Exploring AI’s Impact: Developer Insights on Tools, Productivity, and Workflow Optimization

Nowadays, AI is making a huge impact for developers, helping them work faster and smarter. To see how it’s changing their day-to-day, we asked a few devs to share how AI tools—like code completion and task management—are fitting into their workflow. Their answers show how AI helps them solve problems quicker, collaborate better, and boost productivity overall. Here’s what they had to say!

What specific AI tools or software do you use most frequently, and why?

• • KP: ChatGPT. Use it to do mundane tasks.

• • JB: ChatGPT. Specifically the 4o model since I use it a lot for coding. I found the earlier OpenAI models to be unreliable, or too limited in the amount of input they accepted. This model also seems less “lazy”, sometimes even doing too much “work”. The older models would give me partial responses for code which I had to puzzle piece together whereas 4o seems to do too much “work” and keep displaying the entire solution when I actually only need the snippet! Haven’t used much other tools, but interested in trying out Claude, Perplexity and the newer OpenAI models as I’ve heard good things.

• • Tere: ChatGpt – b/c everyone in the company is using it, you can ask anything. The interface is user-friendly.

• • Lydia: Chatgpt. It knows everything.

• • Matt: Currently using chatgpt (but will look at alternatives in the future: maybe a locally hosted LLM)

• • Norwin: ChatGPT, I use it the same way I do Google Search but it presents information better.

• • MJ: chatGPT.

• • Kirstin: ChatGPT and Copilot. They are readily available. And they’re a step-up compared to a Google search, where I have to sift through search results. ChatGPT and Copilot provide answers and sources immediately.

• • LA: I use ChatGPT at times to help me code but only partially. I also use Notion to manage my daily/weekly/monthly routines, plans, notes, etc.. Basically everything that goes through my head, in written/typed format.

Which AI-powered code completion tools (e.g., GitHub Copilot) do you use, and how do they impact your coding speed?

• • KP: I don’t use any AI code auto complete.

• • JB: Don’t use any yet. Mainly been using ChatGPT which does impact my coding speed positively. I’m almost never stuck now. I can offload a coding task to it while I move on to another part of what I’m programming. I do have to come back to the response and incorporate it, but it lets me keep moving forward. I think even just this has had a huge effect on my code production. I was at a point where I almost didn’t produce any code anymore. I wasn’t building anything anymore because I had no time. We all know programming normally requires big blocks of uninterrupted time, time which I didn’t have (and still don’t). But I’ve started coding for 1 hour early in the morning before the rest of the team or my customers start their days, and in that 1 hour with the help of ChatGPT I’m able to produce what I feel used to require a solid 3 or 4 hours of work.

• • Lydia: Chatgpt. It helps to speed up programming and debugging by saving time on searching online for the matching articles.

• • Matt: I just use Chatgpt for programming. But it has increased my ability to understand old code quickly (by asking for a line-by-line explanation)

• • Norwin: Our IDEs already had good code completion and suggestion tools even before the AI boom. These features work very well.

• • MJ:  None. And I don’t want to rely on them. A good LSP is enough.

• • Kirstin: I don’t use code completion tools yet

• • LA: I just use the normal IDE features. Decent speed, but would like to learn more.

• • Marben: GitHub Copilot helps suggests code snippets and speeds up writing test scripts, improving overall efficiency.

In what ways has AI improved your ability to manage and prioritize development tasks?

• • KP: AI has improved my ability to manage tasks that require pattern based text changes.

• • JB: I don’t use it to directly manage or prioritize dev tasks. Instead, it’s given me a different view when I’m thinking about tasks for myself and my team. With these AI tools, I have a lot more confidence that a given task can be completed in a short amount of time. I used to be face with a task that I estimated would take a month of dedicated work and procrastinate on it. Now I think less, and spend an hour or two of side-by-side coding with AI to come out with a functional proof of concept that often is good enough to be the start of a whole new feature or module. I do use a tool called ChatPRD which helps me make better thought-out, structured product requirement documentation for our developers. Its useful when taking a feature request or idea, and turning it into a requirement for the development team to work on. Since we don’t have dedicated product managers yet, it’s been a valuable tool to help sprinkle in some product manager-esque ways of thinking and questioning. We use ChatPRD for any substantial feature we’re about to do, to help us all get more on the same page.

• • Tere: It can easily offer solutions without needing to visit other websites, allowing me to save time for other activities.

• • Lydia: N/A. I don’t use it for prioritization.

• • Matt: helps me get up to speed on new technologies much faster

• • Norwin: Easier to do QA because you can feed ChatGPT code and therefore ask contextual questions. With Google / StackOverflow, you would have to search more generically.

• • MJ:

• • Kirstin: Haven’t applied this yet

• • LA: I haven’t touched much on AI besides when trying to know stuff or code help/snippets.

• • Marben: AI helps me prioritize testing tasks by analyzing deadlines and dependencies, making test execution smoother.

How does AI assist you in generating or optimizing algorithms and logic in your code?

• • KP:  I dont use it much for in production code (I prefer documentation).

• • JB: On many occasions now, I’ve coded a quick and ugly (i.e. inefficient) version of something, made sure it worked correctly, then asked ChatGPT to optimize it. Sometimes it produces code that ends up with incorrect results, but a little back and forth with it and we end up in a position better than before. I also often have it write me little helper functions which I used to normally have to write by hand. I’ll just explain what I need in plain english, give it some context and let it work while I move onto the next part of my project. As mentioned before, this is really how I’m able to speed up my code production, by offloading these tasks and continuing on with my work. 

• • Lydia: When I ask it a generic question, it gives out several solutions that helps with designing the right one. When I ask it a more specific question, it is able to analyze and gives out an improvement.

• • Norwin: N/A as I don’t directly dev and therefore don’t implement algos, but ChatGPT has also been helpful with asking coding best practices. Like “Is it necessary to call Dispose() on an object if it’s within a loop” or “How can you implement logging for web applications given there is no console output”

• • MJ: Sometimes I provide it a snippet and ask if there is something more optimal than this or to provide alternative solutions. Then I pick the one that makes the most sense. most of the time it just rearranges my code though

• • Kirstin: Coming from different programming languages, there are times where I know how to implement something in – let’s say – Typescript but not in C#. I type the logic in Typescript and then ask AI to convert it to C#.

• • LA: Better readability and efficiency, also explains why something is done the way it is.

• • Marben: AI Suggests better ways to write test scenarios or improve test case logic for better coverage.

Can you describe how AI has improved the accuracy and efficiency of your tasks?

• • KP: it has improved efficiency for simpler tasks.

• • JB: At our current levels of AI and LLMs, I can’t actually say it’s helped with accuracy. Often times the results from AI end up incorrect. Or they only capture cases you’ve explicitly instructed for. Maybe it’s operator error, and not giving enough instruction. I’m waiting for the day when these AI tools can think ahead and consider edge cases, and extrapolate past what was explicitly instructed.

• • Tere: It helped me identify errors more quickly and offers alternative methods for addressing problems.

• • Lydia: It provides summarized and to the point solution. In best scenario, the solution it gives runs without modifications and solves the problem in the development environment. It doesn’t always give out a working solution though. 

• • Norwin: Accuracy I think not so much; you still have to vet ChatGPT answers. Efficiency definitely, as they way it is able to find and present information is much faster and precise/direct than Google.

• • MJ: it’s good at breaking down concepts/ideas into digestible chunks of information. it’s good at solving known problems/problems with an existing solution. if something is already known (like you feed it a dataset or the information exists somewhere on the internet), ai seems magical.

• • Kirstin:  AI has helped my learning significantly (in new concepts to me, like accounting). It has made accounting concepts easier to understand (Example prompt: “What is accounts receivable and how is it connected to orders? Explain like I’m five”)

• • LA:  It helps lessen time in coding when the code required is not large.

• Marben: AI assists in quickly identifying bugs or discrepancies in code, which improves the accuracy of my tests and reduces manual effort.

How has AI improved your ability to troubleshoot and debug issues in your code?

• • KP: I use a debugger and StackOverflow

• • JB: On first thought, it hasn’t helped much in troubleshooting or debugging code. I think this is because it lacks context on your environment and what you’re trying to do. It can look at code and see that there’s no syntax errors, and that logically it should work, but it doesn’t consider edge cases. But if you give it enough information (e.g. conditions you’re running the code in, values you are giving it, the results you’re getting vs what you’re expecting, etc.) I’ve found that it does help. So for me, it does help with fixing code if you already know what the issue might be.

• • Lydia: It analyzes the error message and gives working solutions based on earlier encounters and extensive knowledge base, which is better than me guessing the cause. 

• • Norwin: Tremendously, since as previously mentioned, the ability to feed it code / snippets and ask contextual questions is a huge boost to traditional web search.

• • MJ: when i know that something is wrong in an area of a code snippet, but i don’t know where exactly, i can ask the AI to pinpoint where or to give me ideas what might be wrong with the code. but i try to use this sparringly, i feel like my problem solving skills deteriorate whenever i use AI coz i’m not the one solving the problem. it can be an easy way out sometimes, like yes, the problem is solved, but on the cost of my growth as a dev.

• • Kirstin: With code snippets where I’m expecting a result but getting a bug instead, AI made finding these bugs quicker. Traditionally, we have to manually sift through lines of code to find bugs. But now AI can perform root cause analysis and explain where things went wrong and how to improve upon it.

• • LA: Explanations from AI, when it is not mentioned, I ask why.

• • Marben: AI tools quickly analyze logs and provide insights on where the issue might be in the code, helping in faster bug identification.

Can you give an example of how AI helped you solve a technical problem faster than traditional methods?

• • KP: N/A I prefer books. for ex: I use the index of “The Algorithm Design Manual” book as a lookup for algorithms.

• • JB: Recently we added a feature that taps into a web browser’s webcam. Normally I’d google and have to sift through 5 to 10 articles or sets of documentation. With AI, I posed the problem and gave it the context of our application (e.g. programming language, web framework, what we needed to do). With a little back and forth, it produced a solution tailored to our existing codebase. I also went back and forth with it assessing the technical method, something that I would normally take some time to research. Of course I still did my own research, but AI helped me whittle down the options.

• • Lydia: For example, when I encounter an error on resolving nuget package. The traditional way will be me to google the error message, read top posts (stackoverflow or alike threads), identify the relevance, and try out the suggestions. Whereas, using chatgpt, I feeds it the same error message, it will list out the relevant possible solutions in one place. Sometimes, the first solution it gives works with luck.

• • Norwin: There was an issue with a program erroring out during file-handling and I suspected that it was when a file already existed. I asked ChatGPT a question like, “Will File.Write() in C# error if a file already exists”. It not only gave the correct answer, but also noted that the behavior changed in a more recent version of the language. This could’ve been found by Googling or reading C# docs, but it would’ve been slower.

• • MJ:  When I can describe a solution, but don’t know exactly how that might look with the tech I’m currently using to solve the problem. But I try to use this sparingly, since often times I get code that works but I don’t understand why it works, like it looks correct. this is fine, if the goal was to solve the problem and call it a day, but the problem with software development, is that any code you write now will be a liability in the future. and because I’m not the one who wrote the code, this will bite me in the ass in the future. like with ai you more problems faster but with the cost of significantly longer debugging sessions in the future.

• • Kirstin: There was a project where the program had to calculate the due date of a payment while skipping weekends and statutory holidays. I asked AI to give me a solution for C# and it worked when tested.

• • LA: I just ask away instead of looking for answers here and there, but still, stackoverflow remains unparalleled IMO

• • Marben: AI provides immediate suggestions and solutions when encountering issues in testing, speeding up the problem-solving process.

In what ways does AI enhance collaboration with other developers, such as in code reviews or shared projects?

• • KP: Maybe have the AI pitch in for code review of code.

• • Lydia: If I want to continue the exploration of a topic with other devs, I can share conversations between me and chatgpt with others as a link. Other devs can continue to ask chatgpt questions using the same thread.

• • Norwin: You can review AI-generated code and information and use it as a starting point for discussing/brainstorming. There is also some comfort in that what you get back from GPT is the most common/popular answer (by virtue of how GPT models work i.e. probability), so you know that what you’re dealing with isn’t something completely random and there is at least some visibility bias (I guess the good kind? Lol) to the answers.

• • MJ: because of the language/cultural barrier on working with multi-national teams, code reviews might be misunderstood as an attack. but with ai, i can ask it to tell me what this code review meant, like what is the author trying to say objectively or to provide me a different perspective or interpretation which helped me a ton.

• • Kirstin: Not applicable yet (haven’t had a shared project)

• • LA: I’m not quite sure who AI will come into play with collaborations yet.

• • Marben: AI enhances collaboration by offering quick insight during test reviews, and helping share feedback with developers more effectively.

What specific AI features have made the biggest impact on your productivity and workflow improvement?

• • KP: AI summarization is nice to summarize large articles that I don’t have time to read

• • JB: See all my input above. It’s been taking what I used to think were hard problems that would take a lot of effort, to being able to whip out quick proof of concepts in about an hour. I’ve used the new ChatGPT “o1” model, to take our technical feature list and help turn it into marketing materials. It does a great job “thinking” about how to take a feature and present it with the end business benefits

• • Lydia: I don’t actually use AI for workflow improvement, but I do think chatgpt helps to improve my development speed with its chatting feature.

• • Matt: but regardless, we still need to ask the right questions and carefully monitor the output to get the results we want.

• • Norwin: Just GPT in itself and how it works (the generative aspect) and how it helps better collect, format, and present information from existing knowledge.

• • MJ:  chat feature, save chats, AI learning from previous chat

• • Kirstin: The AI features that have made the biggest impact on my productivity and workflow would be its capability to do root cause analysis, enhanced code reviews and optimization, and its ability to make learning faster when it comes to unfamiliar technology.

• • LA: Simply ask and an answer shall be given to you with much desired accuracy as long as you’re using the standard(talking about ChatGPT).

• • Marben: The most helpful AI features are automated test generation, error detection, and task prioritization.

---

So, there you have it—AI is quickly becoming a must-have for developers, helping them code faster, troubleshoot better, and keep everything running smoothly. Whether it’s automating routine tasks, speeding up coding with tools like GitHub Copilot, or finding clever ways to debug, AI is definitely a game-changer.

At the end of the day, it’s all about working smarter, not harder. And from what these developers shared, AI is making their workflows not only more efficient but way more enjoyable, too. If you haven’t hopped on the AI train yet, you might want to start exploring what it can do for you!