SEO, GEO and the CMS on one autopilot.
Growzo runs the loop that most teams run by hand: mine the queries, gather the evidence, draft against it, gate it, publish it, ping IndexNow, then probe the AI engines to see whether anything cites you. It ships with its own headless CMS, so the pages it writes have somewhere to live.
Collector, Researcher, Writer, Editor, QA, Publisher, Monitor and Nurture, in that order.
Perplexity, ChatGPT, Claude and Gemini. Three more sit in the registry as stubs and are not counted.
The audit names each agent, because a blanket robots rule hides a site from all of them at once.
The entry tier. Self-hosting on your own keys is free; both carry a hard runtime cost cap.
What Growzo is
Growzo is an open-source, self-hostable SEO and GEO platform with a headless CMS inside it. Eight named workers run in sequence — Collector, Researcher, Writer, Editor, QA, Publisher, Monitor and Nurture — taking a page from a Search Console row through drafting, a pre-publish gate and an IndexNow ping, then measuring whether AI answer engines cite it. The CMS is Strapi-v4 compatible, so a site already fed by Strapi can point at Growzo with a one-line base-URL change.
GEO is the part that is genuinely uncommon. Growzo probes four AI engines live — Perplexity with web grounding, plus ChatGPT, Claude and Gemini — and rolls per-engine citation share, sentiment and word association into an AI Brand Score with a trend line. Its audit module checks the thirteen AI crawler user-agents by name, which matters because a single over-broad robots.txt rule can remove a site from every AI surface while leaving Google Search untouched.
Commercially it works two ways, and it is worth being precise about which: the code is open source and self-hostable, so you can run it free and bring your own API keys, or take a hosted tier from ₹3,000 a month. Either way a hard cost cap is enforced at runtime, so the workers throttle rather than overspend your budget.
What it does
Growzo is the CMS and the measurement in one product, which is the reason a page it publishes can be traced to the query that justified it.
The eight-worker loop
Collector pulls Search Console rows idempotently. Researcher gathers the evidence a page will cite, Writer drafts against that evidence rather than from a blank prompt, Editor revises, QA gates, Publisher publishes and pings IndexNow, Monitor probes the engines and Nurture follows up.
A headless CMS you can migrate onto
A block-based editor with a per-page SEO panel — meta, slug, OG, canonical, robots and JSON-LD — plus taxonomy, authors with credentials and sameAs links, and thirteen content-type field types. Import wizards read Strapi, WordPress and Astro, with a rollback path.
AI citation tracking
Live probes against Perplexity, ChatGPT, Claude and Gemini, rolled into an AI Brand Score out of 100 that combines per-engine citation share, sentiment and word association. It is measured per probe, and the daily probe cap is visible on the plan.
The pre-publish gate
QA holds a draft on a velocity throttle, an AI-content score and YMYL guardrails; a page either passes to queued or lands in qa_failed. Nothing publishes because a scheduler said so.
On-site RAG chat and forms
A chat widget that answers strictly from the project's own published pages, cites the pages it used, refuses when nothing relevant is retrieved, and captures an email as a lead on the second message. Embeddable forms carry a honeypot and per-IP rate limiting.
Three free public tools
An llms.txt generator with a ten-check validator and a 0–100 score, an AI-crawler check that returns a red-green matrix across thirteen agents, and an AI-readiness audit that produces a shareable report. No signup, and the audit covers up to ten pages.
How it runs
One pass of the daily loop, in the product's own module names.
-
Collect
Connect Search Console by OAuth or a service account and Growzo pulls up to sixteen months of history, plus GA4 conversions and Microsoft Clarity aggregates where you connect them. Rows arrive idempotently, so a re-run does not double-count.
-
Research and write
The Researcher assembles the evidence the page will cite; the Writer drafts against that evidence with an E-E-A-T structure and your brand-voice samples as few-shot anchors, rather than inventing from the topic alone.
-
Edit and gate
The Editor revises structure and internal links. QA then applies the velocity throttle, the AI-content score and the YMYL guardrails, and a draft that fails stays failed rather than shipping with a warning.
-
Publish and ping
Publish into Growzo's own CMS, or push to WordPress over REST, to your own site by webhook, or to any generic webhook. IndexNow fires on publish — free, near-immediate, and the fastest route into Bing and the engines downstream of it.
-
Monitor
Probe the four AI engines for citations and check Brave SERP positions, then watch the AI Brand Score trend. The loop runs on a manual trigger or a scheduler you point at it, deliberately, so a new account never spends your API budget before you ask it to.
Who it is for
Three hosted tiers scale by pages published per day, queries tracked and daily probe budget, and the self-hosted path has no tier at all.
Founders publishing on a budget
The entry tier is scoped for one person trying to get cited and ranked without an agency retainer: three pages a day, thirty tracked queries, and a cap that cannot be exceeded.
In-house growth teams
Teams that already have a CMS problem as well as a ranking problem, and would rather solve both in one system than wire a content tool to a separate publishing stack.
Agencies and content-heavy teams
Multiple connected sites, higher publishing volume and a bigger probe budget, with the per-project cost cap keeping several clients' spend separable.
Engineering-led teams who want to self-host
The code is open source and the deployment is yours: your Postgres, your keys, your infrastructure. This is the only version where "free" means zero cost to you beyond the model vendors.
What it does not do
Growzo's own marketing page publishes its limits, which is unusual and worth repeating exactly rather than softening.
- Rank tracking is Brave SERP only. It does not track positions on Google, Bing, Yandex, Baidu, DuckDuckGo or Naver, and it does not detect AI Overview presence — Brave's API does not expose it, so the indicator is not claimed.
- It does not discover backlinks or report a domain rating, and it does not provide keyword volume or difficulty. There is no crawler behind those numbers, so they render as an em dash rather than as a fabricated figure.
- It does not integrate with Ahrefs, Semrush or Moz, and it does not crawl a site site-wide. The AI-readiness audit is capped at ten pages.
- It does not use Google's Indexing API for ordinary content — that API is sanctioned for JobPosting and BroadcastEvent only, and submitting blog posts through it is a documented violation. Discovery runs on real lastmod values, IndexNow and internal links instead.
- It does not place links or run digital PR. Earned coverage and the entity signals behind it are a service: Digital PR →
Answers
Is Growzo free or paid?
Both, depending on where it runs. The code is open source and self-hostable, so you can run it yourself for nothing and pay only the model vendors on your own keys; the hosted product sells tiers from ₹3,000 a month. Either way a hard runtime cap stops the spend.
Which AI engines does it actually probe?
Four, live: Perplexity with web grounding, plus ChatGPT, Claude and Gemini. Three more engines exist in the adapter registry as stubs with no live network call, and they are not counted as capability here.
Do I have to move my CMS?
No. Growzo's own CMS is the default, but it can publish out to WordPress over the REST API, to your own site by webhook, or to any generic webhook. Its read API speaks Strapi v4, so an existing Strapi-fed frontend can point at it by changing one base URL.
Does the on-site chat make things up?
It is retrieval-augmented and system-prompted to answer only from that project's published pages, it names the source pages it used, and it refuses when nothing relevant is retrieved rather than filling the gap.
Does the autopilot start on its own?
Not on signup — deliberately, because the first loop would spend your API budget before you asked it to. You trigger the first run, or point your own scheduler at it.
Can I try the GEO tooling before signing up?
Yes. Three public tools need no account: an llms.txt generator and validator scored out of 100, an AI-crawler check across thirteen named agents, and an AI-readiness audit that produces a shareable report.
Two ways to start
Talk to the group and a senior lead scopes it in writing, or go straight to the product's own site and look at it yourself. Neither route commits you to the other.
Name the number you need to hit. A senior lead replies within one business day and a costed plan follows within three working days.
The product's own page, which states its scope limits as plainly as its features — worth reading before you decide, and the route to the app and the free tools.
Related
Everything the group sells around Growzo — the company that delivers it, the nearest siblings, and the full list.