Built the job-hunt workspace I wanted while running my own search
Hi all, Li You from Manchester.
I spent 8 years in product roles across enterprise software, EdTech and banking. Earlier this year I found myself job hunting again, and the part that got to me was not the rejection. It was the silence.
So I built the system I wanted. WorkstationAI scores a job ad before you apply, tailors your CV to it, researches the company, and preps both interview rounds. Every run logs itself, and once you have a few outcomes it runs a retrospective across your whole search.
It's for job seekers who are applying to a lot of roles at once. Career changers, recent graduates, people doing this properly rather than spraying.
Why I actually built it: I ran that retrospective on my own 47 applications. Where I met the core requirement of the role, I converted at 71%. Where I was stretching, 8%. My own confidence score at the point of applying predicted nothing at all. Months of effort aimed at the wrong half of the market, and no tool I was paying for could have told me that.
The feedback I want most: I'm not sure my page makes the difference clear. Every CV tool sounds the same from the outside. If you look at it and think "another resume builder", tell me, because that's what I need to fix before the 12th.
Happy to look at anyone else's page in return.
koecall.ai
AI phone companion for elderly people — always available, always kind.
Comments (13)
Welcome to the community, Li! The 71% vs 8% split is the most impressive thing in your post. Could you drop your SaaS Hive product page link here? Easier for people to check it out and leave feedback.
Thanks Olga, and good to be here. Here it is: https://saashive.com/view-product/0d98cde4-3925-4c55-90ac-77b3ca7f1e71
Fair warning, it's set to launch on the 12th, so it may still be gated until then. Once it's up I'd really value your eyes on it — you've already spotted the thing I keep burying, so I trust your read.
Hello Li You! Welcome to the community :) I don't see your product listed yet but just in case it is still scheduled to be launched, you can share its SaaS Hive product page URL here. Otherwise, we can't see your product screenshots and etc. And there is now a separate space to request for a feedback if you are interested. Here is the page: https://saashive.com/founder-community/request-feedback
Thanks Sergey. You're right that it's not listed yet — it's scheduled for the 12th. Page is here in the meantime: https://saashive.com/view-product/0d98cde4-3925-4c55-90ac-77b3ca7f1e71
And thanks for pointing me at the feedback space, I hadn't found it. I'll post there once the listing is live so people can actually see the screenshots rather than take my word for it.
↳ Replying to You Li
I looked at WorkstationAI page on SaaS Hive and it was not easy to understand what WorkstationAI is. I would add a clear one sentence answer to a question "What is WorkstationAI?" by saying something like "WorkstationAI is ...". Maybe something like "AI workspace is the all‑in‑one environment that runs your entire job‑hunt workflow - from “Should I apply?” to Research, CV, Interview, Cover Letter, and Onboarding - in one place"... I am sorry. I know you did not ask for this feedback but I really felt like sharing it because it was quite challenging fo me to understand what it actually does quickly.
You Li, the 71 percent versus 8 percent number is the strongest thing in this whole introduction, and I almost missed it in the middle of the post. That statistic alone is a better hook than anything about CV tailoring or company research. Have you thought about leading with that number before anything else, since it is the one claim a generic resume builder could never make?
Stacy, you're right and it's slightly annoying how right you are. I buried it in paragraph five of my own introduction.
You've also put your finger on why: 71% vs 8% is a claim about which jobs I should have applied to, and I've been describing the product as a set of tools instead. A resume builder can't say that sentence because it never sees what happened after you hit send.
Leading with the number on the product page before the 12th. Thank you.
The retrospective feels like the real category difference, but I would make the closed loop even more explicit: score before applying, support the application, then learn from outcomes and improve the next decision. A resume builder stops at the document; this sounds more like an application learning system. Does the retrospective currently feed back into future scoring automatically, or is it mainly descriptive at this stage?
Honest answer: descriptive at this stage.
Right now the log fills itself from every job score and every CV tailor, you add the outcome, and then two things run — a triage on a single application (fit gap, writing problem, or just noise) and a retrospective across the whole search. Both tell you the pattern. Neither changes what the scorer does next time.
So the loop is closed for the human and open for the system. That's the deliberate order, because I wanted the finding to be legible before it became automatic — "you convert at 71% when you meet the core requirement" is a sentence someone can act on and argue with. A silently adjusted score isn't.
Where I'm stuck is the cold start: the retro has to say something useful at 3 applications, not 20, or nobody gets to the point where the feedback matters. Any thoughts there would be welcome.
You Li, at 3 applications I would not try to make the retrospective statistical at all, the sample is too small to say anything true. What still works at that size is the single application triage you already built, fit gap, writing problem, or noise, since that needs zero history to be useful. I would hold the pattern report back until there is a real sample, maybe 10 or so, and say that plainly in the product rather than showing a percentage that will just be wrong. People trust a tool more when it tells them it does not know yet than when it guesses early and gets it wrong once.
This is the rule the product is already built to, and it is good to hear it back from someone else.
Two different things run at two different points. The single application triage needs no history at all, so it runs from the first decline: fit gap, writing problem, or noise. The retrospective across the whole search will not run until 8 outcomes are logged, and below that it says so rather than showing a percentage.
8 is my own line, not a statistical one. It is roughly where the clean fit and stretch groups stop being 2 applications each.
Your last sentence is the part I would have got wrong a month ago. I ran the 71 and 8 on 47 of my own applications and I know exactly how tempting it is to show a user their version of that number on day one. It would be wrong once, and once is all it takes.
What makes WorkstationAI different to me is not simply the tailored CV—it is the feedback loop across the entire job search. Discovering a 71% conversion rate for well-matched roles compared with 8% for stretch roles is exactly the kind of insight that could save applicants significant time and emotional energy.
I would emphasize that retrospective more prominently on the page, because “learn which applications actually work for you” feels much more distinctive than another promise to optimize a résumé. As a computer science student applying for software engineering opportunities, I would find the job-fit scoring and outcome analysis especially useful. Your personal results also make the problem and the product’s value easy to understand.
One precision on the 71% versus 8%, because it changes what you do with it. The 8% group is not stretch roles. Plenty of them scored 5 out of 5. The gap was that the missing thing was what the role fundamentally is, and the ad had filed it as "nice to have". For your side that looks like an ad calling itself a Go backend role, listing Go under desirable, and hiring someone who writes Go. If the missing skill appears in the job title or the reporting line, it is the spine of the job, whatever the JD calls it. That test costs nothing and it is most of the value.
You are right that the retrospective is the distinctive part, and it is not the headline on the page yet. Fair hit. The honest reason is that the retro currently runs on my own application history, not inside the product, so I am not putting it on the homepage until it works for someone who is not me.
What is live today and free is the fit check: paste an ad, get a score with the reasoning behind it and a ghost-job check. Worth running on a few of your applications, and I would like to know whether the spine-gap logic reads correctly for engineering roles. It was tuned on product roles, and I suspect SWE ads hide the spine differently, more in the stack than in the domain. If you try it, tell me where it gets you wrong.
Sign in to comment or upvote.