Now accepting guest editorial from SAP, Oracle, Workday and ServiceNow practitioners.Pitch a guide
The Sign-Off Table
All guides

Guide · SAP Business AI

Cloud ALM AI Requirement Generation for Fit-to-Standard

Turn Fit-to-Standard workshop transcripts into SAP Cloud ALM requirements with AI, then review, test and sign off the output. Steps, checks and pitfalls.

Matt Angerer
November 20, 2026 · 8 min read · 10 sources

SAP Cloud ALM can now draft business requirements from a Fit-to-Standard workshop transcript. You create a requirement template, paste the transcript, generate one requirement at a time, and then review every field before it is saved. The AI draft is a starting point. The consultant and the requirement approver own the content, and the output is only as good as the template and the transcript behind it.

This guide covers the working sequence, the checks that matter before approval, and what it means for the test team. It is written for the people who inherit these requirements and have to test them.

Key takeaways
  • AI requirement generation in SAP Cloud ALM works from a requirement template and a workshop transcript, one requirement topic per run.
  • The generated draft is a proposal. Review, refine and approve it before it becomes a test basis.
  • Check AI Unit entitlement before you roll it out. SAP's product pages and community posts do not describe the cost the same way.
  • Keep the transcript, template version and approver name as evidence. Re-test the feature when the model or Joule changes.

What changed from ECC-era workshop documentation

In an ECC project, fit-to-standard notes usually lived in workshop slides, Word documents and spreadsheets, and the requirement list was rebuilt by hand. Cloud ALM is the SAP-side tool for this now. SAP describes Cloud ALM as the place where fit-to-standard workshops are prepared, and where the requirements that come out of them are recorded, for S/4HANA projects on the SAP Activate roadmap. The support pages list a Fit-to-Standard Assistant among the Cloud ALM scope assistants, and SAP says its AI accelerates Fit-to-Standard workshops.

The timing matters for ECC shops. Solution Manager is the older tool, and one third-party comparison notes that Solution Manager has mainstream maintenance until 31 December 2027. If your program still runs requirements in Solution Manager, plan the move before you build AI-assisted requirements on a tool you will later migrate away from. Our SAP AI for Practitioners guide covers the wider Business AI picture if your team is new to it.

  1. Build templateDefine sections and instructions for each requirement field
  2. Load transcriptPaste the full workshop transcript into a saved document
  3. GenerateOne topic per run, with scope and solution processes set
  4. Review draftCheck facts, As-Is, To-Be and acceptance criteria against the transcript
  5. ApproveRequirement approver signs off, and the status is recorded
  6. Hand to testAcceptance criteria feed test design and the coverage map
Figure 1. The requirement generation path in SAP Cloud ALM, from template to approved requirement

Step 1: Build the requirement template first

The template is the most important step, and most teams skip it. SAP's community guidance says the template forms part of the prompt sent to the large language model, so each section needs clear instructions about what to fill in. Vague sections produce vague requirements.

For a standard template, use sections that your approvers already check: the detailed description, the current process (As-Is), the target process (To-Be), business value and acceptance criteria. The output fields shown in SAP Community examples include description, As-Is, To-Be, business value and acceptance criteria. Give each section a one-line instruction that says what good looks like, such as "acceptance criteria as observable results in the system, with a document number or status field named."

Version the template. When you change a section, the output changes, and you will need to explain the difference at audit time.

Step 2: Load the transcript and keep it complete

Paste the whole transcript, not a summary. SAP's guidance describes creating a document, pasting the transcript in its entirety and saving it for the generation process. A summary removes the exact words the business used, and those words are what you test against later.

Before you paste anything, check what the transcript contains. Workshop recordings often include employee names, customer data, pricing or bank details mentioned as examples. Strip or pseudonymize them under your data policy. Our EU AI Act and SAP Joule guide covers what to document about AI use before go-live.

Step 3: Generate one requirement per topic

Navigate to the requirements application and select Create, then Requirement, then Generate. Enter a requirement title first. Each requirement is generated individually, so the title is what the run is about.

The practical rule from SAP's own guidance is to target one requirement topic per generation. Do not generate one requirement for a whole process, and do not generate one per transcript. A run on "sales order blocking and release" gives you something you can review. A run on "order-to-cash" gives you a document nobody can approve.

Select the template, define the scope, and name the solution processes discussed in the workshop. You can also add SAP help documentation under Advanced Context. Use that where the workshop touched a standard process, so the draft can reference the standard behavior instead of the customer's memory of it.

Step 4: Review the draft line by line

The draft is a proposal. SAP's own product description says consultants can review, refine and validate the generated requirements before saving them. Treat that as a requirement, not a suggestion.

Read each section against the transcript, not against your memory of the meeting. Check four things:

  • Every As-Is statement traces to something a named participant said. Anything the model supplied without a source gets deleted or flagged.
  • The To-Be matches the standard process as the consultant understands it, including any SAP Best Practices reference. Do not accept a standard-process claim you have not checked in the system.
  • Acceptance criteria are testable. "User can release the order" is not testable. "Order moves to status released and a release document is created" is.
  • No gap has disappeared. If the transcript shows an open decision, the requirement must show it as open.

SAP's expert content for requirement management lists AI-assisted requirement generation alongside the "Requirement Approver" project role. Set that role up before the first run, so the approval step has an owner.

Step 5: Set up approval and evidence

The approval step is where sign-off happens. Keep a record for each requirement that shows the transcript version, the template version, the reviewer, the approver, the date and what the reviewer changed. The changes are the most useful evidence. If the approver changed nothing across fifty drafts, the review was probably not real.

SAP has also published a pattern where an AI model scores requirements against quality criteria such as completeness and correctness, with a policy-controlled approval workflow in SAP Build Process Automation. The SAP Community post behind it says it is a solution pattern based on SAP Consulting Best Practices and is not provided out of the box. Treat it as an option for a later phase, not something you get on activation.

Step 6: Hand the acceptance criteria to test design

The acceptance criteria become the basis for test design. SAP says its Fit-to-Standard Assistant helps define test scope, generate test cases and produce test data. Generated test cases still need a tester's review. The same rule applies as for the requirements: a generated test case that you did not check against the approved requirement is not evidence.

Map each approved requirement to at least one test case and one process step in the coverage map. Our Fit-to-Standard → Test Coverage Map template does this in a spreadsheet. For the design method, the free Generative AI for SAP Test Design course covers test cases, test data and traceability from process documentation.

Cost, activation and what to check first

Cost is where the sources disagree, so check your own contract. SAP's product page for requirement generation says AI Units are currently required to use this AI offering and points to the AI Estimator. A recent SAP Community post says Joule with SAP Cloud ALM is available at no additional cost and will not consume AI Units, while the agents added to it may have their own licensing. Requirement generation is a separate feature from the Joule assistant, so confirm which applies to your tenant in writing.

For activation, SAP's knowledge base note 3777513 is the entry point for questions on activating Cloud ALM AI features and the cost of using them. Our SAP AI Units cost estimate guide shows how to build the estimate before you buy.

What can break, and when to re-test

The generated output can change when the underlying model or Joule version changes. Treat the requirement generator as an AI feature under change control. Keep a fixed set of three or four past transcripts, run them after each release or model change, and compare the drafts. If the As-Is or acceptance criteria shift in a way the reviewer would not accept, stop and raise it before the next workshop. Our guide to testing SAP AI features after a model or Joule update describes that check.

For the people who need the broader setup, the free AI in SAP Delivery Tools course covers Cloud ALM requirement generation alongside SAP Signavio AI, and the Testing Joule Agents and AI Features course covers evaluation sets and sign-off for non-deterministic output. For prompt wording on the preparation side, the SAP Joule prompts guide covers the Activate prompt library.

Frequently asked questions

Can SAP Cloud ALM write requirements from a workshop recording?

SAP's documentation describes the feature as generating requirements from Fit-to-Standard workshop transcripts, not recordings. Transcribe the recording first, check the transcript for errors, and then paste it into the saved document. The consultant still reviews every requirement before saving.

Do I need a separate requirement template for each workshop?

No. Keep one template per requirement type, versioned, and reuse it across workshops. Change the template only through the same change control you use for test scripts, because a template change alters every draft generated after it.

Does Joule with SAP Cloud ALM replace the requirement approver?

No. SAP's own product description places review and validation with the consultant. Your approval workflow, the Requirement Approver role and the sign-off evidence stay with the project, whatever the AI does in the draft.

Is this a replacement for Solution Manager?

For cloud and RISE projects, Cloud ALM is the target tool, and Solution Manager is the older system you may still be moving off. If you run both, define which tool holds the approved requirement record before the first AI-generated requirement is saved. Otherwise you will have two versions of the truth.

Sources

  1. AI in SAP Cloud ALM, SAP Support Portal, undated
  2. SAP Cloud ALM, requirement generation, SAP, undated
  3. AI Goes CALM: AI-Assisted Requirement Generation, SAP Community, early 2026
  4. Create Requirements with AI in SAP Cloud ALM, SAP Community, recent
  5. Joule with SAP Cloud ALM available to all SAP Cloud ALM customers, SAP Community, recent
  6. SAP Cloud ALM for Implementation Expert Portal, SAP Support Portal, updated October 2026
  7. AI-based evaluation for requirement approvals in SAP Cloud ALM with external workflows in SAP BPA, SAP Community, early 2026
  8. 3777513 - Where to find information on activating SAP Cloud ALM's AI functionality?, SAP Support, undated
  9. SAP Cloud ALM: features, pricing, Solution Manager migration, Gambit Consulting, 2026
  10. Transform Your Fit-to-Standard Workshop Preparation With AI-Assisted Procedures, Saptix, 30 July 2026

More on SAP Business AI

How this guide is made: researched with AI-assisted search across SAP documentation and practitioner sources, with every fact linked to its source. Found an error or have field experience to add? Tell us.