How Developers at profiq Use AI

The advent of large language models disrupted many fields. Software development was one of the most impacted. This makes a lot of sense. The overlap between AI and software development is large. AI folks understand software development pretty well, and they also wanted to solve their own problems.

As a result, we can now choose from a plethora of AI-based tools promising productivity improvements in software development. We have:

• In-line code completion tools (GitHub Copilot or Augment Code)
• Chatbots answering our “How to …” questions (ChatGPT, Le Chat, or Claude)
• Editors that let us write prompts instead of code (Cursor)
• UI and app prototyping tools (Bolt.new, v0)
• Autonomous AI software engineers (Devin)

But since these are all very recent developments, we have no established best practices on how and when should we use each tool to increase our productivity and/or learning. Things are also changing fast, and developers struggle to keep up with what’s happening. And while some developers embrace these tools with open arms, others remain skeptical and cautious.

To learn from each other, we conducted a short survey across different teams at profiq, and asked our developers (but not only them) how they are using AI in their day-to-day work. Here are some of their testimonials:

Hana Klingová

Frontend developer, IndyKite

I use Augment Code in my IDE — mainly the chat interface. Especially when I get some errors, I need to write some CSS or tests. Perfect as a rubber duck replacement to brainstorm errors. Augment is also aware of the whole project, so it often points out that we have some config file or that it seems like we are using some specific pattern. It is also quite good at explaining things. (I like to challenge myself and make sure that I understand why and not just how).

I use ChatGPT to check my English and as a personal assistant (fashion advice, what to cook based on what I have at home, when I need to evaluate some decision or compare options, all kinds of stupid questions). But I stopped using it for development related questions as it did not perform well.  But this was at the beginning of the AI era, so it may do better now.

Michael Tichopád

Full-stack developer, Serenity

I use different AI tools depending on what I’m working on. For coding, Cursor has been a game-changer. I just tell it what I need and it handles the scaffolding and edits. It’s pretty reliable — 60% of the time, it works every time.

When I need to write something important like customer replies, release notes or testimonials, I hop over to Anthropic’s workbench where I keep my favorite carefully crafted prompts.

For quick questions, Gemini is right there in my browser and my phone, which makes it super convenient. When I’m bored, it’s fun to mess with Midjourney for images or Suno for audio (getting that hardcore punk version of Stayin’ Alive).

Ondřej Fuchsík

Software engineer, Ping

I use ChatGPT for tasks I’d normally handle with Stack Overflow or Google (though not always). ChatGPT’s responses are more helpful, and it enables me to utilize my expertise by asking follow-up questions. It was particularly useful when I was repairing a Windows Server and needed guidance on PowerShell commands.

I use GitHub Copilot to automate tedious coding tasks. I also sometimes use it to troubleshoot issues in unfamiliar languages or when the problem isn’t immediately clear. Rather than blindly accepting its suggestions, I carefully analyze Copilot’s proposed solutions and ask clarifying questions as needed. Additionally, I find the /explain command incredibly useful when exploring a completely new project, unfamiliar code, or when I’m unsure about a specific method or language feature.

Jan-Kryštof Zahradník

Test automation engineer, Correlium

I avoid coding assistants because I like the feeling of being “in the zone”. Waiting for and reading through in-editor suggestions takes me out of it. I think they are great for some people, but not me. I like to have a lot of control over what code I write, and I want to know what it is doing. I also think that you can substitute a lot of the speed gains that code assistants provide by using custom snippets and knowing your development environment well.

However, I use ChatGPT and similar chatbots, for example, to create Regular expressions or to give me ideas on how to implement certain things. When the output is not ideal, I tend to look for the answers elsewhere (Stack Overflow, official documentation). This gives me the feeling of retaining control while still having some assistance.

I also use it to structure written text. Whenever I write down my thoughts, they tend to sound a little chaotic. AI helps me to make the text a little more coherent and sound better overall.

Viktor Nawrath

Software engineering manager, Tech Research/Acuvity/Tandem

I try to stay up to date with all new models and tools, but it’s getting harder and harder.

My daily driver is Cursor, where I mostly use Claude 3.5/3.7 Sonnet. I love Cursor’s Composer feature, especially in the “agent” mode. I prefer this style of AI coding over “tab completion”, and the models are getting much better at it. To improve reliability, I instruct the model to create markdown files with documentation, implementation plans and other helpful information.

In production codebases, the AI-generated code still needs a thorough review. But I also try to find use-cases where this is not necessary — PoCs, R&D projects and internal tooling I would never even dream of building before, as it would just take too much time. I love using prototyping tools like bolt.new, v0 or lovable to work on such projects.

And last but not least, I’m using ChatGPT and Claude as educational tools. It’s like having a personal tutor to guide me along the way.

Rostislav Nevoral

Student, Student pool

For people just starting their careers, AI tools are both useful and dangerous. They are excellent at helping you understand miscellaneous concepts in programming, and at performing many tasks in little to no time.

But many people that don’t have much experience with programming abuse AI to simply make lives easier for themselves. I did this too at first, and it truly did not help me much in the long term. Yes, you will complete the task, but what do I get out of it?

Nowadays, I mostly use ChatGPT to understand parts of code — how it works, why it works that way, how I can make it better, why an error occurred etc.

AI is an important and useful tool for experienced developers. But it can be detrimental for students like me that need to learn.

Kateřina Hronová

Marketing Coordinator

I use ChatGPT to create proposals for marketing content plans, draft social media posts, and generate slogan ideas. It also helps me brainstorm activities for company events and research various companies’ products and services. However, the results always require review, editing, or expansion. Sometimes, I use AI-generated content purely for inspiration and then write everything from scratch. ChatGPT is also an excellent tool for translation and proofreading.

Additionally, I’m experimenting with Gemini’s reply drafts in Gmail and LinkedIn comment suggestions. For visuals, I use DALL-E and Canva’s AI tool to generate illustrations for campaigns.

I’ve also used the ChatGPT mobile app to transcribe recorded audio into text, making it easier to work with transcripts. The app serves as a great tutor for learning English as well.

As you can see, people at profiq have diverse perspectives on how to best use AI in their daily work. Some, like Viktor or Michael, are more open to most recent AI tools, while others, like Jan-Krystof or Rostislav, are more reluctant. Kateřina’s testimonial also demonstrates that AI tools can be very helpful outside of software development, too.

And what about you? Have you fully embraced AI, are you staying cautious?