How JobLog Built a User-Focused Job Application Tool that Prioritizes Accuracy

It was pretty simple. I was looking for work and getting very few responses, so I decided to focus on sending more applications. But filling out the same forms again and again took a lot of time. When an employer replied, I sometimes could not remember which position they were contacting me about. Repeating this process without getting answers was discouraging. After only a few days, I felt frustrated and wondered whether I was doing something wrong. I also started to feel that many of the listings might not even be real. I tried different apps, browser extensions, and AI tools, but none felt right. Some made mistakes, some were too slow, and others were paid. Eventually, I stopped trying to combine tools I disliked and decided to build the tools I actually wanted for myself. I built JobLog and improved its accuracy, then added old-listing signals through JobLog Filter.

After building the first version of JobLog, I expected my website, reviews, and Reddit posts to bring users. That approach mostly failed. Direct links were blocked in many communities, and I had no marketing experience, so my posts were not very good. I tried getting help from AI, but the copy felt dramatic and artificial. I published one of those posts and received fair criticism immediately. Things improved when software directories started listing JobLog. I began using those platforms myself, and website traffic increased. Then I applied something I learned from working as a trainer, do a little every day instead of expecting one big result. My TikTok demo videos now have more than 10,000 views, and installations are increasing daily. People who install the extensions rarely remove them, so growth is beginning to compound organically. JobLog and JobLog Filter also link to each other, helping both products grow together. Some traffic sources have not produced results yet, but I see them as additional streams that may support future growth and help me improve the products as demand develops.

Most of the problems were not reported by users. I found them myself because I actively use both products during my own job search. There were accuracy problems with autofill, issues with how job cards were highlighted on LinkedIn, and cases where information saved to the application tracker was incorrect or incomplete. The original purpose of JobLog was to save time. I did not want to simply replace the time spent filling out forms with time spent correcting everything the extension had filled incorrectly. Because of that, I repeatedly returned to the same features and improved the accuracy of field matching, saved application information, and the way JobLog Filter reads details from job listings. I also received criticism that the website looked AI-generated. When I first started building the extensions, I studied more established websites and products. Some early design decisions were influenced too heavily by things I had already seen. As the products gradually developed, I moved further toward my own visual direction and started paying more attention to smaller details. This included removing unnecessary elements, keeping the products free from advertising, improving cover-letter autofill, and making the overall workflow feel simpler. The only repeated complaint I received directly from users was that JobLog did not always open automatically on certain websites. Sometimes they had to click “Enable on this site” before using it. I changed the activation logic and reduced how often this manual step is required, although website behavior can still vary depending on how each application page is built.

One thing many people get wrong about the job-search tools market is the expectation that they should be able to press one button and receive a good job. Creating a tool that can accurately fill application forms, save information about each vacancy locally, and submit applications without supervision requires an extremely high level of accuracy. This is why JobLog does not automatically submit completed applications. A tool can easily send incorrect or incomplete information, which may reduce the applicant’s chances and ultimately make the job search take even longer. The problem is also not only the number of applications someone sends. People often spend a lot of time applying to vacancies that may no longer be active. A listing might have been published months ago, or several hundred people may have already applied for one position. This also wastes time. I created JobLog Filter so users can immediately see visual signals across listings based on the thresholds they choose. AI is useful as a tool, but I do not believe it can currently replace a person in the job-search process. You still need to sit down, review vacancies, make decisions, and apply. The best we can do is make that process faster, clearer, and more efficient.

The next step for both products is improving accuracy even further. JobLog needs to recognize and autofill more application formats because employers do not only use the largest platforms. There are many customized forms, and the information still needs to be entered correctly. I also plan to expand the number of available cover-letter options and reusable answers for questions such as “Why are you interested in this role?” and other common application fields. For JobLog Filter, I want to add more filters and signals so users can personalize the highlighting more precisely for their own type of job search. A reasonable number of applicants for one position may be completely different for another, so the extension should adapt to the user rather than apply the same rules to everyone. Other planned features include interview reminders and assistance with preparing follow-up questions or emails to employers after applying. I am also planning a third product. JobLog Filter helps users choose where to apply, and JobLog helps them complete and track applications. The third product would focus on creating a stronger and more presentable resume, completing the product line around the full job-search process.

Over 1,000 job applications completed using JobLog in its first month.

I had written books and launched digital products before. While I was still living in Russia, I also built a website related to coaching and sports. From time to time, I made small desktop applications for myself, but I had no experience building browser extensions like this. What I did have was firsthand experience searching for jobs, managing production processes, and recruiting employees, but not developing browser extensions or job-related websites. I created those products and the website because I found the traditional job search process slow and inefficient. Now I’m simply trying to make the job search process easier and more efficient.

At the beginning, I tried to hand off part of the work to AI and kept looking for some way to make the product suddenly “take off.” But in reality, I was mostly doing nothing while searching for that shortcut and wasting time again. Now, I’ve taken a step back and focused on gradually growing the user base. I no longer obsess over outdated or delayed user numbers in the Chrome Web Store and similar platforms. I just keep working on improving the quality of the product. After all, the whole idea was to make the process easier, not to chase numbers for the sake of numbers.