Forcelis Platform
Your systems already record what happened. The platform is what reads them together, keeps the history comparable, and turns it into the handful of numbers a management team actually decides on.
Nothing is replaced. The platform sits above the systems you already run.
Diagram: eight data sources feed the Forcelis platform, which produces financial, operational and accounting intelligence, which in turn feeds business decisions.
These are engineering constraints, not marketing claims. Each one closes off a shortcut that would have been easier to take.
Every figure comes from a deterministic analytical engine with tests that pin known answers — four customers at 25% each gives an HHI of exactly 0.25. No number in the product is produced by a language model, because a number you cannot reproduce is a number you cannot defend in a management meeting.
Isolation is enforced at four independent layers: an automatic tenant filter in the data layer, ownership stamped on every write, authorization checked before any scope opens, and row-level security inside PostgreSQL itself. The last one matters because the first three live in application code, and application code can be bypassed.
There is no field for an e.firma or a tax portal password anywhere in the product, and there never will be. Ingestion is by files the client downloads themselves from their own systems. A credential you never hold is a credential you cannot lose.
When the information needed for a metric is not there, the platform says so and explains why, instead of returning a zero or a plausible-looking estimate. A silent zero is indistinguishable from a real result, and that is how bad decisions get made confidently.
Software companies routinely describe a roadmap in the present tense. This is the honest split: the left column is running; the right column is not something you can use yet.
CFDI ingestion
Issued and received invoices, payment complements, payroll and withholding documents, uploaded as files. Handles the mess that comes with real data: duplicates, malformed XML, unnormalizable units.
Normalization and managerial curation
Concepts and products grouped into a stable managerial catalogue, so the same item is the same item across suppliers and periods.
Five analytical engines
Price dispersion, customer concentration and churn, input cost inflation, business pulse, and forecast — each tested against scenarios with known answers.
Real cash collection
Collection measured from payment complements rather than assumed from invoice dates, with receivable and payable aging and a stated coverage percentage.
Managerial income statement
Restated results with history, comparable period over period, reconciled against the underlying documents.
Interactive dashboards
Charts, filters and drill-down, rendered by a purpose-built engine so the product works on a slow connection.
Multi-tenant with channel
An accounting firm can administer dozens of companies under one hierarchy, with its own branding on the deliverables.
Accounts, login and MFA
Working user accounts with time-based two-factor authentication, encrypted secrets and single-use recovery codes.
Deliverables
Word report, self-contained HTML dashboard and Excel export — the formats a finance team already circulates.
Anonymized sector benchmark
Compare against your sector rather than only against your own history. Built and tested with a k≥3 anonymity floor; it switches on only when there are enough companies per cell for the comparison to mean anything and to stay anonymous.
AI narrative layer
A copilot that drafts the written narrative around figures the engine has already computed. The privacy guard that constrains it is built and heavily tested; the copilot itself has not yet been run against a live model, so it is not being offered.
Direct connectors
Automatic ingestion from ERP and banking systems. The plugin contract exists; no concrete connector has been built. Today, ingestion is by file.
External economic context
Public INEGI and Banxico series alongside your own figures, to separate a company-specific movement from a market-wide one.
Second country
The core is already country-agnostic — base currency per client, country on the data model. What is missing is the tax document parser for the next country.
The most useful demonstration is not a demo account. It is a diagnostic run on your data.