The Software Lifecycle, run by agentic AI
Two entry points – a new greenfield idea or an existing legacy application – pass through ten governed phases with hard Go/No-Go gates and measurable KPIs. In every phase a ReqPOOL platform tool does the heavy lifting: hosted in the EU, no training on client data, no licence costs for the client.
Agentic AI compresses analysis, specification, implementation and acceptance from quarters to weeks.
Every statement traceable to source code, screenshot or specification clause – end to end.
All platform tools are part of the ReqPOOL consulting service – no tool procurement needed.
The Software Lifecycle, run by agentic AI
Two entry points – a greenfield idea or an existing legacy application – through ten governed phases with hard Go/No-Go gates and measurable KPIs. In every phase, a ReqPOOL platform tool does the heavy lifting: EU-hosted, no training on client data, no licence costs for the client.
Greenfield
A new idea, product or capability – built from a clean sheet with no legacy constraints.
Ideation
Structured ideation document: business value, users, core capabilities – feasibility checked against benchmark data.
- Business value, users and core capabilities captured in a structured way
- Feasibility checked against benchmark data
- Decision-ready ideation document as the basis for scoping
Decision-ready ideation document in days, not weeks
Scoping
IFPUG function points from 7 parameters – estimate in minutes. Three delivery variants compared: classic · agile · AI-coded, with Best/Expected/Worst bands and a versioned scope baseline.
- IFPUG function points from 7 parameters – estimate in minutes
- Three delivery variants compared: classic · agile · AI-coded, with Best/Expected/Worst bands
- Versioned scope baseline and scope document for the Go/No-Go decision
Estimate in minutes, not days · versioned scope baseline
G1 · Go / No-Go · Idea approved
Legacy
An existing legacy application – undocumented, business-critical, due for modernisation.
Reverse Engineering of Legacy Systems
Blackbox – from usage
An AI agent operates the app like a user, or observes real users; screens, rules and processes become the AS-IS specification.
Whitebox – from source code
Reads any code base – SAP/ABAP, COBOL, NATURAL, Java … – into source-proven documentation, a queryable knowledge graph and transformation advice.
- Blackbox: An AI agent operates the app like a user, or observes real users – screens, rules and processes become the AS-IS specification
- Whitebox: Any code base becomes source-proven documentation, a queryable knowledge graph and transformation advice
- Every statement evidence-linked – traceable to a screenshot or a line of source code
100% functional coverage · every statement evidence-linked · analysis in days, not months
Streams merge – one governed delivery stream
Streams merge – one governed delivery stream
Requirements Engineering
- Workshop videos, protocols and ideation documents turned into specification documents
- Requirement quality checked automatically, improvements suggested
- AI coding intent created as machine-readable build order
100% testable requirements · quality-scored spec · weeks, not quarters
Design Software Architecture
- AS-IS landscape: processes, applications, interfaces, infrastructure
- AI-derived recommendations decided on the EAM decision board
- Target architecture and transformation roadmap per year
Decision-ready target architecture · every recommendation drill-down traceable
Implement Software
- EU-based AI codes the software from the AI coding intent
- Scope changes tracked as versioned change requests
- Human engineers review, harden and integrate
Delivery in months, not years · scope drift measured against baseline
Test Software
- Specification decomposed into atomic Gherkin requirements
- Test cases incl. edge and negative cases generated automatically
- AI agent executes tests, collects evidence, suggests bugfixes
Full spec coverage · traceability matrix from clause to test result
User Acceptance & Sign-Off
- Guided playbooks for business users; screen recordings evaluated multimodally
- Independent AI review model rates every test case
- Revision-proof acceptance report – final acceptance stays with the client
Acceptance in days · audit-proof report · zero untested clauses
Deployment
- Cut-over plan, data migration and interface switch
- Parallel run and reconciliation against the AS-IS baseline
- Controlled legacy switch-off, landscape updated
Zero unplanned downtime · legacy switched off on date
Run Software
- Lifecycle and enterprise architecture management in one inventory
- End-of-support detection and technical-debt monitoring
- New needs feed the next ideation cycle – the loop closes
Falling run costs · managed technical debt · continuous roadmap
Run Software feeds the next ideation cycle – the lifecycle closes into a continuous loop. ↑ 01
Nine agentic tools, one suite
Hosted in the EU, no training on client data, included in every ReqPOOL project without licence costs. Proprietary – available only at ReqPOOL, exclusively for our partners.
Estimate effort
Function-point estimates from 7 parameters in minutes; ideation and scope documents; versioned baselines and change requests.
Phase 01 · 02 · 06
Understand applications
Observes an application in use and compiles an evidence-linked AS-IS specification with processes and business rules.
Phase 03
Understand legacy code
Sysparency Legacy-Code Intelligence Platform: reads any source code into source-proven documentation and a knowledge graph.
Phase 03
Generate specifications
Turns workshop videos, protocols and ideation documents into specifications and creates the AI coding intent.
Phase 04
Specify at top security
Requirements engineering for the highest security level: symbolic instead of generative, no external LLMs – requirements stay entirely in-house.
Phase 04
Check requirements
Checks the quality of requirements and specifications, gives structured feedback and suggests improved, testable formulations.
Phase 04 · 06 · 07
Build software agentically
The complete, fully automated Agentic Software Development Lifecycle: EU-based AI develops, tests and delivers software from the AI coding intent – with gates, human validation and a complete audit trail. Already in internal use at ReqPOOL.
Phase 04 · 05 · 06 · 07 · 08 · 09
Accept software
Generates and executes test cases against the running software, suggests bugfixes and produces a revision-proof acceptance report.
Phase 07 · 08
Steer architecture
Software lifecycle management, EAM decision board, target landscape and IT transformation roadmap.
Phase 05 · 09 · 10
Evidence instead of promises
All figures are backed by scientific and industry evidence. Across full projects including strategy, specification and steering, the evidenced productivity value ranges from 20 to 45 percent – we communicate this distinction transparently.
Try reqChecker right now
No login: paste requirements and let every sentence be checked against IREB and ISO 29148 criteria – the analysis runs directly in your browser.
The AI dividend stays with you
Our platforms are not sold on the market and are used exclusively in consulting projects. You pay only for expertise and effort – no tool licences, no platform fees.
- EU hosting: Microsoft Azure in European regions, AI processing in the EU data zone
- No AI training: Your data never trains models – not even through AI processing
- Traceability: Every statement traceable to source, screenshot or clause
- Intellectual property: Specification, code and documentation are 100% yours
Get in touch
Arrange a no-obligation initial conversation with our contact person.
Christian Buchegger
Chief Sales Officer & Authorised Signatory
