one-off requests by email. One approval queue.
What survives if you cancel
Everything. Fixes are already in your own CMS or code.
They are written into your own CMS or code repository, so they are already your source before you cancel and they are still your source afterwards. Articles sit on your site. Profile updates sit on Google. Review replies sit on the review platforms. Cancelling CLEO removes the engine, not the work. That is the whole of test 3, and it is the one line in this table you can check without talking to anybody.
Read every row. The pattern is the point.
What Orbit runs today
CLEO runs on CLEO
From 20 September 2026, CLEO's own presence runs all five surfaces on Orbit.
Maps | Local
Google Business Profile and Google reviews. Google only.
Search | SEO
Audits, rank tracking and seven classes of fix written server-side into WordPress, Wix, Shopify, Sanity and code repositories.
AI Search | GEO
Prompts watched six days a week across the AI engines at 95% confidence. Articles commissioned from citation gaps, written by Quill, published on approval.
Social
Listening on X, LinkedIn, Reddit, Medium, YouTube, Quora and Bluesky. Publishing by itself wherever the platform permits: LinkedIn, X, YouTube and Bluesky today, with Instagram to follow once Meta grants API access. Elsewhere the engine prepares the post and a person places it.
Reviews | Perception
G2, Capterra, Trustpilot, TrustRadius and Tripadvisor fetched and replies drafted for a person to post, in line with each platform's terms. Google reviews are answered directly under Maps.
In beta
Leads and Ads. The claim does not rest on them.
In productionOne agency runs 46 client sites on Orbit for SEO. CLEO's own site moved AI Readability 35 to 96, SEO 40 to 95 and GEO 14 to 50 in thirty days. DisburseCloud moved AI citation share from 17% to 67% in ninety days. 200+ sites live in 8 countries. Client site audits are recorded site by site at regencleo.ai/results, with the third-party SE Ranking audit score before and after, and the days each site took.
See the record →Measured, not projected.
How we reviewed
The method, and its date
We read each vendor's public product pages, documentation, pricing pages and release notes on 18 September 2026, with the rows re-checked on 20 September 2026 marked in the Source column. This is a point-in-time review. Vendors ship quickly. We re-check every quarter and after any major release. If a page has changed since our date, tell us. We re-check the table quarterly because what each vendor ships changes. The date on the claim does not.
Challenge this
Challenge the claim
Being first is a matter of record, not of opinion, so there is only one way to disprove it: show us a platform that passed all four tests before 18 September 2026. Send it to contact@regencleo.ai. We answer within five working days, publish the outcome here, and withdraw the claim if the evidence holds.
Platforms that pass all four tests after that date have caught up, not arrived first. We will record them here as they do.
Challenges received to date: 0.
Show us prior art →Change log
- 21 September 2026
- First publication.
- 21 September 2026
- Correction. An earlier version of this page stated that no AI crawler renders JavaScript. The study cited here (Vercel and MERJ, The Rise of the AI Crawler, December 2024) names two exceptions that do render, Google's Gemini and Apple's AppleBot, so the page now says most AI crawlers do not render and names both. The claim was restated everywhere it appears on this site, not only here.
- 21 September 2026
- Clarified that the priority claim is fixed at 18 September 2026 and is disproved only by prior art, not by later parity. The quarterly cycle re-checks the evidence table, not the claim.
- 21 September 2026
- Corrected the Search Atlas row. Test 3 wording tightened to specify default, cross-stack and persistence. Vercel and MERJ finding restated to name Gemini and AppleBot as rendering exceptions. What survives if you cancel column added. Clarified that the priority claim is fixed at 18 September 2026 and is disproved only by prior art.
- 22 September 2026
- Engine and Autonomously definitions and test 2 restated to describe how the engine runs: self-directed, to a goal and on demand, all of which the product did on or before 18 September 2026. "Finds and plans the work" column added. HubSpot row expanded and rows added for Salesforce, Adobe, Braze, Jasper and ActiveCampaign. "What other platforms do better" now records that the marketing suites lead on campaign orchestration across email, ads and CRM. The claim and its date are unchanged.
- 21 December 2026
- Next scheduled re-check of the evidence table.
A claim that invites challenge expects to win.
FAQ
Frequently asked questions
Is this just marketing?
No. It is a defined claim with four public tests and a vendor-by-vendor evidence table, dated and open to challenge. If anyone shows a platform that passed all four tests before 18 September 2026, we withdraw the claim. If one passes them later, we record it here as a platform that has caught up. Marketing states things. This page shows its working.
What does autonomous mean here?
CLEO plans, executes and publishes across the five surfaces without a person starting each task. It runs three ways. Self-directed: with no instruction at all, it finds its own opportunities, ranks them and works through them day by day. To a goal: you state an objective and it writes a dated plan, runs it over weeks and re-plans from measured results. On demand: you email Orbit a request and it does that one thing. In every case the work lands in one approval queue, and you set how deep approval goes, per surface, at any time.
Does anything publish without my approval?
No. Every action routes to one approval queue, at the depth you set for that surface. Every action is logged with its before and after, and any applied fix reverses in one click.
Why does server-side matter?
Most AI crawlers do not run JavaScript. The crawlers behind ChatGPT, Claude, Perplexity, Meta and ByteDance fetch your HTML and never execute the scripts. Gemini and AppleBot are the exceptions, because they ride on existing search infrastructure. A fix injected by a script tag is invisible to most of the engines answering your buyers, and it disappears if you leave the tool. CLEO writes the fix into your own CMS or code, so every crawler sees it and the fix survives CLEO.
What about Profound, Yext or Search Atlas?
Each is strong. Profound's builder and engine coverage, Yext's 200+ listing endpoints, Search Atlas's execution across many sites. None writes site-wide technical fixes server-side into your own stack across all five surfaces from one approval queue. They show you the gap. CLEO fills the gap.
What is in beta, and where does a person step in?
Leads and Ads are in beta, and the claim does not rest on them. On Social, Orbit publishes by itself wherever the platform permits automated publishing: LinkedIn, X, YouTube and Bluesky today, with Instagram to follow once Meta grants access. Where a platform does not permit it, the engine prepares the post and a person places it. No one can automate what a platform forbids.
Is Orbit in beta?
Orbit is the mode the engine runs in, and it runs all five surfaces on real sites today: 200+ sites live in 8 countries, and one agency runs 46 client sites on Orbit for SEO. Two capabilities inside the engine, Leads and Ads, are in beta, and the claim does not rest on them. Where the product still shows a beta label on a surface, we say so on this page and not in the claim.
Does it need instructions?
No. Connect a site and Orbit finds its own opportunities across the five surfaces, ranks them by expected impact and works through them day by day. You will see what it proposes in one approval queue, and the scores move as you approve. You can also give it a goal, and it will plan to it, or email it a single request. What it never does is publish without the approval depth you have set.
Don't HubSpot, Salesforce and Adobe already do goal-driven campaigns?
Yes, and they do it well. You can give each of them a goal and get a multi-channel plan that re-plans from results, across email, ads, SMS and CRM. That is real, and it came before us on those channels. The difference is the surfaces. CLEO takes one goal and runs it across Maps, Search, AI Search, Social and Reviews, and it writes the site fixes into your own CMS or code, where every crawler can read them and where they stay if you cancel. That combination is the seam.
Next
See what the machines see. Then let the engine run.
Everything runs. You only approve.
See CLEO Orbit →