Legacy System Reverse Engineering
Recover how your legacy system works, even without the source code
We combine AI-assisted code and binary analysis with narrated recordings of the people who use the system. You get a verified account of its business rules, workflows, and data, ready to modernise, rebuild, or replace. We work only on software your organisation owns or holds the rights to examine.
When nobody can say what the system does
Modernisation stalls before it starts when the knowledge of how a system works has gone.
The source code is lost or will not build
The repository went with a previous supplier, the build environment no longer exists, or you only ever received compiled binaries. Nobody can prove what the deployed system was built from.
The people who built it have gone
The original developers have left, the agency has closed, or the vendor no longer supports the product. The system still runs a critical process, and nobody can say exactly how.
The rules live in people and spreadsheets
Calculations, validations, and workarounds have built up over years. Some sit in the code, some in the heads of experienced users, and some in spreadsheets that correct for things the system gets wrong.
Reverse engineering capabilities
AI reads code and screens quickly. People confirm what it means. We use both.
AI-assisted code and binary analysis
Narrated workshop recordings
Business rules recovery
Data model recovery
Rights and data protection checks
Human verification throughout
What we will and will not reverse engineer
The rights your organisation holds in the software decide what we can do, so we check them first on every engagement.
Software you own or licence to modify
If your organisation owns the copyright, or holds a licence that lets you copy and modify the code and have a supplier do so for you, the system is yours to examine. Decompiling and analysing it is part of recovering how it works.
Is REA legal? Reverse engineering with AIA vendor product you run under licence
Where you hold only a licence to use a vendor's product, UK law still lets you study how it behaves and decompile what is needed for interoperability. We use that to document the system and get your data out, not to copy the product.
Lawful reverse engineering from recordingsA SaaS product
We do not decompile or recreate a SaaS vendor's code. A SaaS replacement starts from a functional specification written in your own words, from your use of the product and your business knowledge.
Is SaaS reverse engineering legal?Talk Think Do is a software development consultancy, not a law firm. Our guides set out how we understand the law, and they are not legal advice. Where the position on a system is unclear, we recommend taking advice from a solicitor qualified in intellectual property law before analysis begins.
From rights check to verified specification
Code analysis and recorded workshops run side by side, and nothing reaches the specification without verification.
Outputs you can build from
Every artefact records the evidence behind it, so the replacement team can check it rather than trust it.
Pain-point register
What does not work today, who it affects, and how often. This is what makes the replacement better than the original rather than a copy of it.
Legacy information architecture
A map of the screens, navigation, and workflows of the current system, organised by functional area.
Conceptual data model
The entities and relationships the system holds, recovered from the database, the screens, and the people who use them.
Business rules register
Each calculation, validation, and edge case, with the evidence behind it, a confidence grade, and a list of what has not yet been found.
Requirements catalogue
Requirements for the replacement, each cross-referenced to the pain points it addresses and the rules it must preserve.
Interactive prototype
A working prototype of the replacement, so stakeholders can test the recovered understanding by using it rather than reading about it.
How an engagement runs
The rights check comes first on every engagement, before any code or recording is analysed.
Rights and scope check
We read the original development agreement or licence with you and confirm what rights your organisation holds in the system. We identify third-party components, agree what may be analysed and by which tools, and assess the personal data involved.
Evidence capture
We collect what exists: binaries, databases, outputs, logs, and any surviving documentation. We run narrated workshops with the people who use the system, ideally in a training or masked environment.
AI-assisted analysis
Coding agents connected to decompilers and analysis tools work through the code and data, while a separate pipeline indexes the workshop recordings. The two are brought together by functional area.
Verification
Recovered rules are tested with worked examples and controlled probing in a safe environment, and checked with the people who know the system. Anything we could not confirm is recorded as unresolved, not guessed.
Specification and prototype
You receive the registers, the data model, the requirements catalogue, and an interactive prototype. From there you can modernise, rebuild, or replace with an accurate picture of what the current system does.
Related reading
Reverse Engineering a Legacy System That Has No Documentation
Recovering Business Rules When You Do Not Have the Source Code
Is REA Legal? AI Reverse Engineering, SaaS, and UK Law Explained
Is It Lawful to Reverse Engineer a System From Recordings and Screenshots?
What You Can (and Can't) Legally Copy When Replacing SaaS
Do You Actually Own Your Software? A Control Checklist
Modernise, Rebuild, or Replace: A Decision Framework
Go deeper
Legacy Application Modernisation
Our full legacy modernisation service, from assessment through re-platforming and phased replacement.
Data Migration
Automated nightly migration of the data model this work recovers, with anonymised test data.
Internal Systems Handover
Capturing knowledge of a system before the person who understands it leaves.
Legacy Application Support
Keeping the current system running safely while its replacement is planned and built.
Seen enough? Let's talk through your requirements.
Book a free consultationFrequently asked questions
Can you reverse engineer a legacy system without the source code?
Yes. Where your organisation holds the rights, we use AI coding agents connected to decompilers to analyse the compiled application and its database. We combine that with narrated recordings of people using the system and controlled tests in a safe environment, because code alone does not show which rules matter or what users work around.
Is it legal to decompile our own legacy system?
Generally yes, if your organisation owns the copyright or holds a licence that lets you copy and modify the code and have a supplier do so. Commissioning software does not transfer ownership by default, so we check the original contract with you first. Our guide on whether it is legal to reverse engineer software with AI explains the position.
What if the system is a vendor product we only licence?
We can still help you understand it and get your data out. UK law lets a lawful user study how a program behaves and decompile what is needed for interoperability, such as reading a data format for migration. It does not allow copying the vendor's product, so a replacement is built from a description of what the system does.
Do you use REA (Reverse Engineer Anything)?
REA is one of several ways to connect a coding agent to established tools such as Ghidra and IDA Pro, and we choose tooling per engagement. The legal question is the same whichever tool is used: whose software it is, and what you hold the rights to do with it.
Will you reverse engineer a SaaS product so we can replace it?
No. We do not decompile or recreate a SaaS vendor's code. We replace SaaS products from a functional specification written from your use of the product and your business knowledge, which our SaaS replacement legal guide sets out in detail.
How do you handle personal data in recordings and databases?
We minimise it before analysis starts, preferring training or masked environments, and crop recordings to the application so participants' faces never enter the dataset. We document the lawful basis, assess any AI service through the supplier management process in our ISO 27001:2022-certified ISMS, and set retention before anything is recorded.
Can we trust what the AI recovers?
Only once it has been checked. A model reading code and screens will produce a fluent account that is partly invented. Every rule is verified against a worked example or a person who knows the system, and given a confidence grade. Anything we could not confirm is listed as unresolved.
See how AI is changing how we build software
Our quarterly AI Velocity Report tracks real delivery metrics from live projects: how much of our code is AI-authored, how delivery timelines compare to baseline, and which tools are making a measurable difference. No marketing spin, just honest data from a team that builds software every day.
Read the latest AI Velocity ReportNeed to understand a system before you replace it?
Tell us what you have: source, binaries, a database, or just the people who use it. We will start with the rights check and tell you plainly what can be recovered.
Book a free consultationor call 01202 006729