Start with a document, a repository, or both
Each source works on its own. With both, the agent combines the document's facts with what it finds in the code.
Upload a current-state SAD
Word, PDF, HTML, Markdown or text. The document is read in your browser.
Scan a GitLab repository
The token stays in this browser tab and is sent only to GitLab. It isn't saved. Use a short-lived, read-only token.
Architecture from code
Existing architecture designs
Diagrams, decision records and design documents found in the repository, with the differences between what they describe and what the code contains.
What the agent found
Technologies detected
Compliance scope detected
Security and operational gaps detected
Workload inventory
Brownfield migration questionnaire: on-premises to Azure, including AKS
Covers every application type in a brownfield estate: web and application servers, SQL Server with linked servers, SSIS, SSRS and SSAS, Oracle and other databases, mainframe and midrange, desktop and VDI, integration and batch, vendor products, containers and AKS. Pick the application types first; matching sections appear. Answers marked "from SAD / code" were detected; change any answer and every result updates.
Refactor plan
Refactoring is the default for each workload. The agent steps back to replatform, replace or retain only where your answers or the technology make refactoring unwise.
Migration waves
Future-state Azure architecture
Current to future infrastructure
Why Azure: the business case
Five-year cumulative cost of each option
Technical benefits
Business development benefits
Risks and how we manage them
Alternatives: Azure vs colocation vs SaaS
Colocation moves your hardware into a provider's building. It fixes facilities, not the architecture. The comparison below uses the same workloads and five-year horizon.
Where Azure beats colocation
Where colocation is still reasonable
SAD quality and completeness
The uploaded document scored against the SAD 6.0 template: which sections exist, how complete each one is, how they rank, and what is missing. Missing and incomplete sections are marked in red.
Data and IO: ingress, egress and data-heavy limitations
Every input and output the document and code reveal, and the limitations that data-heavy or chatty workloads face on Azure.
Budget: labor and non-labor
One-time migration budget split into labor hours by role and wave, and non-labor costs, compared with the budget band in the questionnaire.
FinOps: subscriptions and commitment options
Landing zone subscriptions, Azure resources, and the cost of pay-as-you-go against 1-year and 3-year reservations and savings plans.
Delivery plan: Jira backlog and sprints
The refactor list turned into Jira epics and stories, assigned to teams and scheduled into sprints by team velocity and wave order.
Bill of materials
Every Azure resource, AKS namespace, API, data service, pipeline and identity the plan needs, by application code, with owner and monthly cost. Export it, or compare it with what is already deployed in the subscription.
Generate the landing kit
Terraform, AKS namespace bundles, API Management definitions and policies, Liquibase changelogs, pipelines and runbooks, generated for each application code. Download the pack or open a merge request for the platform team to review.
Ask the agent
The agent knows your document's findings, questionnaire answers and generated plan. Ask for the executive recommendation or paste an objection from the team.