---
title: RQA
description: LLM-based for semantics. The RQA detects ambiguity, absolutes, compound requirements, and more — and generates an improvement proposal instantly.
---

[![Logo basewise](https://basewise.ai/hubfs/Logo%20basewise.svg "Logo basewise")](https://basewise.ai/en/home?hsLang=en)

[![Logo basewise](https://basewise.ai/hubfs/Logo%20basewise.svg "Logo basewise")](https://basewise.ai/?hsLang=en)

- [Applications](https://basewise.ai/en/applications)
- [About Basewise](https://basewise.ai/en/about-basewise)
- [Insights](https://basewise.ai/en/insights)
- [Contact](https://basewise.ai/en/contact)

More results

En

- English
- [Nederlands](https://basewise.ai/nl/rqa)

[Request your demo](https://basewise.ai/en/demo?hsLang=en)

[Request your demo](https://basewise.ai/en/demo?hsLang=en)

- [Applications](https://basewise.ai/en/applications)
- [About Basewise](https://basewise.ai/en/about-basewise)
- [Insights](https://basewise.ai/en/insights)
- [Contact](https://basewise.ai/en/contact)

[Book free demo](https://basewise.ai/en/demo?hsLang=en)

![RQA](https://basewise.ai/hs-fs/hubfs/RQA.png?width=500&height=328&name=RQA.png)

 LIVE - demo.basewise.ai

# Defining requirements has never been easier

Evaluates every requirement for ambiguity, singularity, and feasibility — based on INCOSE ISO 29148.

 

[Back to applications](https://basewise.ai/en/applications?hsLang=en)

## Generates an improved formulation for each requirement

Requirements Quality Analyser

Select a requirement below to view the full INCOSE quality analysis.

· SET ANALYSIS 7

1

2

3

4

5

CLASSIFICATIONAll

RULEAll rules

SET ANALYSISAll

7 / 7

| Score | ID | Title | SOI | Classification | Current requirement text | Findings | DUP / CONTRA |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | REQ-GEN-001 | Defect-free lifetime | Structure | Requirement | The viaduct shall always be completely free of any defects and shall never require maintenance. | R14 R22 |  |
| 2 | REQ-GEN-002 | Loads & fire safety | Structure | Requirement | The structure shall withstand traffic loads and wind loads and meet fire safety standards etc. | R18 R9 |  |
| 4 | REQ-PLT-001 | Deck inspection | Deck | Requirement | The deck surface shall be inspected every 12 months by a certified inspector. | None | DUP |
| 4 | REQ-PLT-002 | Annual inspection | Deck | Requirement | The bridge deck shall be inspected annually by a qualified structural engineer. | None | DUP |
| 2 | REQ-GEO-001 | Clearance height A | Geometry | Requirement | The minimum clearance height shall be 4.50 metres above road level. | R22 | CONTRA |
| 2 | REQ-GEO-002 | Clearance height B | Geometry | Requirement | The free height above the carriageway shall not be less than 4.20 metres. | R22 | CONTRA |
| 5 | REQ-GEO-003 | Deck width | Geometry | Requirement | The bridge deck shall have a minimum width of 9.50 m between inner faces of parapets. | None |  |

7 requirements loaded — select a row to analyse

GET STARTED

# Ready to accelerate your SE workflow?

 Request a free demo or start a pilot program. Our consultants guide you from onboarding to your first results.

[Request your demo](https://basewise.ai/en/demo?hsLang=en) [Contact](https://basewise.ai/en/contact?hsLang=en)

How it works

## Two layers. One reliable verdict.

The RQA analyses every requirement through a two-layer approach: first deterministically on hard rules, then via AI on semantics and context. This produces a substantiated quality verdict per requirement.

LAYER 1 · DETERMINISTIC

Algorithmic check on hard INCOSE rules

The first layer operates via a rule engine with word lists for Dutch and English. Fast, consistent and fully repeatable.

- Detection of vague terms: "sufficient", "adequate", "optimal"
- Open-ended formulations: "etc.", "among others", "including but not limited to"
- Check on shall/dient formulation (R31)

LAYER 2 · LLM-SEMANTIC

AI analysis on context and meaning

The second layer uses Claude to assess rules that require contextual understanding — singularity, verifiability and solution freedom.

- Detect compound requirements and generate split proposals
- Assess verifiability: are criteria measurable?
- Generate improvement proposal per requirement in INCOSE structure

Quality rules

## Eight INCOSE rules. Checked per requirement.

The RQA assesses every requirement against eight quality rules based on ISO/IEC/IEEE 29148 and the INCOSE SE Handbook. Every finding is substantiated and traceable.

Code

Rule

What is checked

Layer

R7

Vagueness

Vague terms without measure: "sufficient", "adequate", "appropriate", "efficient", "optimal", "reasonable"

L1 + L2

R9

Open ends

Open-ended formulations: "etc.", "and similar", "among others", "including but not limited to"

L1

R14

Absolutes

Technically unachievable absolute values: "100%", "always", "never", "completely", "absolute"

L1

R18

Singularity

Compound requirements using "and"/"or" that link two different subjects or performance requirements

L1 + L2

R22

Verifiability

Are concrete criteria, measurable values or test methods present? If not: requirement is not verifiable

L2

R28

Solution freedom

Does the requirement prescribe the solution instead of the need? Does not apply to constraints

L2

R31

Shall formulation

Use of "shall" as modal verb. "Must", "will", "can" are non-compliant with INCOSE

L1

C10–C14

Set-level

Completeness, consistency, feasibility, understandability and traceability of the full requirements set

Set

Export: .xlsx

Export: .csv

Score per requirement

Improvement proposal included

Classification

## Classify first. Then analyse.

Before the RQA assesses a requirement, the system determines whether it is a requirement at all. This prevents process descriptions or definitions from being incorrectly evaluated.

Input — three texts

The system shall provide a usable floor area of at least 350 m² per building level.

Group rooms groups 1,2: 6 units, 62 m² net, total 372 m².

This chapter describes the background of the programme of requirements.

= Category 1: REQUIREMENT

= Category 2: LATENT REQUIREMENT

= Category 3: NOT A REQUIREMENT

RQA  
classification

Output — different action per category

CATEGORY 1 — REQUIREMENT → full INCOSE analysis

Text contains a modal verb and expresses a need. All 8 quality rules are applied.

CATEGORY 2 — LATENT REQUIREMENT → reformulate

Implicit need via quantitative info or standards. RQA proposes an explicit reformulation.

CATEGORY 3 — NOT A REQUIREMENT → skipped

Process description, explanation or definition. No INCOSE analysis performed.

Improvement proposal

## Not just flagging. Also improving.

For every finding, the RQA immediately generates a concrete improvement proposal. From vague formulation to INCOSE-compliant requirement — in one step.

!

Original requirement

"The system must provide sufficient ventilation for the spaces, always and under all circumstances."

R7

Vagueness detected

"Sufficient" is a vague term without a measurable value. Specify a concrete standard, e.g. a ventilation rate in m³/h or m³/h per person.

R14

Absolute detected

"Always and under all circumstances" is a technically unachievable absolute. Replace with a concrete condition, e.g. "under normal operating conditions".

R31

Non-compliant formulation

"Must" is non-compliant with INCOSE. Use "shall" as the modal verb.

Quality score

2.1

out of 5.0 — Insufficient

R7 Vagueness CRITICAL

R14 Absolute CRITICAL

R31 Formulation WARNING

Improvement proposal

"The ventilation system **shall** deliver a minimum fresh air supply of **30 m³/h per person**, under normal operating conditions in accordance with EN 15251."

✓ "Sufficient" → 30 m³/h per person

✓ "Always" → under normal operating conditions

✓ "Must" → shall (INCOSE-compliant)

Use cases

## Proven in practice.

The RQA has been tested on requirements sets from construction, infrastructure and energy — from school buildings to complex infrastructure contracts with hundreds of requirements.

Construction · PvE

School building Programme of Requirements

Live RQA analysis on a complete school building PvE. Vagueness, absolutes and non-compliant formulations immediately flagged with improvement proposals.

Live tested · school PvE

Energy · Technical spec

Energy system specifications

RQA analysis on technical requirements for energy systems. Verifiability and measurability of performance requirements centrally assessed.

RQA testing successfully completed

Infrastructure · UAV-GC

Contract requirements Trapo Zuid

Quality analysis on requirements from the UAV-GC contract for rail project Zwolle–Dalfsen. Singularity and solution freedom particularly relevant in contract context.

Hegeman · BAM · ProRail

Universal

Large requirements sets (600+ requirements)

CAP-16 set-level analysis scales to 600+ requirements via a two-step architecture: Claude Haiku for the index, Claude Sonnet for set-level reasoning.

Scalable to 600+ requirements

Review · Audit

INCOSE quality audit

RQA output is fully traceable and audit-ready. Every finding is linked to an INCOSE rule code and supported by substantiation.

Audit-ready output

Future

Set-level duplicate detection (roadmap)

Automatic detection of overlapping and contradictory requirements within a requirements set. CAP-16 extension is on the roadmap for 2026.

Roadmap · 2026

[Back to applications](https://basewise.ai/en/applications?hsLang=en)

#### BASEWISE

**Basewise** provides modular AI solutions and applications for process optimisation.

Basewise Consultancy B.V.  
Chris Schutweg 3  
3632JK Loenen aan de Vecht  
The Netherlands  
KVK-number 98925717

<https://x.com/><https://github.com/Basewise/framework>

[Applications](https://basewise.ai/en/applications?hsLang=en)

- [Platform](https://basewise.ai/en/platform?hsLang=en)
- [DRE](https://basewise.ai/en/dre?hsLang=en)
- [RQA](https://basewise.ai/en/rqa)
- [REF](https://basewise.ai/en/ref?hsLang=en)

[Company](https://basewise.ai/en/about-basewise?hsLang=en)

- [Contact us](https://basewise.ai/en/contact?hsLang=en)
- [AI Disclaimer](https://basewise.ai/en/ai-disclaimer?hsLang=en)
- [Terms and Conditions](https://basewise.ai/en/terms-and-conditions?hsLang=en)
- [Privacy policy](https://basewise.ai/en/privacy-policy?hsLang=en)

 Copyright © 2026 basewise.ai