SEOBOT ↗: an autonomous SEO AI agent
- MRR
- $34,936 Verified 2026-10-22
- Total revenue
- $1,910,331
- Active subscriptions
- 496
Verified by TrustMRR
In one line: An AI agent that runs SEO on autopilot, sold as a monthly subscription.
- Founder
- John RushX ↗
Timeline
- Founder says he started SEObot
- Launches on Product Hunt asking for support and feedback
- Revenue that month: $99,433 (TrustMRR)
- Revenue that month: $63,803 (TrustMRR)
Compared with similar products
Moves SEOBOT made
No public information yet
How to copy it
Build for yourself first: automate the one task you do every day, end to end
Build v1 for your own needs: pick a task you keep doing on your own projects and would hire someone for, have an AI agent do it end to end with no confirmation by default, and log each run's result and API cost. What the founder did: he builds tools for himself first and argues you should become a fan of your own tool (X post by @johnrush), designing the product to be turned on and forgotten (X post by @johnrush).Expand ▾Collapse ▴
Build v1 for your own needs: pick a task you keep doing on your own projects and would hire someone for, have an AI agent do it end to end with no confirmation by default, and log each run's result and API cost. What the founder did: he builds tools for himself first and argues you should become a fan of your own tool (X post by @johnrush ↗), designing the product to be turned on and forgotten (X post by @johnrush ↗).
- One run on one of your own projects: the agent completes a full cycle and the result is live and publicly reachable.
- The agent's output is indexed or publicly reachable at its destination.
Write me a fully automatic AI agent for my own projects that does <the task you do every day>: 1) take my project URL or files and read the existing content; 2) call an LLM API to produce a work plan; 3) run the plan on a schedule and publish output directly to my <destination>; 4) log what it did, the result URL and API cost each time. Read secrets from environment variables; follow official docs and flag anything uncertain for me to check.
Step 1: produce the 10-keyword content plan from the site's content and write one article a week.
Step 1: publish articles straight to the site through the posts endpoint with an application password.
Run it on all your own projects for a few weeks and record output and change
Turn the step 1 agent on for every project you own, don't edit by hand for a few weeks, and record each project's key numbers for equal periods before and after, plus what the agent's own output brought. What the founder did: his method is to build a B2B tool for yourself, be your own customer, then share publicly how you use it (X post by @johnrush).Expand ▾Collapse ▴
Turn the step 1 agent on for every project you own, don't edit by hand for a few weeks, and record each project's key numbers for equal periods before and after, plus what the agent's own output brought. What the founder did: his method is to build a B2B tool for yourself, be your own customer, then share publicly how you use it (X post by @johnrush ↗).
- A table with one row per project: output count, key numbers before and after, and what the agent's output alone brought; plus a screenshot of the best project's data.
- The data shows the agent's own output producing results, not just old content rising.
Below is data from running my own fully automatic agent on <N> of my projects for a few weeks: each project's output URLs, key numbers for equal periods before and after, and numbers broken out by output. Please: 1) compute each project's change and separately the agent output's contribution; 2) compare the best and worst projects; 3) tell me which number best shows the agent worked when I share it, and which not to use (e.g. ones already rising before); 4) use only my data. Data: <paste>
Step 2: in the performance report, compare impressions and clicks for the 4 weeks before and after, and check impressions on the new articles by page.
List the product on directories and post publicly with your own project's real data and a waitlist link
List the product on directories your target users browse, then post where your target users gather: a screenshot of the step 2 project's data, how much was done, how the number moved and how much of your own time it took, ending with an email-only waitlist link. What the founder did: he tested on his own products first, then listed the product on directories and built a waitlist within weeks (X post by @johnrush).Expand ▾Collapse ▴
List the product on directories your target users browse, then post where your target users gather: a screenshot of the step 2 project's data, how much was done, how the number moved and how much of your own time it took, ending with an email-only waitlist link. What the founder did: he tested on his own products first, then listed the product on directories and built a waitlist within weeks (X post by @johnrush ↗).
- The directory listings are submitted, the post is live, and the screenshot's numbers match the step 2 table; the waitlist link opens and accepts emails.
- The waitlist collects emails from strangers.
I built an agent that does <the task> fully automatically for <your target users>, and ran it on my own projects for a few weeks. Data: <project>, <x> pieces done, key number from <a> to <b>, my time spent <minutes>. Write a short English post to go with my screenshot: 1) the first line is the change in the number; 2) one line on what the agent did for me; 3) a last line asking people who want it to join the waitlist at <waitlist link>; 4) only my numbers, no guaranteed results. Also write a 4-part follow-up thread on why I built it for myself first.
Onboard a small batch from the waitlist one by one, with a free diagnosis
Don't open signups all at once: pick a small batch of the best-fit people, ask them to grant access to the data you need, onboard each by hand and connect the agent, and send a free diagnosis: where they stand, what the agent will do first and why. What the founder did: early users were onboarded one by one so outputs could be moderated and servers wouldn't break (X post by @johnrush).Expand ▾Collapse ▴
Don't open signups all at once: pick a small batch of the best-fit people, ask them to grant access to the data you need, onboard each by hand and connect the agent, and send a free diagnosis: where they stand, what the agent will do first and why. What the founder did: early users were onboarded one by one so outputs could be moderated and servers wouldn't break (X post by @johnrush ↗).
- The batch is onboarded, each person has a diagnosis, and each first output has published automatically.
- Someone replies with feedback after their first output publishes.
I'm onboarding early users of a fully automatic agent one by one, with a free diagnosis. Below are a user's project URL, existing content list, recent data they shared and the agent's work plan for them. Write them an English diagnosis under 400 words: 1) where they're strongest now; 2) what the agent will do first and why; 3) when the first output goes out and the cadence after; 4) only my data, no promised results. Material: <paste>
Step 4: once the user adds you as a restricted user, review their last 3 months of queries for the free audit.
Send agencies that do this for clients a white-label resale offer
Find agencies that do for clients what your product does and DM each one: they offer your product to clients under their own brand and you share revenue, with your step 2 project data attached. You set the split. What the founder did: he knew people earning more than him by reselling his product as an agency (X post by @johnrush), and publicly offered white-labeling with shared revenue (X post by @johnrush).Expand ▾Collapse ▴
Find agencies that do for clients what your product does and DM each one: they offer your product to clients under their own brand and you share revenue, with your step 2 project data attached. You set the split. What the founder did: he knew people earning more than him by reselling his product as an agency (X post by @johnrush ↗), and publicly offered white-labeling with shared revenue (X post by @johnrush ↗).
- A list of agencies (handle, site, services) and the DMs sent.
- An agency turns on the product for one of its clients.
I built an agent that does <the task> fully automatically for <your target users> and want agencies to white-label and resell it. Below are agency profiles: handle, site, service list, latest post, plus my own project data. Write each an English DM under 120 words: 1) open with a service of theirs related to this task; 2) say they can offer this product to clients under their brand while I share revenue at <my split>; 3) attach my own project data; 4) promise no results. Profiles: <paste>
Create a monthly subscription payment link in Stripe with a full refund if the first output disappoints
On Stripe's Payment Links page, create a monthly subscription link with no trial and send it to the step 4 users and step 5 agencies that replied. First confirm with the step 1 per-run cost that the price covers a month of costs, rather than discounting, and state a full refund if the first output doesn't satisfy. What the founder did: his margins are too thin for discounts (X post by @johnrush), and the site offers a full refund after the first article if you're not satisfied (seobotai.com).Expand ▾Collapse ▴
On Stripe's Payment Links page, create a monthly subscription link with no trial and send it to the step 4 users and step 5 agencies that replied. First confirm with the step 1 per-run cost that the price covers a month of costs, rather than discounting, and state a full refund if the first output doesn't satisfy. What the founder did: his margins are too thin for discounts (X post by @johnrush ↗), and the site offers a full refund after the first article if you're not satisfied (seobotai.com ↗).
- The payment link opens with no trial; its description states the first-output refund; the "pieces per month x cost per run < price" math is written down.
- Pay once in Stripe's test environment: it bills monthly with no trial.
My fully automatic agent's first plan is $<price>/month, no free trial, full refund if the first output disappoints. Step 1 measured an API cost of $<x> per run, with <n> runs a month by default. Please: 1) compute "runs per month x cost per run" against the price and, if margin is negative, how many runs to cut; 2) write the Stripe product name and a one-line description; 3) write one English refund line: contact us after the first output if not satisfied for a full refund; 4) use only my numbers.
Step 6: on the Payment Links page, create a $49/month subscription link with no trial; TrustMRR shows SEObot's revenue verified through the Stripe API.
Get your first revenue: the first payment lands and the user's first output is live
Confirm the first payment in your Stripe payments list, then confirm the agent's first output for this user is live. After it goes out, ask whether they're satisfied and refund per your step 6 promise if not. What the founder did: his Stripe-verified revenue is public on TrustMRR (TrustMRR).Expand ▾Collapse ▴
Confirm the first payment in your Stripe payments list, then confirm the agent's first output for this user is live. After it goes out, ask whether they're satisfied and refund per your step 6 promise if not. What the founder did: his Stripe-verified revenue is public on TrustMRR (TrustMRR ↗).
- Stripe shows the first real subscription payment (not test mode) and the user has the agent's first published output.
- The user doesn't request a refund after the first output and renews for a second month.
My fully automatic agent just got its first $<price>/month customer. Below are their project, the first output's URL and the plan for the next few weeks. Write an English email under 120 words: 1) link the first output and say why it came first; 2) list what happens each week next; 3) per my refund rule, ask if they're satisfied and say a reply gets a full refund if not; 4) promise no results. Material: <paste>
Step 7: check the first real $49 subscription payment in the payments list; TrustMRR shows SEObot's revenue verified through the Stripe API.
Sources
Next case
ChatSEO: conversational AI SEO agent
Business model: B2B SaaS subscription (four tiers priced in euros: Starter €29 / Pro €49 / Ranker €79 per month, Agency "Custom"; paid plans start with a trial of "7 days + 30 credits", trial added 2026-09-07; for site owners and SEO consultants)
Data source: TrustMRR, checked 2026-10-18