A complete engineering estate — requirements, traceability, mission analysis and training — deployable inside your own network, with the AI models themselves running on hardware you control. No foreign cloud in the data path, and no dependency you cannot terminate.
lifecycle phases it takes part in — from first requirement to as-operated record
The six phases
Each phase has a lead product and the others that support it. The same six phases the in-product launcher and the guided tour narrate.
01
Intelligence & R&D
Supremacy — See the gap first.
WeProject.ai — Anchors SSOT & KPI contract.
SMAT-AI — De-risks the concept fast.
02
Specification & Requirements
WeProject.ai — Owns the requirement baseline.
SMAT-AI — Proves the KPI early.
Supremacy — Positions the winning bid.
03
Design & Analysis
SMAT-AI — Proves the mission, browser-native.
WeProject.ai — Frames the design space.
04
Procurement
WeProject.ai — Runs acquisition end-to-end.
SMAT-AI — Compares proposals, one yardstick.
05
AIT · Verification & Validation
SMAT-AI — Verify by model, not luck.
WeProject.ai — Traces V&V to requirements.
06
Operations & Sustainment
SMAT-AI — Rehearse the mission you designed.
WeProject.ai — Holds the as-operated record.
Cross-cutting · trains every phase
Aerospace Academy
Workforce readiness — trains the people who run every phase, in every domain. 94 simulators · 500+ assessments.
The spine · under every phase
Data Domain
329 KPIs · 1,300+ platforms · 90+ reference APIs — the A&D data spine that seeds every product, with offline reference bundles in WeProject, Supremacy, SMAT and Academy.
Licence only what you need
The rail carries the products you own, and follows you between them.
Every product licensed — the rail carries all five.
WeProject.ai
Flagship
AI-native MBSE & programme management.
Graph-native specification & requirements, traceability, cost estimation, risk, and compliance.
It is the same component in every product, so a preference set once holds everywhere — and an operator moving from mission design to the requirement baseline keeps their context instead of rebuilding it.
Theme
Light, dark, or follow the operating system.
Contrast
A high-contrast mode for daylight and control-room screens.
Display scale
Type and target sizes adapt to the display it is opened on.
Audience lens
The same page, filtered to what one role needs to see.
Mission context
Carries the selected mission across every product you open.
Engineering knowledge is six kinds of thing
A mission objective, a thermal constraint, a test report and an approval record answer different questions. ISBEGO names the six, so "what have we not covered" becomes a query instead of a debate.
I
Intent
10
The purpose and direction — WHY the system or organization exists and what it is trying to achieve.
S
Structure
23
The make-up and composition — WHAT the system and organization are built from.
B
Boundaries
19
The limits and rules that constrain the system — what it must and must not do.
E
Evidence
17
The artefacts and proof that things are true or done.
G
Governance
21
Decisions, authority and oversight — how the work is steered and controlled.
O
Operations
32
The activities and execution — how the work actually gets done over time.
Real workflows from the shipped corpus — some finish inside one product, some cross three. Each one says how much of it a machine can redo and how much still needs a person.
Five worked examples from the workflow corpus.
Mission Definition
WeProject mission workflow
Say what the mission actually needs, then plan the workflow that delivers it — two decisions a person takes inside WeProject.
WeProject.ai
1Define the mission requirement— WeProject.ai, a person decides
2Plan the delivery pathway— WeProject.ai, a person decides
Within one productWorkflowEvery step is a person's call (0%)
Mission Definition
Authority thread (organization → KPI)
Trace who authorised what — from the directive an organisation issues down to the measure it is judged by. Every step is a person's call.
WeProject.ai
1An organisation issues a directive— WeProject.ai, a person decides
2The directive tasks a mission— WeProject.ai, a person decides
3The mission calls for capabilities— WeProject.ai, a person decides
4Each capability produces its measure— WeProject.ai, a person decides
Within one productWorkflowEvery step is a person's call (0%)
Analysis & Decision
SCP capability capacity roll-up
Add up what a capability can really field, from its children upward. Fully computable — no decisions in it.
Data Domain
1Sum every descendant capability, deterministically— Data Domain, computed and checked
Within one productChainRecomputes end to end — nobody re-decides anything (100%)
Analysis & Decision
Threat-countering trade-off (kill-web → Pareto)
Prove the capability fabric really does counter the threat, then rank what survives. Nobody has to decide anything — it recomputes end to end.
Data DomainSMAT-AI
1Compose the kill-web and find the binding constraint— Data Domain, computed and checked
2Rank what survives on the Pareto front— SMAT-AI, computed and checked
Across 2 productsChainRecomputes end to end — nobody re-decides anything (100%)
Analysis & Decision
MOD capability sufficiency (strategy memo)
Show a capability is enough for the mission and actually counters the threat, before the memo is signed.
WeProject.aiSMAT-AIData Domain
1Scope the requirement, or the gap— WeProject.ai, a person decides
2Name the target capability— WeProject.ai, a person decides
3Prove the mission is sufficiently covered— SMAT-AI, computed and checked
4Prove the fabric counters the threat— Data Domain, computed and checked
5Rank the trade-off— SMAT-AI, computed and checked
6Close it with a compliance review— WeProject.ai, a person decides
Across 3 productsWorkflow3 of 6 steps recompute; the rest are decisions (50%)
These are five of the eight workflows seeded in the shared corpus. The other three — SCP strategy decomposition, SCP bottom-up synthesis, SCP policy governance completeness — are each a single human step inside WeProject, so they would repeat an example already above rather than add one.
Who it's for
Eighteen segments across defence, space, industry and education. Each has a brief covering the challenge, what the platform does about it, and the modules involved — no sign-in needed to read one.
How the estate actually fits together — one sign-in, one authoritative store for shared reference data, and a model provider you can host yourself
Drawn from the estate's own integration map. The same diagram, in more detail, is what our engineers work from.
Advanced Data Engine
Proprietary relationship mapping technology for complex dependency tracking
Scalable Architecture
Enterprise-grade modular design for high-performance and reliability
Real-time Collaboration
Instant synchronization and live updates across distributed teams
Flexible Deployment Options
Cloud Deployment
Scalable, managed cloud infrastructure with automatic updates and global availability
• Runs in europe-west1
• Scales to zero and back with demand
• Updated for you, no upgrade window to plan
Air-Gapped Deployment
Secure on-premises installation for classified and sensitive environments
• Complete network isolation
• Local data sovereignty
• Custom security compliance
Backend Technologies
Graph store:Neo4j over Bolt, with immutable entity versions
API:Express REST, served from api.weproject.ai
Real-time:Socket.IO impact stream
Reference data:Postgres single source of truth, per-app RBAC
Inference:OpenAI, Anthropic, Gemini — or self-hosted Ollama / vLLM
Frontend & Performance
Interface:Next.js and React, TypeScript throughout
Document export:Markdown, PDF, DOCX and PPTX, rendered server-side
Visualisation:ECharts, AG Grid, and Three.js for 3D scenes
Ingest:Word, Excel and legacy databases
Deployment:Google Cloud Run, or air-gapped on your own hardware
Numbers you can check
Every figure below is re-derived by a named check that runs on each build. If one stops being true, the build fails rather than the page quietly ageing.
329
KPIs in the shared reference registry
Checked by verify-launcher-claims.mjs counts the mission-contract catalogues the Data Domain seed loads — measured 359 at the last run.
1,300+
platforms in the aerospace & defence data spine
Checked by verify-launcher-claims.mjs queries the live Data Domain database
94
interactive simulators in Academy
Checked by verify-launcher-claims.mjs counts the simulator files that ship — measured 104 at the last run.
90+
reference data APIs
Checked by verify-launcher-claims.mjs counts the route modules Data Domain serves — measured 280 at the last run.
These are capability counts — what the platform contains. They are not customer outcomes, and we do not publish outcome percentages we have not measured. Ask us for a reference and we will put you in touch with someone who is using it.
Questions we get asked
Including the ones with awkward answers.
Is WeProject.ai one product or several?
Both, and the distinction matters. WeProject.ai is the platform; WeProject is its flagship application, the graph-native requirements and programme governance tool. Alongside it run SMAT-AI for space mission analysis, Academy for training, Data Domain as the shared reference-data source of truth, and Supremacy for market and competitive intelligence. They share one sign-in and read the same reference data, and you licence only the ones you need.
Can it run air-gapped, with no outbound connection?
Yes, and this is the capability the platform is built around rather than a bolt-on. The AI layer routes to whichever model provider you configure, and two of the supported options — Ollama and vLLM — run entirely on hardware you own, so no prompt or document leaves your network. The data stores are self-hostable, and each product ships an offline reference bundle so it still works when the shared data service is unreachable.
Which AI models does it use? Can we bring our own?
The model provider is a configuration choice, not a fixed dependency. The AI service routes to OpenAI, Anthropic, Google Gemini and DeepSeek for hosted inference, or to Ollama and vLLM for self-hosted models. You can point it at your own deployment, and switching providers does not change the product surface.
Do we have to buy all five products?
No. Each product is licensed separately and the navigation adapts to what you own — a product you have not licensed simply does not appear. A space operator might take SMAT-AI, Data Domain and Academy; a programme office might take only WeProject and Data Domain.
How do people sign in across the products?
One central authentication service issues a session that every product accepts, so a user signs in once and moves between products without signing in again. It supports password sign-in, Google and Microsoft SSO, and multi-factor authentication. Permissions resolve on two axes: what the person is allowed to do, and which application is asking.
If several products hold the same data, which one is authoritative?
Data Domain is. It is a single Postgres source of truth for the shared reference data — missions, platforms, subsystems and the KPI registry — and it is the only service that writes to it. Every other product reads from it. Applications that hold read-only access can still propose a change, which an operator reviews and approves rather than editing in parallel. It carries 329 KPIs, 1,300+ platforms and 90+ reference APIs.
How do our engineers learn the platform and the domain?
Academy is a full product rather than a documentation site: 94 interactive simulators and over 500 assessments covering orbital mechanics, satellite operations, electronic warfare, model-based systems engineering and leadership. It is licensed like the others and shares the same sign-in.
Can we bring in requirements and data we already have?
Yes. The platform ingests Word documents, Excel workbooks and legacy databases, and exports finished documents server-side as Markdown, PDF, DOCX and PPTX. The export pipeline versions what it produces, so a document you issued can be reproduced later exactly as it was issued.
Does the platform make us compliant with DO-178C, ISO 14971 or DFARS?
No tool can. What the platform does is support the engineering workflow those standards require: bidirectional requirement traceability, immutable baselines, an audit history of every change, verification and validation evidence linked back to the requirements it satisfies, and documents generated from that record rather than maintained beside it. Certification remains an assessment of your organisation and your evidence, by your auditor. We can show how the platform produces the evidence; we cannot and do not certify you.
Where does our data physically live?
Wherever you choose. The hosted deployment runs on Google Cloud Run in the europe-west1 region. The self-hosted and air-gapped deployments run entirely on your infrastructure, in which case no data reaches us at all — including prompts, if you use a self-hosted model.
How do we get access, and what does it cost?
The platform is in beta and access is granted from the waiting list rather than by self-service signup. Commercial terms depend on which products you licence and whether you deploy hosted or air-gapped, so pricing is quoted rather than published. Join the waiting list below and tell us which products and which deployment model you need — that is enough for us to come back with a straight answer.
What happens to our data if we stop using the platform?
It comes with you. Requirements, baselines and the document record export to open formats — Markdown, PDF, DOCX, PPTX and structured data. The reference data model is documented rather than proprietary, and the self-hosted deployment means you can already be holding your own database.
Something not answered here? Ask it in the form below and we will answer it directly.
Join the beta
Three fields is all we need to start.
We use this to reply to you and nothing else. No newsletter, no third parties.