Quality intelligence
Connect requirements, business rules, test cases, defects and execution evidence to expose coverage uncertainty.
A clearer view of enterprise software risk
Know what changed. Understand what matters.
Project Ignite connects requirements, applications, data, tests and delivery evidence to help teams uncover quality blind spots and make better engineering decisions.
Interactive product concept using fictional banking data. Results are illustrative and are not a certification, audit opinion or production-system assessment.
One connected intelligence layer
From fragmented evidence to informed action
Explore the core capabilities
Data intelligence
Profile datasets, identify anomalies and examine integrity, boundaries and execution trends.
Engineering intelligence
Understand application dependencies, repositories, services and the likely impact of change.
Change & release risk
Bring together change context, dependencies and evidence to help prioritise validation.
AI engineering teammate
Plan and assist with development, testing, review and documentation under controlled permissions.
Evidence & governance
Preserve source, timestamp, version and provenance so findings can be independently reviewed.
Try a real-world pattern
Bank transfer quality scenario
Explore how business rules translate into test conditions.
$1–$10k
Single transfer range · fictional rule
$20k
Daily transfer limit · fictional rule
OTP
Additional verification required
Interactive banking quality lab
Explore the edges of a transfer.
Change the transfer amount and account conditions to see how a small set of illustrative business rules can generate meaningful test scenarios.
Sample business rules
Minimum transfer$1.00
Maximum single transfer$10,000
Daily cumulative limit$20,000
Available balanceMust be sufficient
One-time passwordRequired
These rules are demonstration assumptions, not verified banking requirements.
Try a transfer
Enter values and evaluate the sample rules.
Autonomous engineering · controlled assistance
A teammate that explains its work.
Explore a scripted example of how an AI engineering teammate could reason about a quality concern. This demo does not connect to a repository, execute code or call a live AI service.
✳ Ignite engineering teammateSample mode
What should we test for a transfer timeout after debit?
I would investigate whether the debit was committed before the response timed out. The key risk is a customer retry creating a duplicate transfer or an unclear transaction state.
Suggested checks
1. Retry with the same idempotency key.
2. Confirm the ledger contains one debit.
3. Verify status and customer notification.
4. Reconcile pending transactions.
1. Retry with the same idempotency key.
2. Confirm the ledger contains one debit.
3. Verify status and customer notification.
4. Reconcile pending transactions.
Controlled autonomy model
Analyse requirementsAdvisory
Draft test casesAssisted
Modify code in sandboxApproval-gated
Merge or deployHuman approval
In a real deployment, agent identity, permissions, workspace isolation, audit and independent verification would be required.
Ways to explore Project Ignite
One product. Two entry points.
The public demonstration and the planned ChatGPT experience will share the same product concepts and, once implemented, the same governed backend.
Public web demonstration
Explore sample workflows, quality scenarios, risk observations and the product vision in a browser.
Project Ignite in ChatGPT
A future ChatGPT app could provide conversational access and interactive views. It must be built, tested and published separately.
Governed intelligence services
Common knowledge, evidence, permissions and business logic, with clear tenant boundaries for any real customer use.
Current status: This is a self-contained demonstration preview. This is a self-contained demonstration preview, publicly deployed on Cloudflare Pages. The ChatGPT integration is planned and is not yet a published ChatGPT app. All records and analysis shown are fictional and is not yet a published ChatGPT app. All records and analysis shown are fictional.