Priced per agency. Every engine included. In rand.
Four tiers, sized by how many client brands you run rather than by which engines you are allowed to see. Annual billing takes roughly two months off. There is no free tier, because there is no self-serve tenancy to land in — accounts are provisioned.
Studio
A small agency proving the line item
R3,250/mo billed annually
- Up to 5 client brands
- 50 prompts per brand
- Weekly
- All six engines
- White-label reports — your logo, colour and domain
- God-view across every client
- Raw-answer click-through on every figure
- Owned-domain sets, aliases, competitor sets
- English and Afrikaans prompts
- CSV and branded PDF export
- 3 seats
Growth
Most agenciesThe working tier for a full client book
R9,080/mo billed annually
- Up to 20 client brands
- 100 prompts per brand
- Weekly, with daily on flagged prompts
- Everything in Studio
- Daily cadence on up to 10 flagged prompts per brand
- Alerts — movement, stability change, citation loss, coverage failure
- Scheduled email reports, weekly or monthly
- AI-readiness diagnostics
- Proposal brands for pitching
- Model-event annotations and optional control brands
- 10 seats
Scale
A large book, or an agency reselling at volume
R20,750/mo billed annually
- Up to 50 client brands
- 150 prompts per brand
- Daily available on every prompt
- Everything in Growth
- Daily cadence, unrestricted
- Read API with tenant-scoped tokens
- Looker Studio connector and alert webhooks as they ship
- Custom report schedules
- Priority support, SAST business hours
- Unlimited seats
Custom
More than fifty brands, or a market we have not built
- 50+ brands
- Custom
- Custom
- Everything in Scale
- A new market configuration, built and validated with you
- AI-crawler analytics from your clients' logs or CDN
- Bespoke prompt-set design engagements
- Data-processing terms negotiated rather than accepted
Prices exclude VAT. Overage is priced per additional brand and per block of 25 prompts, quoted at signature so it never arrives as a surprise.
Bundled in every tier. 1 of 6 adapters are sourcing answers today (as of 2026-08-27) — the rest are in build, and we publish which.
A brand
One client, one competitive landscape: their owned-domain set, their aliases, their declared competitors and their prompt set. A client with two genuinely separate product lines is usually two brands, and we will tell you which we think it is rather than billing you for the ambiguity.
A prompt
One tracked question, run across every engine in the market at three samples per cycle. A prompt group counts as one prompt up to three phrasings — because breadth is what makes the measurement better and pricing it per phrasing would punish the thing we are asking you to do.
A seat
A person in your agency with a login. Client-side viewers do not consume seats — the client-facing report is a page you share, not an account you provision.
Fair use
The brand and prompt allocations exist because a cycle has a real per-run cost. They are set well above normal use, and if you are near a limit we would rather move you a tier than throttle you quietly mid-month.
What it actually costs us to run a brand.
Published because you are pricing on top of it, and because “all engines included” is only credible if the numbers behind it are. These are measured from live runs, not modelled.
- Cadence is the big lever. Daily costs roughly seven times weekly and moves a slow metric barely at all — which is why weekly is the default and daily is a flagged premium.
- About 70% of the cost is input tokens, and those tokens are search results rather than our prompts. Prompt caching cannot fire on them at all.
- Below a certain model size the web-search fee becomes the floor — 42–59% of the bill. There is no optimising past a fee that does not scale with the model.
Measured, 210 live runs
| Configuration | Per run | Per brand-year |
|---|---|---|
| As first built — large model, synchronous | $0.159 | ≈ $7,420 |
| Batch submission + mid-tier analysis model | $0.0337 | ≈ $1,580 |
| The same, fortnightly | $0.0337 | ≈ $790 |
| 20 prompts instead of 50, weekly | $0.0337 | ≈ $632 |
The 79% reduction changed nothing about the output: mention rate was 30/30 against 30/30 across the bake-off. Every per-brand figure assumes 50 prompts across six engines at three samples weekly.
The ones worth asking.
Why is there no free tier?
Because there is nothing for a self-serve account to land in. A tenancy is an agency with branding, client brands and provisioned users; there is no single-user workspace and no plan to build one. A demo on your real category is the free thing we offer instead.
Do engines cost extra?
No. Six engines in every tier, including Custom. If we ever add a seventh, it will be included too — the per-engine add-on is the structure we are deliberately not shipping. That is a pricing commitment; which adapters are sourcing answers today is a separate and more perishable fact, and we publish it on the engines page.
Can I cancel?
Monthly plans cancel at the end of the period, no notice required. Annual plans run their term. On termination you can export everything, and you can require erasure of the stored captures — both are contractual, not a favour.
What happens if I go over?
You are told, and the cycle still runs. Overage is billed at the rate quoted at signature, or we move you a tier if that is cheaper — which we will say, because a client who discovers a better price from a competitor is an expensive way to learn it.
Is my clients' data used for training?
No. Prompts, brand configuration, captured answers and results are never used to train models — ours or anyone else’s. See the trust page for the sub-processor list and where data physically sits.
Why rand?
Because you invoice your clients in rand. USD pricing on a cost-of-goods base that is also USD is a currency risk we can carry far more sensibly than you can, and passing it to you would be pricing our convenience.
These tiers are pre-launch and being validated with South African agency partners. If the shape is wrong for how your book actually works, that is a conversation we want to have now rather than after it hardens.
See it against your own client list
A working demo runs your prompts, in your market, on live engines — not a sandbox with seeded data. Bring one client brand and three competitors.