広告効果測定
同意した場合のみX Pixelを使用します。
メールアドレスや制作内容は送信しません。
Some content is automatically translated from Japanese. Untranslated sections are shown in the original language.
TIMELINE ARCHIVE
Follow public updates, creation logs, and related releases from newest to oldest.
Updates
8
Creation logs
0
Related releases
0
TIMELINE
Public records are arranged in one timeline without duplicating the same event.
Oct 26 08:58
In the test environment, we confirmed that importing 100 job listings at once continues processing even after closing the screen. The AI's functions for resume organization, job-specific proposals, and recommendation letter creation are operational, allowing for the review of generated results and editing/saving of recommendation letters. We have also added a mechanism for user consent, and before using real candidates, we will first test it with genuine materials.
We had been progressing with synthetic data, but we determined that we should implement a measurement of effectiveness by inputting actual job listings and our own career backgrounds before moving to practical use.
Using my own resume and work history, I will combine them with actual job listing URLs to verify how my background is perceived and the content of my proposals.
Oct 7 11:30
Candidates can now swipe through job listings and select "Interested," "Postponed," or "Wish to Apply," with the results being verifiable from the CA side. The person in charge of introductions has made progress in summarizing comparisons with current jobs, desired conditions, commuting time, and their connection to experience for each job listing.
When I actually tried using it myself, I felt that the screen with all the information laid out was overwhelming, and I thought, "Excel is easier to use than this." Therefore, while automating processing, I am shifting towards narrowing down the information that people typically see to only the necessary items.
Organize the job detail screen to prominently display only the information necessary for everyday decision-making, while separating the commute confirmation into a smaller, distinct screen.
Sep 25 07:50
On the job seeker side, we have summarized the profile display and improved the response to screen transitions and swipes. Using fictitious candidates and job listings, we generated three personalized proposals with AI, separating the explanations for candidates and the confirmation items for the introducer. Integration of the proposals into the screen has not yet been completed, and actual candidate information has not been provided to the AI.
The job description I saw on the actual machine was generic and did not feel personalized at all. I want to delve into not just whether the conditions match, but also the significance of that person going to that company, similar to the proposals I usually write.
To establish the minimum necessary handling conditions and safety measures for managing actual candidate information, without increasing the daily tasks for job seekers, recruiting companies, or myself.
Sep 16 01:14
I have progressed to the point where I can actually interact with the beta version on my iPhone. Upon using it, I realized that it requires too much input and confirmation from the candidate side, making the process of viewing job listings cumbersome. Moving forward, we will change the flow so that the CA pre-enters candidate information, allowing candidates to view job listings immediately after logging in.
When actually using it as a candidate, I disengaged with the long profile screen. Also, I felt that having just matching job titles was a weak reason to feel that the job openings suited me.
Adjust the flow so that candidates can view job listings immediately after logging in, and confirm the job cards and recommendation reasons again on an iPhone.
Sep 11 00:32
R. decided to move forward by releasing a free beta version instead of introducing the product to companies one by one through sales, allowing the product's quality to naturally replace existing systems. The input platform, which enables career advisors and job seekers to create candidates, register documents, add job listings, input preferences, and log in via a browser, has reached the testing screen. However, during practical verification, it was found that issues such as screen transition delays, slow document and job importation, and the difficulty in using automatic transcription and preference input still remain.
We believe that if the growth is driven by product strength rather than sales capability, it can be operated at low cost by a single person and can be scaled at a price lower than that of typical SaaS.
Fix the waiting times for screen transitions, document uploads, and job imports, and address any discomfort during input, solidifying the input interface to a state that can be used in practical work.
Sep 8 11:26
R. is advancing its policy as a system provided to recruitment agencies with employment introduction permits. It has progressed to an operational flow where job seekers can swipe through job listings, and career advisors can review and determine matches. We have confirmed connections to Supabase, candidate segregation through dedicated URLs, saving swipe results, and the separation of stated preferences, work history basis, and inferred tendencies from behavior in the actual database. Furthermore, we have also confirmed a route to retrieve job information from the job listing URLs shared with candidates, allowing for incorporation into the existing job data without duplication.
I already have several acquaintances with job placement permits, and I can conduct experiments there.
Complete the input interface to allow career advisors and job seekers to upload resumes and work history documents and enter job URLs entirely on-screen.
Sep 4 18:31
We have shaped a swipe-type job application experience for candidates called "R." and prepared a public demo that friends can access without logging in. Additionally, we restructured the MVP to avoid requiring a resume and work history from the start. Instead, users can begin with minimal information and increase recommendation accuracy through swiping and adding information.
In actual personnel introduction, there is significant friction when there is no response from candidates after sending job listings and requests for document preparation. Therefore, I determined that it is more natural to turn the act of viewing the job listing into a hearing, rather than asking for complete information from the start.
Using one actual candidate and one actual job listing, we will streamline the process from presenting job offers based on simple preferred conditions, saving the swipe results, to confirmation on the CA side.
Sep 3 15:28
From the proposal for a job change management tool, we expanded the concept into an AI-native new business product aimed at recruitment agencies. This will be offered to companies with job placement licenses, and we are considering significantly streamlining the labor-intensive aspects of job searching, comparing, and obtaining application intent, using AI, from resumes and work history.
We determined that it would be easier to advance to paid implementation in a short period by providing our services as a new business to recruitment companies that already have the necessary permits, job listings, and operational infrastructure, rather than starting the acquisition of recruitment permits, developing job companies, and attracting job seekers simultaneously.
Create the minimum demo and sales materials to propose to the first recruitment agency, and start sales for paid implementation.
9 2026 3
The R. project was started.
Creators who are quickly testing what can be done with AI.
