Skip to content

Aircraft certification, computed

One connected model for requirements, safety analyses, test evidence and the regulations. Change one thing, and the answer arrives in seconds

Not a pile of disconnected documents

One living model, everything connected

Change one thing and the whole case answers

Evidence always ready to hand over

Building the aircraft is no longer the bottleneck, proving it is

The proof that an aircraft is safe lives in pieces. Requirements in one tool, safety analyses in another, test evidence in spreadsheets. Nothing joins them but a person

REQUIREMENTSin one tool
no link
SAFETY ANALYSESin another
no link
TEST EVIDENCEin spreadsheets
01
A change landsA new sensor, a battery, a software release
02
Nobody can say what it brokeThe links between the pieces were never machine readable
03
So it is all re-checked by handBy the most expensive engineers on the programme
Footage NASA Langley Research Center · tiltwing model in the 14 by 22 foot tunnel, 2025

More new kinds of aircraft than the rules were written for

Electric
No precedent to copy from
Hydrogen
Failure modes the codes never met
Uncrewed
The airframe changes in weeks
Every one of them still has to be certified, against rules written for programmes that changed once a decade.

One model instead of a filing system

TodayCertProve
A requirements tool
Requirements
A safety analysis tool
Safety analyses
Spreadsheets
Test evidence
Documents in a drive
The regulations
An engineer reconciling them
One connected model

Certification is the price of flight

For novel aircraft, certification now costs more than development itself

92M+
ABOUT £80M · $100M
to certify a commercial aircraft
23M
ABOUT £20M · $25M
for a general aviation type
92
ABOUT £80 · $100
per line of DO-178C software code

SOURCES · INDUSTRY CERTIFICATION COST STUDIES · DO-178C BENCHMARKS

REQUIREMENTSSAFETY SHEETSDOCUMENTSTESTSCADEMAIL

Eight to sixteen tools per programme, and none of them talk, senior engineers keep them consistent by hand, after every design change

Everything on one graph

Requirements, hazards, architecture, tests and evidence are one connected model, graded against the live rule and recomputed on every change

CertProveReference airframe / Jet A-1 fuel system / Programme graphCS-25 · 25.1309
Living model · requirement to evidenceone connected graph
SYSTEMHAZARDREQUIREMENTVERIFICATIONEVIDENCECOMP-3COMP-7HC-01HC-04REQ-014REQ-021REQ-042FT-09FT-17TC-61EVID25.1309SEAL
0 nodes · 0 edgesrev 6c5176cgraded against EASA and UK CAA rule text

Smarter certification, built to be trusted

It cites the rule or abstains, the trace is the structure not a maintained link, and every gate ladders up to one honest verdict

CertProveReference airframe / Jet A-1 fuel system / AssuranceCS-25 · 25.1309
Trust · governor, trace, verdictcite or abstain
The governor cites the rule or abstains
FACT≤ 1×10⁻⁹ per flight hour
COMPUTATION2.1×10⁻¹⁰ · meets target
ABSTAINwithheld, needs a source

Every answer is fact, computation, precedent or abstain, and it refuses to invent a number.

Traceability that cannot rot
REQ-014 → FT-09pass
REQ-021 → no testunverified
FT-17 → nothingorphan
alloc → deleteddangling

Verification truth, reverse orphans and dangling references are found before the authority asks.

One readiness verdict
0%ready

2 objectives open · FT-17 running, TC-61 not started

Readiness recomputes live as evidence lands, always answering can we certify yet.

trace health checkedrev 6c5176c2 objectives open

The machine you must prove

Every failure condition, graded against the rule

Proven, not claimed

Watch the case build itself

One description of a system, graded against the live rule. Hazards, requirements and verification resolve into one computed verdict, recomputed on every change, never a status someone typed.

CertProveReference airframe/Jet A-1 fuel system/Safety caseCS-25 · 25.1309
Lattice · system to verdictseverity at or above MajorDAL flow-down on auto-recompute
SYSTEM1HAZARD8 conditionsREQUIREMENT8 typedVERIFICATION6 pass · 1 run · 1 openVERDICTJet A-1fuel system01Loss of fuel feed to engineHC-01CATDAL AREQ-014FT-09FTAPASS02Undetected fuel leak, dry bayHC-02CATDAL AREQ-021AN-12ANALYSISPASS03Ignition source within tankHC-03CATDAL AREQ-063AN-31CCAPASS04Fuel starvation, imbalanceHC-04HAZDAL BREQ-042FT-17FTARUN05Misleading fuel quantityHC-05MAJDAL CREQ-035TC-44TESTPASS06Uncommanded fuel transferHC-06MAJDAL CREQ-051AN-23ANALYSISPASS07Tank overpressure, vent blockHC-07MAJDAL CREQ-058TC-61TESTOPEN08Contaminated fuel supplyHC-08MINDAL DREQ-070INS-4INSPECTPASS

Worst-case catastrophic rate

0.0×10−10/ FH

Target at or below 1.0×10⁻⁹ /FH · 25.1309 catastrophic

Meets target
Objectives closed 0 / 8
Catastrophic verified 0 / 3

2 objectives open · FT-17 running, TC-61 not started

live0 nodes · 0 edgesrev 6c5176crecompute ··synchronising
Reference airframe · Jet A-1 fuel system · 8 failure conditions · 3 catastrophicARP4761A · ARP4754B · CS-25 25.1309 · DO-178C DAL A

Watch it catch the mistake as it happens

One failure condition, typed against the live rule. Word it with a cause, or downgrade its severity, and the check fires that moment, cited and computed

CertProveReference airframe/Jet A-1 fuel system/Functional Hazard AssessmentCS-25 · 25.1309
FHA-018 · new rowseverity at or above Major checking as you type
FHA-018#1unclassified
FUNCTION
LIVErev 6c5176crecompute 38 msgraded against the live rule

A different kind of machine

Tools that draft certification documents with language models now exist. Useful, and not this

A drafting machine

The detector shall alarm at 2 per cent

plausible · unverified

Generates likely text, fast. A human verifies every line, because the machine cannot

The computing machine
evidenceTEST-014 · passverdictobjective met
CS 25.1309sha256 6c5176c…recomputed
no evidence on file?abstains · never invents

Derives the verdict from your evidence, cited to its source and reproducible bit for bit. Where evidence is absent, it says so

everything load-bearing runs with AI offthe optional assist, off by default, on your key and your endpointsame input, same output, always

From the first hazard to the sealed case

Six analyses, one linked flow, every verdict graded against the 25.1309 ceiling, catastrophic ≤ 1e-9

  1. 01
    FHA

    Functional Hazard Assessment

    Failure conditions, severities and safety objectives, the source of truth

  2. 02
    FMEA

    Failure Modes & Effects

    Component failure modes derived from the model, with a criticality matrix

  3. 03
    PSSA

    Preliminary System Safety

    Architecture allocated against 25.1309 targets, with DAL and budget checks

  4. 04
    FTA

    Fault Tree Analysis

    Quantitative fault trees, cut sets and single points of failure versus target

  5. 05
    CCA

    Common Cause Analysis

    Common-mode groups and particular risks that verify independence

  6. 06
    SSA

    System Safety Assessment

    Closure, every objective marshalled into one computed verdict

  7. SEAL

    Sealed case

    Tamper-evident, anyone can verify

One condition, end to end: total loss of lift · Catastrophic → DAL A → target 1.0×10⁻⁹ per flight hour → two independent lift channels at 3.0×10⁻⁵ each → computed 9.0×10⁻¹⁰ objective met. Its neighbour stays open until the vendor rate arrives, and the case says so.

Change one thing, and the answer arrives in seconds

IN
One changeA battery, a sensor, a software release
01What it affectsRequirements, hazards and interfaces, with the path that connects them
02What survivesEvidence that still counts, and the reason it still counts
03What must be redoneNamed, never quietly reused
04The plan to fix itOrdered by what unblocks the most
Minoryou approve it yourself
Majoran authority cycle

That one question sets the timeline of the programme. Answered from the recorded evidence, every factor named, the rule cited, and undetermined where the evidence is missing.

All the engineering in one place

GRAPH

Programme workspace

Requirements, hazards, analyses and verdicts on one connected graph

SAFETY

ARP4761A safety chain

FHA, FMEA, PSSA, FTA, CCA and SSA as one linked, computed flow

RULE

Live regulatory grounding

Answered against the live rule text, cited or refused, never guessed

TRACE

Deep trace health

Verification truth, orphans, dangling refs and DAL flow-down

SEAL

Evidence integrity

Tamper-evident baselines anyone can verify, no exported files

CREW

The grounded crew

Drafts the next row and gap with its sources attached, or abstains

Retire tools, recover time

Software earns its place two ways, better work, or the same work with less resource, here is the arithmetic

A TEAM OF SIX, TODAY

Requirements managerlicence
Safety worksheetslicence
Trace matrix toollicence
Test managerlicence
Document controllicence
Risk registerlicence
Compliance trackerlicence
Keeping them consistentsenior engineering hours
Half the tooling cost
one platform in place of the licence stack, indicatively
Half an engineer back
reconciliation hours recovered over six months, spent on engineering instead

INDICATIVE FIGURES · WORKED CASE BY CASE WITH EACH PROGRAMME

One licence per programme per year, from 175k EUR for a single approval and 350k EUR at type certificate scope

The depth behind every surface

CITED

Every answer carries its source

The engine cites the rule or it abstains, it will not invent a number, and an unsourced fact becomes an honest gap

WHOLE

Reasoned across the entire aircraft

Systems, structures, powertrain and software, each graded against the live rule rather than a generic checklist

LIVE

A verdict that recomputes itself

Readiness updates the moment evidence lands, computed on every change, never a status someone typed

One evidence base, every authority

A programme certifies at home while validating elsewhere, and every authority wants the case in its own terms

THE METHODS ARE GLOBALARP4754A and ARP4761, and the FHA, PSSA, FTA, FMEA, CCA and SSA chain, are the same standards in every jurisdiction, so the analysis core is authority independent
EASAEuropean Union Aviation Safety Agencybuilt today
UK CAAUK Civil Aviation Authoritybuilt today
FAAFederal Aviation Administrationnextharmonised with CS-23 and CS-25
ANACBrazilon customer demand
TCCATransport Canadaon customer demand
JCABJapanon customer demand
one connected evidence baseproved oncepresented to each authority in its own termsconcurrent validations, one record

The wave is here

Three forces make a living certification case necessary now, not eventually

20+eVTOL companies in active certification
10+countries they are certifying in
3type certificates issued worldwide
WAVE

A whole new category, still ahead

The certification workload of an entire aircraft category is still in front of the industry

RULES

The rules move mid-programme

Special conditions for novel aircraft evolve while the programme runs, a case reconciled by hand goes stale against a rule that changed last month

PEOPLE

The scarcest resource is judgement

Experienced certification engineers are too valuable to spend on keeping spreadsheets consistent, the platform returns their hours to engineering

Every discipline on one graph

Systems, structures, powertrain and software, every team designing and verifying on the same living model, so a change in one ripples to every other

Three questions a certification manager will ask

Does it make the team faster?
The same submission, with fewer senior hours spent making four tools agree with each other
Does it make the case better?
Every link re-checked on every change. Nothing asserted without a source, and gaps named rather than filled
Does it add a burden?
No new tool in the compliance loop to qualify. Deterministic engines, auditable output, your own AI key or none at all

Bring us a slice you have already signed off

Fixed fee, two to four weeks, all seven engines, and a report you keep either way

frozen and content hashedset beside your own conclusionsreference by referencethe fee comes off year one

We may find things in it.