Frequently Asked Questions

PhoenixCode AI is an enterprise platform that transforms legacy applications into modern cloud-native solutions on .NET 10 + Angular 22 + SQL Server. Delphi is the source taken end to end today, with a public run and published prices; PowerBuilder, FoxPro, VB6, COBOL, Oracle Forms, Java and .NET Framework have a parser and a translator and start with a paid pilot module. It combines deterministic, rule-based transformations with generative AI where a rule cannot apply, and reports the coverage it actually achieved.

Manual rewriting is priced in developer-months and delivered at the end. PhoenixCode publishes a fixed price up to 200K lines of code (a small Delphi project is $35,000 – $60,000, a medium one $75,000 – $180,000), quotes larger systems per wave after a discovery, and invoices each wave against acceptance cases you sign. We do not publish a timeline: nothing measured supports one yet. What is measured is the 76-step deterministic blueprint and the reference run on the platform page.

Clean Architecture .NET 10 backend, Angular 22 standalone components with DevExtreme, SQL Server migrations, Docker configurations, CI/CD pipelines, comprehensive documentation, and auto-generated tests — all production-ready.

Yes. PhoenixCode Lens runs 100% locally on your machine. Only small code fragments are sent to our AI proxy for business rule extraction. We offer full on-premise deployment for the Platform. Your code never leaves your network.

Full support for Delphi/Object Pascal (DFM forms, PAS units, Paradox databases), Angular/TypeScript (components, services, routes), and .NET/C# (controllers, Entity Framework, NuGet dependencies).

Analyze up to 10 files and 5,000 lines of code with no license key. You get a dependency graph, component inventory, and complexity analysis. AI business rule extraction requires a paid license (Starter: $2,000).

Scanning is a local, deterministic pass and is typically quick; AI business-rule extraction takes longer and scales with the size of the codebase and the number of AI calls in your tier.

HTML interactive dashboard with D3.js visualization, professional DOCX report, JSON structured data export, and CSV inventory export. The dashboard includes dependency graphs, risk matrices, and effort estimates.

Download the self-contained binary for your OS (Windows, macOS, or Linux) from phoenixcode.ai/download, unzip, and run phoenixcode-lens. No .NET install required; Node.js 22+ is needed for Angular/TypeScript analysis.

Yes. Lens v1.5.0 includes a push command that sends your assessment blueprint directly to the Platform for one-click migration start. Run phoenixcode-lens push after your scan.

19 specialized AI agents wired into the migration pipeline: 7 master agents that coordinate each phase plus 12 specialists (performance, security, API-contract, architecture, SRE, and more). Masters use Claude Opus for reasoning; specialists use Claude Sonnet for focused tasks.

1) Build Verification — code must compile. 2) Unit Tests — 90%+ must pass. 3) Integration Tests — 100% must pass. 4) Security Scan — no critical OWASP vulnerabilities. 5) Code Quality — score >= 70%. 6) Functional Equivalence — migrated code produces same outputs as legacy.

76 steps across 6 phases: Analysis (A00-A10), Design (D01-D10), Backend Generation (G01-G20), UI Transform (U01-U10), Quality (Q00-Q13), and Deployment (DP01-DP08). Each step is deterministic and traceable. Every transformation is logged for audit compliance.

Yes. The Platform supports full on-premise deployment behind your firewall with Docker/Kubernetes. Air-gapped deployment is available using Ollama as the local LLM provider.

Claude (primary), GPT-4 (fallback), Gemini (cost-optimized), and Ollama (local/air-gapped). The Platform automatically selects the best provider based on task complexity and cost constraints.

PhoenixCode Lens has a free tier: 10 files, 5,000 LOC, 1 stack (Delphi), 7-day access, no credit card. We assess any Angular app with Leap for free. For the Platform the entry point is the pilot module: one vertical business flow up to 50K LOC for $25,000 fixed, accepted against agreed cases.

Per project, not per month. Pilot module $25,000 fixed; small project (up to 50K LOC) $35,000 – $60,000; medium (50K – 200K LOC) $75,000 – $180,000. Above 200K LOC: a validated discovery, then a fixed price per wave, invoiced 30% at wave start and 70% at acceptance. Every invoice separates platform license, services, third-party licenses and support. Angular upgrades with Leap start at $18,000 fixed per portal.

We do not publish a saving percentage: the ones this page used to carry came from a cost model, not from a measured project. The return is calculated with your own assumptions during the discovery — your maintenance spend, your team cost, the modules in scope — and it goes into the proposal as a scenario, not a promise. What you can check before paying anything: the Lens assessment on your code, and the measured reference run on the platform page.

Yes. Enterprise tier includes unlimited projects and custom SLAs. Contact sales@phoenixcode.ai for multi-project or portfolio-wide migration pricing.

We are pursuing SOC 2 Type II certification. Our Platform includes complete audit trails, role-based access control, encryption at rest and in transit, and secret rotation.

Precisely, because this is the question that decides deals. Lens scans your code, builds the dependency graph and writes the report entirely on your machine — nothing is uploaded for any of that, and the free tier never calls out at all. The one exception is AI rule extraction: to describe a business rule in plain language, the source of the component being analysed is sent to our API, which forwards it to the model. Not your whole codebase, but real source — so if that is not acceptable to your security team, run without AI extraction and everything stays local. The Platform can be deployed on-premise; on our SaaS, code is processed in isolated containers, encrypted, and deleted after migration completes.

We don't claim to satisfy any specific framework — your auditor decides that, and they'll want to see the evidence rather than a logo. What the Platform gives them: every transformation step is recorded with who ran it, when, on which entity, and the values before and after, plus the outcome and any error. That trail is queryable, exportable, and append-only — nothing in the pipeline updates or deletes an entry once written.