100 % of our software developers use AI — and it shows in our clients’ results
AI has become a natural part of everyday software development. It gives developers new ways to work more efficiently, but it also demands an understanding of what it can do and how to use it responsibly. At Identio, we use AI in a range of ways: to support software development, to build clients’ own solutions, and as part of everyday problem-solving.
For us, using AI doesn’t mean taking shortcuts — it means finding smarter ways to do even higher-quality work for the client. We’ve created our own AI guidelines and, in autumn 2025, ran an internal survey on how our developers use AI in their client projects.
In this article, I’ll look at how Identio’s consultants use AI in their work, how we ensure responsible use, and what benefits it brings to our clients.
AI as part of software development
With AI already a fixed part of everyday development work, we’ve wanted to make sure that Identio has broad, up-to-date expertise around it. Every one of our consultants needs to master a range of AI solutions — but the final way of using them is always defined by the client company and their operating environment. Whatever the situation or purpose, AI has to be used professionally and in line with ethical principles.
AI isn’t just a tool that makes a developer’s day easier. It also delivers concrete benefits to the client: faster delivery, better quality, and lower risk of errors. At the same time, it frees up the developer’s time for higher-level thinking, namely seeing the bigger picture and developing the logic of the solution. These are things AI can’t do – at least not yet.
100% of our consultants master AI tools
What do we mean when we say that 100% of our consultants master AI tools?
The figure is based on a survey we ran and on the conversations we’ve had with every single one of our consultants. In this context, daily use of AI means using AI- and language-model-based tools to support development work. Among the tools our consultants use are GitHub Copilot, ChatGPT, Gemini, Claude, Google AI Studio, Cursor, and Windsurf.
Our clients’ AI guidelines take precedence and bind us too, and sometimes that means development work is done entirely without AI. This is often the case in projects involving sensitive data or where security requirements are high. Because AI is a megatrend and a big part of our daily work, we think it’s important that every one of our consultants masters these tools, regardless of project-specific requirements.
AI will never replace an expert, but with AI-assisted software development we can bring more efficiency to the work and, in doing so, create even more value for our clients.
How our developers use AI
Our internal survey gave us a good overall picture of how our developers use AI in their daily work. Since we don’t define the terms of AI use on our own – the client’s guidelines and the consultant’s own preferences carry more weight — the ways of using it vary widely. That’s why the survey was a good way to map out AI use and get a comprehensive picture of where things stand.
The most common uses for AI include the following:
Code-related tasks Using AI for things like code completion while writing (basic autocomplete), generating test data, and writing SQL queries. AI also helps with fixing bugs and quality assurance. It can go through huge volumes of log files and monitor the state of a system to identify the causes of bugs. More advanced AI tools can also suggest concrete changes to the code to fix them.
Learning and grasping context AI helps explain and summarize new libraries or languages in a way that fits the context. Our consultants also use it to understand new concepts and apply them to an existing codebase, to gather background information for various tasks, or to get summaries or advice that used to come from Google or Stack Overflow. So it helps solve specific problems and reduces the need to hunt for technical information through other channels.
Ideation Exploring alternative implementation approaches and prototyping have become easier with AI. It speeds up finding different solution options considerably, especially in situations where it would be easy to fall back on more familiar patterns.
Drafting and communication AI helps draft and polish documents, create message templates, and condense documentation. As a result, they have more time to focus on what matters most: problem-solving and writing code.
Prompt strategies and custom GPTs We’ve found AI useful for things like iterating on prompts. Custom GPT versions – ones pre-loaded with background information – have also made work considerably more efficient.
A concrete, practical example with language models
Language models predict probabilities based on the inputs they’re given, but there’s always a degree of uncertainty involved. If a model doesn’t know the answer, it tries to infer it – sometimes incorrectly. That’s why we make use of the “tool calling” principle and the Model Context Protocol (MCP), a system for managing and connecting the context information AI models need, securely and in a distributed way.
With MCP, we can, for example, connect a language model to Figma, which lets it read the spacing and dimensions of elements down to the pixel. Once that information is fed to the model, it can automatically code the corresponding view. After that, we can run a testing tool that confirms everything has been done correctly. This way we avoid manual information-gathering and give the model better, more accurate source data — and the end result is both faster and higher quality.
This removes the need to pull information from Figma by hand, and we get the full picture directly through the integration. In other words, we give the AI better source data.
Cursor, for instance, codes excellently in a text-based environment, but it can’t directly interpret the contents of a graphical user interface. In that case it searches for solutions online that may not suit the project’s purpose. With an MCP integration, we can solve this and give the model context straight from the right source.
Why does this matter to our clients?
AI tools don’t make coding consultants better – they make them faster. When a consultant who bills by the hour uses these tools fluently, they save the client significant resources. What’s also important is that AI shifts the emphasis toward thinking. When time is saved, the consultant can focus on higher-level questions: why the solution is being built, what the client really needs, and how the end result best serves its purpose.
So using AI to support software development brings the client at least the following concrete benefits:
- More time to understand the client’s needs
- More time to ensure the solution is fit for purpose
- Shorter development times
- Fewer recurring errors
- Better code quality
- More efficient documentation and communication
Responsible use and risk management
AI is only as good as the developer using it. You can only get good results with it if you’ve mastered the fundamentals of the field along with the rules and principles of programming. Without that expertise, checking AI’s output is impossible. Only an experienced developer can produce high-quality code with the help of AI, which is why it’s a far more powerful tool in a professional’s hands. That’s why we’d argue a senior developer without AI is a better bet than a junior developer using AI.
AI makes a lot possible, but it always comes with certain risks. Alongside our clients’ AI guidelines, we’ve created our own set of guidelines that serve as a general framework for using AI to support our own work. Every consultant knows the rules for how AI can be used, and on the other hand, the things it should never be used for.
For the sake of clarity, I’ll mention some of the basic principles we always follow in AI-assisted software development:
We don’t feed AI confidential or personal information. We know what kind of information we can give to which model, and whether the information we provide is used to train the model. We never feed AI a client’s source code, strategic documents, personal data, or other sensitive content without the client’s explicit permission and a secure environment.
We don’t give AI material protected by copyright or IP rights without permission. We don’t use AI to produce content that violates copyrights, trademarks, or intellectual property rights. All material to be provided is checked for its usage rights.
We never use unchecked AI output as-is. The consultant has to be the more skilled party than the AI. We check, edit, and finalize all AI output before using it.
We never accept or enable illegal, unethical, or harmful activity. We don’t use AI to create spam, malware, scam content, or discriminatory material. Our work is based on responsibility and ethical principles.
The future of AI-assisted software development
We’re currently living through the peak of the AI hype. When the dust settles, only the genuinely good implementations and applications will survive.
In our experience, the pace of model development has already started to slow, and the situation is relatively stable. We keep up with developments continuously, but we don’t track the release of every shiny new AI model too closely, because the differences from previous versions are by now relatively small. The sources of models’ training data are also starting to matter more, now and in the future. In October 2025, Futurism and many other outlets wrote that more than 50% of the material online is AI-generated. Feeding that in as training data doesn’t produce good results.
Whatever the direction, as AI tools evolve, our guidelines and processes are updated continuously too, and we actively develop our expertise around the topic.
More efficiency and proactivity to support your development team?
Even if your team already uses AI, projects still need people with precise expertise and broad experience across different kinds of projects. Identio’s consultants combine the power of AI tools with strong professional skill. We bring independent problem-solving and results to your team from day one.