AI Labs · 9 lab tools

Nine specialist SAP AI tools.

The lab tools that round out the A²AI pipeline alongside the six Core Products. Each opens into its own working MVP page with client-side demo output. For the unified view of all fifteen tools, see Full Portfolio.

Pre-sales

Lab Tool 01

Document Intelligence

Ingests RFPs, AS-IS docs and process flows in any format. Structured extraction with source-linked summaries.

OCR + layout-aware parsing + LLM extraction

Open lab demo

Pre-sales

Lab Tool 02

Requirements Extraction

Auto-builds a requirements register with Functional / Non-Functional split, MoSCoW priority and SAP module tagging.

Sequence labelling + rule-based MoSCoW classifier

Open lab demo

Pre-sales

Lab Tool 03

SAP Module Router

Semantic mapping of every requirement to the right SAP modules, sub-modules and licence implications.

Bi-encoder semantic search over SAP module catalog

Open lab demo

Pre-sales

Lab Tool 05

Project Risk Estimator

A 0–100 delivery risk score with SHAP attribution — exactly which factors are pushing risk up.

Gradient-boosted classifier + SHAP explanations

Open lab demo

Pre-sales

Lab Tool 06

RFP Chat

Natural-language Q&A over every uploaded document, with inline citations back to the source page.

Retrieval-augmented generation with citation grounding

Open lab demo

Classification

Lab Tool 08

RICEFW Classifier

Classifies every custom development as Report, Interface, Conversion, Enhancement, Form or Workflow.

XGBoost over TF-IDF + 14 hand-crafted features + naming-rule overrides

Open lab demo

Design

Lab Tool 11

Compliance Matcher

Auto-maps requirements to 20+ frameworks across 40+ countries — with the SAP configuration that satisfies each control.

Semantic search + control-to-config knowledge base

Open lab demo

Design

Lab Tool 12

Integration Pattern Advisor

Recommends sync, async, event or batch — with the matching BTP/CPI/API technology and reuse opportunities.

Decision tree over latency, volume, durability constraints

Open lab demo

Delivery

Lab Tool 13

Portfolio Intelligence

Roll-up of scope, cost, risk and compliance posture across every active engagement.

Engagement aggregation + drift detection

Open lab demo

Practitioner Corpus · New

5,400 SAP consultant tasks, structured.

The Practitioner Task Atlas captures the real day-to-day work of senior SAP consultants — across 18 modules and 15 activity classes (Discovery, Legacy T-code Mapping, Master Data, Configuration, Execution, Cutover, Hypercare, Cleansing, Reporting, AI Hand-off and more). Every A²AI recommendation is grounded against this corpus.

5,400

Tasks

Modules

359

T-codes

Open the Task Atlas

Algorithm flowcharts

How the six flagship tools actually run.

Each pipeline is deterministic where it can be, generative only where it must be, and grounded in real SAP engagement data end-to-end.

Flowchart · 01

Landscape Intelligence Pack

Filename detection + regex extraction + ontology aggregation + L1–L5 evidence

Open tool

Step 01

Intake

Documents (BRD/TRD/FSD/CR/RICEFW), screenshots, free text — all in-browser

Step 02

Artefact typing

Filename rules classify each upload into 12 SAP artefact families

Step 03

Regex extraction

T-codes · BAPIs · FMs · tables · Z/Y customs · ZZ_/YY_ fields · roles · formulae

Step 04

Ontology merge

Per-artefact entities aggregated into a client-specific knowledge graph

Step 05

Evidence scoring

L1 user-described → L5 SME-validated · per-artefact toggle

Step 06

Output

JSON manifest · ontology view · downstream tools inherit the context

Flowchart · 02

Cost & Timeline Forecaster

Quantile LightGBM + Monte Carlo (Gaussian copula)

Open tool

Step 01

Inputs

Scope, modules, geos, RICEFW counts, integration map, team mix

Step 02

Feature engineering

180+ engagement features · SAP module weights · complexity tiers

Step 03

Quantile LightGBM

Three gradient-boosted regressors → P10 / P50 / P90 effort

Step 04

Monte Carlo · 10k paths

Workstream PERT-Beta sampling under a Gaussian copula

Step 05

Risk attribution

Shapley values per workstream + sensitivity tornado

Step 06

Output

Effort, cost, duration band · risk drivers · scenario diff

Flowchart · 03

Fiori App Recommender

Bi-encoder retrieval + ColBERT-v2 re-ranking

Open tool

Step 01

Inputs

Role design · process taxonomy · industry · S/4 release

Step 02

Bi-encoder recall

Dense embeddings over Fiori Library + custom catalogue (top 200)

Step 03

ColBERT-v2 re-rank

Late-interaction cross-encoder for fine-grained precision

Step 04

Business rules filter

Licence, edition, BTP dependency, deprecation guards

Step 05

Output

Ranked Fiori apps per role with rationale + SAP Note links

Flowchart · 04

Clean Core Scorer

Deterministic rubric + LLM narration on Cloud ALM signals

Open tool

Step 01

Inputs

Cloud ALM extension registry · custom code analyser · API usage

Step 02

Rule-based rubric

Weighted deviation scoring across 7 Clean Core dimensions

Step 03

Grounded LLM narration

RAG over SAP Clean Core whitepapers · temperature 0.0 · cited

Step 04

Remediation planner

Side-by-side extensibility patterns · effort estimate per finding

Step 05

Output

Score 0–100 · deviation list · remediation roadmap

Flowchart · 05

Change Impact Analyser

Node2Vec embeddings on the SAP object graph

Open tool

Step 01

Inputs

Transports · custom objects · enhancement points · interface map

Step 02

Graph build

8k-node SAP object graph · DDIC, ABAP, CDS, BAdI edges

Step 03

Node2Vec walks

Biased random walks → 128-dim embeddings per object

Step 04

Impact propagation

k-hop similarity + transitive dependency closure

Step 05

Output

Impacted objects · risk tier · suggested test scope

Flowchart · 06

Test Coverage Intelligence

Hybrid RAG (BM25 + dense) + schema-constrained LLM

Open tool

Step 01

Inputs

Test pack · process inventory · RICEFW list · risk register

Step 02

Hybrid retrieval

BM25 lexical + pgvector dense over scripts and processes

Step 03

Coverage scorer

Process · RICEFW · risk-weighted gap matrix

Step 04

Schema-locked LLM

JSON-mode generation of missing test cases · grounded · cited

Step 05

Output

Coverage % · gap heat-map · draft scripts for Tricentis / Cloud ALM

Integrations · In the works

A²AI, connected to the rest of your stack.

Conceptual integrations on the A²AI roadmap. Each pairs an external platform with a specific A²AI tool to extend its reach beyond the SAP core.

Source-to-Pay

SAP Ariba

Concept · In design

Pairs with

Cost & Timeline Forecaster + Compliance Matcher

Pulls live category spend and supplier-risk signals from Ariba to refine the cost model — supplier-led RICEFW (PunchOut, cXML, custom approvals) is auto-detected and folded into the P10/P50/P90 forecast.

Smart Manufacturing / MES

Plex Automation

Concept · In design

Pairs with

Landscape Intelligence Pack + Change Impact Analyser

Reads MES routings, machine masters and shop-floor signals from Plex to enrich the SAP object graph with plant-level dependencies — change-impact propagation now reaches the line, not just the IDoc.

Lakehouse · ML Platform

Databricks

Concept · In design

Pairs with

All six Core Tools (model layer)

Externalises A²AI's quantile LightGBM, Node2Vec and embedding pipelines onto a customer Databricks workspace — models retrain nightly on the client's own anonymised engagement data while inference stays in-browser.

Cloud Data Warehouse

Snowflake

Concept · In design

Pairs with

Test Coverage Intelligence + Clean Core Scorer

Federates SAP Cloud ALM, Solution Manager and custom code telemetry into a Snowflake share — Clean Core deviations and test-coverage gaps become governed, queryable assets with row-level lineage for audit.

Partner names and trademarks belong to their respective owners. A²AI is independent and not affiliated with SAP SE, Ariba, Plex Systems, Databricks or Snowflake.