Perasight is an application platform, not a reporting tool. Here is what that means, feature by feature — with enough detail for your architect and your CFO.
Your team describes what the app should do. Perasight compiles it against your live SAP schema and runs it inside your instance — reasoning in context, not against a stale extract.
Inbound messages are classified by tone and urgency, joined to customer order history, and replied to or routed. Live example in our sandbox.
Every posted invoice is scored against GL history. Outliers land in the controller's queue with reasoning attached.
AR ageing + seasonality + open orders → 13-week rolling cash forecast, updated nightly.
A new app is a prompt, a schema binding, and a trigger. No SAP transport, no data engineer required.
A pair of hands for the analyst in your finance team. Ask a question in plain German, English, or Turkish; see the SQL, the chart, and the interpretation.
# User prompt Show me DSO by region, last 4 quarters, excluding intercompany invoices. # Generated SQL · editable before run SELECT region, AVG(DATEDIFF(paid_at, billed_at)) AS dso, FLOOR_QUARTER(billed_at) AS q FROM fi.invoices WHERE intercompany = FALSE AND billed_at >= NOW() - INTERVAL '1 year' GROUP BY region, q; # Output • Bar chart by region × quarter • Interpretation: DE up 6 days, NL flat, TR -3 days (explained below) • [ save as dashboard ] [ promote to app ]
Studio sees table shapes, relationships, and masked samples — not data contents. The model never sees raw rows.
Generated queries are shown, editable, and versioned. No black box.
An answer becomes a pinned chart, a scheduled digest, or the first step of an autonomous app.
SD and FI ship pre-wired, so value lands on day one while your team builds custom apps on top. No SAP consultant required to light them up.
Pipeline, pipeline velocity, DSO, on-time delivery, returns rate, top accounts by margin.
AR/AP ageing, cash position, GL journal anomalies, intercompany reconciliation flags.
Inventory ageing, supplier lead-time drift, shortage-risk scoring.
Cost-center drift, internal allocations, margin decomposition by product line.
When an off-the-shelf app isn't enough, drop a Python workload into the runner. It executes against your data with the same security posture — and its outputs plug back into dashboards or trigger apps.
Isolated Python environments per customer. No egress to untrusted destinations.
Call Claude, a local model, or an in-house PyTorch forecast — the runtime is indifferent.
A runner result is a table, a chart, or a webhook — cleanly composed into Perasight's UI and into other apps.
Multi-tenant is a compliance liability and we refuse it. Every customer gets their own instance, their own database project, their own keys.
Not a schema in our database — a separate, customer-owned project with its own boundary.
Frankfurt or Amsterdam regions. No US replication. GDPR-native.
We bind to your replica, or to a read-only service user. We cannot write back.
Model calls opt out of provider training. Your data is never a dataset.
SAML / OIDC login. SCIM provisioning. Role-based access mirrors your SAP org.
Every query, every app run, every export — appended, signed, exportable.
Direct connector to ECC 6, batch & CDC.
HDB replication & CDS views.
Live queries via standard SAP interfaces.
OAuth-scoped connector, multi-client.
Snowflake, BigQuery, Databricks, MotherDuck.
CRM pipeline against SAP orders.
Shop-floor streams into runner.
A 12-page PDF with the runtime diagram, security boundary, and a sample SD dashboard. We send it after a first call.
Request the deck