AI & Vibe Code Software Assessments
A little less guesswork. A little more confidence.
Unless you've been living under a rock, you'll know that AI is changing the way we work in ways that were completely unimaginable just a few short years ago. One area where AI is having a significant impact is software development, much of it being positive, some of it being negative.
If you've found yourself here, chances are you've been experimenting with AI for software development yourself or you have ended up with a piece of software that has been built using AI. We'll also take a punt and assume you're now wondering whether your new software is actually fit for purpose, ready for the real world, secure and reliable.
Let us help you with that.
Well, it doesn't involve AI for starters. Our very human development team will take a good look at your software and determine how it's been built, how well it works and whether it's suitable for whatever you intend to use it for.
Depending on the software itself, this might include:
We'll assess whether the code is organised, understandable and maintainable.
Possible findings:
All good, no remediation needed.
The same code is repeated in multiple places, making updates unnecessarily time-consuming. The code is overly complicated for what it needs to do. Functions are too long or try to manage too many tasks at once. Different parts of the application have been built in inconsistent ways, making future development more difficult.
We'll check that everything works as it's supposed to work, including in less obvious scenarios and edge cases.
Possible findings:
Everything works exactly as intended under various different scenarios.
A form accepts invalid information. Calculations produce incorrect results when certain values are entered. Users can bypass a required step in a workflow. The application fails when someone enters unexpected data or follows an unusual sequence of actions.
We'll take a look for vulnerabilities, exposed credentials, weak access controls and other potential risks depending on the software and its intended purpose.
Possible findings:
The software looks to be secure.
API keys or database credentials are exposed in the code. Users can access information belonging to other users by changing a URL. Administrative functions aren't properly restricted. Input fields allow malicious data to be submitted without adequate validation.
We'll determine whether your data is stored, processed and protected appropriately.
Possible findings:
All checks out, your data set up is fine.
Sensitive information is stored without appropriate protection. Records are accidentally overwritten or duplicated. Data is deleted without a reliable recovery mechanism. Personal information is collected or retained unnecessarily.
We'll assess whether the software can cope with realistic usage and identify any potential points of failure.
Possible findings:
The software performs as expected under the conditions tested.
Pages take too long to load when processing larger amounts of data. The application becomes unresponsive when multiple users access it simultaneously. Repeated database queries slow down common activities. An unexpected error causes the application to crash rather than recover gracefully.
We'll review any connections to databases, APIs and third party software work as they are meant to.
Possible findings:
The integrations all work in the way they need to.
An integration stops working when a third-party API changes. Failed requests aren't retried or reported properly. Information isn't synchronised correctly between systems. The application breaks when an external service is temporarily unavailable.
We'll check whether your software is set up in a way that makes future enhancements, additions and changes straightforward.
Possible findings:
The software is set up for success.
Adding a seemingly simple feature requires changes across multiple parts of the application. The software has outdated or incompatible dependencies. There are no automated tests to help identify regressions. The application has been built around assumptions that make new functionality difficult to introduce without breaking existing features.
Following on from the assessment, we'll provide you with a detailed report that covers the key areas listed above, highlighting any risks and opportunities to improve the software. Depending on our findings, there are a few possible routes forward:
If it turns out the software is well built and performs as expected, you can move forward with peace of mind knowing the software has been thoroughly assessed by a team of professional developers.
If we've identified issues that can be resolved, we'll provide you with any options, associated costs and estimated timescales. You may wish to have another development company fix the problems instead, which is perfectly fine.
Sometimes, the underlying architecture or code is simply not fit for purpose and the time and costs to make it so, would probably be higher than rebuilding the whole thing. If that happens to be the case, we'll explain why rebuilding is preferable and should you wish to go ahead, we can do that for you.
Ultimately, only you can answer that question. If you like living on the wild side and have a penchant for risk taking, you may choose not to have your AI software assessed. This can be fine in some circumstances such as if the software is only being used for experimentation, you've built a simple proof of concept, it isn't managing anything particularly important and you're not going to become reliant on it or it isn't customer facing.
However, if you prefer to err on the side of caution and you're planning to use the software to run important business processes, store customer information, manage sensitive data or connect to other business-critical systems, it's definitely worth taking a closer look before you put it to work.
Just because your software appears to work, this doesn't mean it's secure, reliable or ready to cope with real world demands. This is a common trap people fall into when using AI to generate their software and it can be an incredibly costly mistake. Problems aren't always obvious and some may not surface until something goes wrong.
An assessment can help you understand the risks, identify anything that needs attention and decide whether your software is ready for what you have planned.
If you're unsure whether your AI generated software is fit for purpose, let us take a look. You can show us what you've built, tell us what you want it to do and how you plan to use it. We'll discuss your requirements, explain what your unique assessment will entail and help you decide on the best way forward. We don't use scare tactics nor do we flog unnecessary work, we just so happen to know software and can provde you with honest and transparent advice on how to get yours to where you need it to be.
Good software starts with good decisions. Let's make one.
AI & Vibe Coded Software - The lowdown
AI Assisted Software Development
Q) How much does an assessment cost?
A) The cost depends on the size, complexity and nature of your application as well as the depth of assessment required. We'll discuss your software and what you need from the assessment before providing a price.
Q) Can you assess software built using tools such as Lovable, Bolt or Replit?
A) Yes. We can assess software created using AI powered development tools, whether you've built it yourself or had someone else build it for you. We'll look at the resulting application and its underlying code rather than judging it by the tool used to create it.
Q) Can you fix any problems you find during the assessment?
A) Potentially, yes. Depending on our findings, we may be able to resolve issues within the existing code. If the underlying architecture is unsuitable or the software would be more costly to fix than rebuild, we'll explain your options and recommend a sensible way forward.