The questions people actually ask before a pilot.

What is Superiora Planner?

A planning platform with a decision layer above the model. You model the choices a business actually makes — with an owner, a date and a reason — and the operating plan is derived underneath them, so every figure resolves to a formula, its operands and the cells they read. Underneath it is our own calculation engine.

Is it a spreadsheet replacement, or a BI tool?

Neither. A spreadsheet has no shared model, no access control and no history, and a BI tool reports what happened rather than letting you plan what will. Superiora is a multidimensional planning model — dimensions, cubes, measures, formulas — with a governed write path. Excel still gets used for the last mile, and the platform imports from it and exports to it on purpose.

How does it compare with Anaplan?

The modelling semantics will be familiar to anyone who has built in Anaplan — hierarchies that aggregate, measures that are inputs or formulas, a fiscal calendar, dimension alignment in formulas — and the engine has no feeders: it works out the dependency graph itself and refuses an illegal one by name at save time. What is different is the layer above: the decisions are modelled objects rather than context in a deck, and the whole model is exposed to AI agents as first-class users. The names of other products on this page are their owners’ trademarks; see brand use.

What is it not good for?

Transaction processing, general ledger accounting, and reporting that is really a data-warehouse query. It also will not make a decision for you. If your problem is that nobody has decided, a planning platform is not the missing piece — that is a conversation, and we would rather have it than sell you software.

How does data get in?

CSV, Excel, or straight out of a Postgres database. An import is defined once against the model’s own lines — the mapping is asked target-first, so a renamed column in the file does not silently land in the wrong place — and it remembers how your file is written, including the decimal mark and which row the data starts on. It can then be re-run by hand or on a schedule.

Can I get my data out?

Yes, whole and at any time: Excel, CSV, and a query interface that BI tools read. A board pack exports as a printable document and as a workbook. We do not treat your data as a reason you cannot leave.

Where is it hosted?

Amazon Web Services, us-east-1 (Northern Virginia, United States). Backups are nightly, to object storage in the same region. There is no other region today; if data residency in Brazil or the EEA is a requirement, say so before the pilot rather than during it. Subprocessors has the full list.

Who owns the model I build?

You do. Your data, your model, your formulas, your boards. We claim no ownership and acquire no licence beyond what running the service needs. The platform itself stays ours. The terms say it in the language that binds.

Do you have SOC 2 or ISO 27001?

No. We hold no third-party security certification and there is no independent penetration test report. Security sets out what is actually in place — Argon2id passwords, signed sessions with lockout, role plus dimensional access enforced on the read side, period locks, an append-only audit log, tenant isolation, TLS, nightly backups — and, in the same length, what is not.

Can I stop someone seeing part of the model?

Yes, down to the slice: a region, an entity, a set of cost centres. Grants are explicit and are enforced on the read side, so a cell somebody may not see is never offered to them for writing. The order is role, then period lock, then selective access, then dynamic cell access — most restrictive wins.

How do I know a closed month stays closed?

Because the engine refuses it, not a convention. A locked period refuses every write into it — typed, pasted, imported, scheduled, and the platform’s own — and the refusal names the cell and the reason. Closing, reopening and reverting are recorded acts with an author and a time.

Does it support single sign-on?

Not yet. No SAML, no OIDC, and no enforced multi-factor authentication. It is a known gap and it is on security in the list of gaps rather than buried in a roadmap.

What does an AI agent actually do here?

It builds and changes the model, sets up scenarios and baselines, and runs the routine — the imports, the notifications, the pack. It does that through a real interface: read the model and its map, walk the semantic graph, trace what a formula depends on, find everything that references a thing, compute what a dimension change would reach, check a change, apply it, and record why. It is reading the model’s own structure, not guessing from a description of it.

What stops it doing something stupid?

The same four gates that stop a person. An agent is a user with a role; a closed period refuses it; an access grant refuses it; its writes land in the audit log under its own identity. A change can be staged and its impact read before it is applied. And the decision itself is not delegated — the platform shows what each version of it costs, and a person signs it.

Is my data sent to an AI provider?

Not by the platform. No customer data is sent to a third-party model API by Superiora Planner, and forecasting runs in a sidecar on the same host. Where an agent operates on your model, that agent is run by you or by us as an operator against the platform’s own authenticated interface — so the model provider is the operator’s choice and their responsibility. We do not use your data to train anything.

Do I have to use the AI part?

No. Everything an agent does can be done by a person in Studio, and plenty of teams will want it that way to begin with. The interface is there because building a planning model by hand is slow, not because a model built by hand is worse.

What does it cost?

Talk to us. There is no price list on this site and no self-service sign-up, because the number depends on the planning scope and the number of seats, and quoting one before we know either would be a guess. Pricing explains how we arrive at it.

What does a pilot involve?

A real planning process, stood up on your own data, with a decision map first and a model second. It runs against 0.3.0, the pilot release, with the founding team doing the build and the training alongside you. The point is to find out whether the method holds in your business — not to produce a demo.

Which languages does it work in?

This site is published in English, Portuguese, Spanish, Italian and French, and the practice works in all five. The application’s own interface is in English; the terms of art — S&OP, FP&A, plan of record — stay in English everywhere, because that is what your own tools call them.

Is the software open source?

No. Superiora Planner is proprietary. It is built on open-source components, and every one of them is named with its licence on third-party notices.

A question that is not here: hello@superiora.ai. If it comes up twice, it ends up on this page.