Flat fee. Tokens at cost.
Most tools bundle the model cost into the price, then cap your prompts. We charge for the software. You run the models on your own key.
Pay for the year and save 20%.
Starter
One brand, one person watching it.
Billed $228 yearly. Save $120 a year.
See where you stand on all four assistants.
- 1 project
- 1 seat
- Unlimited tracked prompts
- All four: ChatGPT, Claude, Gemini, Perplexity
- Read-only MCP, so your agent can read the data
- Daily, weekly or monthly runs
Pro
In-house teams who act on what they find.
Billed $828 yearly. Save $240 a year.
Two-way MCP: your agent writes the work back, not just reads it.
- Up to 5 projects
- 3 seats
- Unlimited tracked prompts
- All four: ChatGPT, Claude, Gemini, Perplexity
- Two-way MCP into your IDE or CMS
- Retrieval scoring on your own pages
Agency
Agencies carrying a roster of clients.
Billed $2,868 yearly. Save $720 a year.
A separate model key per client, so the run cost lands on their bill.
- Up to 25 projects
- 10 seats
- Unlimited tracked prompts
- All four: ChatGPT, Claude, Gemini, Perplexity
- Per-client keys, billed to the client
Enterprise
For teams with a security review to get through.
Everything above, plus the paperwork your security team needs.
- Unlimited projects and seats
- SSO and SAML
- Dedicated model routing
- Custom pipelines and data export
- A DPA, a security review and a human to call
Prices in USD. These are software fees only. The next section works out what the model calls add on top.
What each plan does
Open a row if a term is new. They are explained in plain English, not sold.
The questions we ask the assistants on your behalf, on a schedule. You pay the model cost, so we do not cap them.
Every prompt runs on ChatGPT, Claude, Gemini and Perplexity, on every plan. Same question, same parsing, so the four can be compared.
Daily for prompts you are working on, weekly for the rest. Changing a cadence changes your model spend the same day.
Assistants answer differently each time they are asked. More samples per run narrows the confidence band on your mention rate. You run on your own key, so you choose how many.
A prompt you are deliberately not optimising, tracked alongside the rest, so you can tell a real move from a platform-wide shift.
Whether the assistant put you forward as a choice, or just named you in passing. Only one of them sends you a buyer.
The searches an assistant runs on its way to an answer. They are rarely your keywords, and they are what you have to be findable for.
Every URL the assistant pulled, and which of them it actually quoted. Shows you the pages standing between you and the answer.
We crawl your pages and score whether the answer chunk clears what retrievers measure, then say what to change.
Pausing or cancelling stops new runs. Everything already measured stays, so the trend line survives a gap.
Connect Claude, Codex or any MCP client to your project. The assistant reads your live position instead of you pasting numbers in.
Two-way MCP lets the agent log an optimisation, set a page as being worked, and pick up where it left off next session.
The AI SEO method itself, served to your agent alongside the data: the processes and skills it should follow.
You add an OpenRouter or OpenAI key and pay that provider directly at list price. We never resell or mark up model calls.
Each client project runs on its own key, so a client's tracking lands on the client's bill rather than yours.
What each prompt, assistant and run cost, so you can see where the bill goes before you change a cadence.
One project per brand or client. Each has its own prompts, competitors, domain and history.
People who can sign in to the account. Seats are free within your plan's number.
Give a teammate access to some projects and not others, and decide who can change billing.
A record of admin actions on the account: team changes, and project pause, cancel and resume.
Sign in through your own identity provider, with access removed centrally when someone leaves.
A signed data processing agreement, and someone to answer your security questionnaire.
The difference a two-way connection makes
Every plan connects your agent to Omnipresence over MCP. What changes is whether the agent can only look, or can also do the work and record it.
Your agent can see
Ask it where you stand and it answers from live data instead of guessing. It reads your mention rate, the fan-out searches, the sources each assistant cited and how you track against competitors.
- Where you stand this week, and whether it moved
- The searches behind each answer, and the pages cited
- How you compare with the competitors you name
You still do the work. The agent tells you what to fix, and you go and fix it somewhere else.
Your agent can act, and leave a record
It adds the prompt it thinks you are missing, marks a page as being worked on, fires a run to check whether the rewrite moved anything, and logs what it changed and when.
Everything above, plus
- Add the prompt you are missing
- Mark a page as being worked on
- Run a check after a rewrite, to see if it moved
- Log what it changed, and when
The record is the part people miss. Because the agent writes what it did back, next week it opens on your project and already knows what it tried, what moved and what it was halfway through.
In practice that is the difference between pasting numbers into a chat window every Monday and saying "carry on where you left off".
What a month really costs
The software fee is half of it. Here is the other half, at the rates the app itself uses.
- Pro platform fee
- $69.00
- Model calls, paid to your provider
- $51.96
- Total, all in
- $121/mo
Model spend is an estimate at $0.12 per prompt per run, across all four assistants (ChatGPT, Claude, Gemini, Perplexity). You pay that to your own provider at their list price. We never touch it, mark it up, or resell it.
Run fewer prompts, or run them less often, and the bill drops the same day.
What your own key gets you
Adding a key takes a couple of minutes. Here is what changes once the compute is yours.
No cap on prompts
A tool that pays for your model calls has to limit them. That is why entry plans track fifty prompts on one assistant. On your own key there is no cap, so you can track the long tail as well as the head terms.
You see the meter
Spend shows up per prompt, per assistant, per run, in your provider's dashboard and in ours. Turn a cadence down and the bill drops the same day. No credits, no overage invoice.
Clients pay for their own runs
On Agency, each project takes its own key. A client's tracking runs on the client's account, so taking on another one does not add to your costs.
Before you hand over a key
Because every prompt costs money to run, and someone has to pay it. A tool that bundles that cost has to guess high and cap your prompts, or it loses money on its heaviest users. We would rather charge for the software and let you buy the model calls at list price. It works out cheaper, and you can see what you are spending.
The scheduling, the parsing and the judgement. We run the probes on a cadence, read what each assistant said, work out whether you were recommended or only mentioned, track which sources got cited, score your pages, and hand all of it to your agent over MCP. The model call itself is a few cents of that.
Keys are encrypted at rest with AES-256 and only used to run your own scheduled prompts. They are never logged, never sent anywhere else, and you can replace or remove one at any time. The security page has the detail.
Yes, on Agency. Each project takes its own key, so a client's tracking lands on the client's bill rather than yours. Your cost stays the flat fee however many prompts you run for them.
Seven days on Pro, no card. You get the two-way MCP and five projects for the week whichever plan you end up on, because the way to judge this is to point it at a real brand and look at what comes back. Agency trials ask for a card up front.
Your data stays. Projects pause instead of deleting, runs stop, and everything you measured is there when you come back. We have never hard-deleted a member's history and do not plan to start.
Still deciding? Point it at one brand for a week and look at what the assistants say about it. Seven days, no card.
Start free for 7 daysStart with one brand and a week.
Seven days, no card, on the plan with everything switched on. Point it at a brand you care about and see what the assistants are telling buyers.
Start free for 7 days