Frequently asked questions

You paste SAP ABAP source — or a payroll schema (PE01) or rule/cycle (PE02) — choose what to analyse, and it produces structured documentation of what the code does, grounded in the source itself. Every result is an AI-generated first draft for a qualified person to review, not a finished or certified document.

Your code is sent to a large language model (Anthropic Claude) with a carefully constrained prompt. The model documents the program section by section, separating what is observed directly in the code from what is inferred, cites evidence with line numbers, and lists the identifiers it touches. You can export the result to PDF.

Technical mode focuses on program purpose, execution and data flow, database operations, performance, dependencies, authorizations, and technical risks.

Payroll mode focuses on SAP HCM/Payroll relevance: payroll inputs/outputs, employee selection, retroactive-accounting exposure, external interfaces carrying payroll data, controls, and audit notes. The mode changes the actual analysis, not just the display — pick it before you analyse.

Yes. Switch the input to Payroll schema / rules and paste a schema (PE01) or a personnel calculation rule / cycle (PE02). It detects whether the input is a schema, a rule, or a schema printed together with its rules, then documents the processing sequence, the wage types read and produced, the calculations and decision logic, and an inventory of the functions, operations, subschemas and rules it uses.

Standard SAP functions and operations are explained from their documented behaviour. Custom or site-specific functions and operations (typically Z-prefixed) are listed and flagged separately — their actual logic can only be confirmed from the underlying ABAP or configuration, so the tool says so rather than guessing.

Tip for best results: run report RPDASC00 (Schema Documentation) in your SAP system for the schema you want and paste its output. It expands the schema with its subschemas and rules into a single, consistent listing — the cleanest and most complete input you can give it.

Each analysis includes:

• A classification (payroll-relevant or no direct payroll impact detected) and an overall risk level.
• Sections (purpose, flow, inputs/outputs, dependencies, controls, risks, review points) — each with Observed / Inferred points, a risk level, a confidence score, and evidence.
• A simple flow diagram of the execution/data flow where one can be determined.
• A code inventory — tables, infotypes, wage-type literals, function modules, RFC destinations, hardcoded values, etc. (listed as written; business meaning is not interpreted).
• Limitations — what could not be verified from the source you provided (missing includes, dynamic calls, config).

The default is a fast, capable model suitable for most programs. Ticking Deep analysis uses a more thorough (slower) model for complex code. Deep analyses take longer and cost 2× the credits.

Up to about 200,000 characters per submission. If a program is larger, analyse the main program and its key includes separately, or contact us — we’re interested in supporting larger programs.

Analyses are priced in credits: 1 credit per ~40,000 characters (rounded up, minimum 1), and 2× for Deep analysis. Most programs cost 1 credit.

The Free plan includes 3 free credits every month. Monthly plans refresh a larger allowance:

• Solo — $29/month, 60 credits/month.
• Pro — $79/month, 200 credits/month.
• Team — $249/month, 1000 credits/month.

You can also buy one-off credit packs that never expire:

• Starter — 50 credits for $25.
• Standard — 150 credits for $49.
• Bulk — 750 credits for $199.

Free monthly credits top up to your plan’s floor each month; purchased credits roll over. Cancel a subscription anytime from the billing page; access continues until the end of the period.

Sign up with an email and password. By default your pasted code is not stored — it’s used to produce the analysis and not kept. If you tick “Save this code to my history”, the code is stored so you can re-open, re-run, and export it; you can delete any saved analysis (and its code) at any time from your history.

Treat every result as a draft. AI output can be incomplete or wrong, and risk/confidence scores are estimates, not guarantees or certifications. Always have a qualified SAP professional verify the output before relying on it. See our AI Disclaimer and Privacy Policy.