Software technical audit before you decide
A software technical audit is an independent review of a platform's code, infrastructure, security and access, done so you can decide on evidence: keep it, refactor it, migrate it, rebuild it or switch vendor.
You need one when a decision is due and nobody in the company can read the code: you inherited the platform from a vendor who left, you have a quote you cannot assess, or the app you generated with AI breaks with real users. We hand you a report with every finding prioritised and a recommendation.
Five situations where a second opinion saves the expensive mistake
It is usually commissioned by a company with no CTO or senior engineer on staff that has to decide about software it did not write. Also by founders and investors who need technical due diligence before they sign.
I inherited a platform and do not know what is inside
The vendor who built it is gone, or the freelancer stopped answering. Before reading any code we find out whose name the repository, domain, cloud account and production keys are in. Then we check whether what runs on the server matches the repository, and what it takes to deploy a change without breaking anything.
I have a quote and do not know if it is reasonable
We read the proposal against what has to be built: what is in scope, what is quietly left out (data migration, testing, deployment, maintenance), which technology it proposes and whether it fits what you already have. We tell you what is missing in writing and what to ask before you sign.
The app we generated with AI does not survive production
Lovable, Bolt, v0 or Cursor produce a prototype that works in the demo. The failures show up with real users, and almost always in the same places. We review what can be kept and what has to be rewritten before you open the app to more people.
I want to switch vendor without losing the code
Before you announce the switch, you want everything needed to carry on without the current vendor in your hands: the repository with its history, the production credentials, the cloud account and how it gets deployed. We draw up the list and check that what is handed over builds and deploys from scratch.
Rebuild from scratch or refactor?
Rebuilding looks cleaner and usually takes longer than planned, because the old system holds business rules nobody documented. Refactoring makes sense if the architecture holds up and the problems are contained. We measure how much code can be kept, which dependencies have lost support and whether there are tests to protect the changes.
Eight areas, from the repository to the contract
The code is one part. The other is everything the software needs to keep running if the team maintaining it changes tomorrow.
Access and ownership
Whose name the repository, domain, cloud account, app stores and payment services are in. Who has access today, including people who no longer work with you.
Architecture and dependencies
How the pieces are split, which frameworks and versions they use, which have lost support and which libraries have gone years without an update. Where the business logic lives.
Security
Secrets in the repository or its history, authentication and permissions, dependencies with known CVEs, and unpatched PHP, Node.js or NGINX versions.
Data and backups
Where the data lives, whether there are backups, when a restore was last tested and whether schema changes are versioned or were made by hand.
Performance and infrastructure
Slow queries, oversized or maxed-out resources, and which part of the cloud bill can be cut without touching the product.
Testing, CI and deployment
What the tests cover, whether deploying is one command or a ritual only one person knows, and how long a rollback takes when something fails.
Documentation
Whether an outsider could run the project locally and deploy it from what is written down. Missing documentation goes into the report as one more finding.
Legal, GDPR and licences
Where personal data is stored and in which country, which providers process it, and whether any third-party licence clashes with how you use it. We flag the technical side; the legal assessment is your lawyer’s.
If something turns up that cannot wait for the report, we tell you as soon as we see it and handle it separately: PHP security patch, Node.js patch, exposed credentials rotation or, if there has already been a breach, hacked website response.
Need to decide what to do with your software?
Get a second opinion →From access to decision
The diagnosis runs on read permissions. Handover is a separate phase, and it only starts if you ask for it.
- 01
Read-only access
The repository and, where possible, the cloud console and logs. For the diagnosis we do not touch production or ask for write permissions.
- 02
Analysis
We run the project locally, read the code and configuration, review dependencies and talk to the people who use it and to whoever maintained it, if they can still be reached.
- 03
Prioritised findings
Each finding gets three ratings: business impact, risk of leaving it as it is, and effort to fix it. That keeps the urgent apart from what can wait.
- 04
Report and decision
Keep, refactor, migrate or rebuild, with the reasons and a phased roadmap. We present it in a meeting with whoever makes the call.
- 05
Handover, if you want it
We can start on the high-risk findings, or prepare the handover to your team or to another vendor. The report works the same in all three cases.
A report to decide with, not to file away
It is written for two readers: the decision-maker, who needs the risk and the recommendation on one page, and the team that will do the work, whether yours, ours or someone else’s.
- A one-page summary for the decision-maker: what there is, what risk it carries and what we recommend
- Inventory of access and owners: repository, domain, cloud, app stores and payment services
- Findings table with impact, risk and effort
- Recommendation (keep, refactor, migrate or rebuild) with its reasons
- Phased roadmap, starting with what cannot wait
- List of the documentation another team would need to carry on
- H-01 Payment gateway key in the git historySecurityHighHighLow
- H-02 Repository in the former vendor’s personal accountOwnershipHighHighLow
- H-03 Framework on a version with no security patchesArchitectureMedHighHigh
- H-04 Daily backups nobody has ever restoredDataHighMedLow
- H-05 Manual FTP deploy that only one person knowsDeploymentMedMedMed
Fictional example: these findings do not belong to any client.
Rebuild, refactor or keep?
Four questions we ask in the first conversation. With what you know today they give a first direction; the firm answer comes from reading the code.
If you are unsure about any of them, that already tells you something: it is exactly what the audit answers.
Your case in four questions
Question 1 of 4Do you have the complete source code and access to production?
What changes when the decision comes with a report
Before the project starts is the cheapest moment to discover that the repository is missing or the database has no backup.
- You compare quotes that use the same words for different things
- The new vendor finds out mid-project that access or parts of the code are missing
- A platform that needed two fixes gets rebuilt, or one that needed rebuilding gets patched
- Security holes surface when someone exploits them
- Every proposal is compared against the same list of findings
- Access and code verified before they change hands
- The decision to rebuild or refactor has its reasons in writing
- The urgent items are fixed before the big project starts
An audit that does not create another dependency
Checking who controls the software would make no sense if the outcome were depending on us.
Code and accounts in your name
If a repository, domain or cloud account turns up in someone else’s name, the first recommendation is to move it to an account of yours. Anything we build afterwards also goes in your accounts.
Documentation anyone can use
The report and any documentation we produce are written for another team to use. They cover how to run the project, how to deploy it and what not to touch without tests.
Confidentiality
We sign the NDA before receiving any access. The report is yours and is not shared with third parties without your permission.
No lock-in
The audit does not tie you to hiring us for development. You can take the report to your own team or to another vendor.
Long-running platforms, and one built with AI
We have modernised platforms that had been in production for many years, and we built our own website with AI agents. We know what AI leaves undone because we had to do it by hand.
Soamee.com
Our own website, written with AI agents and 22 automated checks that block half-finished work. The case study also covers what did not work.
22 automated checkseEvidence
Modernisation of a platform with more than 20 years generating electronic evidence, where security and eIDAS compliance are the product.
eIDAS complianceInfoAdex
Modernisation of the tools of a platform with more than 55 million records and 30 years of advertising market data.
55M+ recordsYou might also be interested in
Software audit FAQ
Do I have to give you access to the code?
What if the current vendor will not cooperate?
Do you review apps built with Lovable, Bolt, v0 or Cursor?
What happens after the report?
Do you sign an NDA?
Is this a security audit or a pentest?
Does it work as technical due diligence before investing or acquiring?
How much does it cost and how long does it take?
Tell us what you have and what you need to decide
In the free 30-minute consultation we look at what the software is, what decision you face and what would need reviewing. Then we tell you whether an audit makes sense and what it would include.
Tell us your challenge. We'll propose a solution.
No commitment. We reply within 24 hours, and within 48 hours you have a proposal with scope, timeline and budget. No fine print.