
software developer @papercheck_ai. i also build my own apps, Rowpin and Driving Permit Practice, and share what i learn here.
Kampala, Uganda
Joined August 2020
- Tweets3.2K
- Following1.1K
- Followers333
- Likes16.4K
Pinned Tweet
Four founders on X this week: 7 to 16 days in Waiting for Review.
My first app: 3 days.
The update after it: about a day.
Both passed first try.
Before every submission I fill in a written yes/no checklist. I can't say it's why.
It's public:
github.com/jordan-jakisa/app…
We just launched PickyProf, @papercheck_ai
but optimized for students in the US
It helps students check their argument, structure, citations, sources, grammar and plagiarism all in one tool.
Here are screenshots from a sample paper.
Link: pickyprof.com
2
4
50
Rowpin asks a crocheter's first name in onboarding.
It stays on the phone, so our privacy label didn't change.
Personalising the paywall on device keeps it that way.
No name sent to an AI API.
Make your paywall feel personal.
Use Apple’s on-device Foundation Models + RevenueCat custom variables to tailor it to each user - privately, with no AI API costs.
Build it in an afternoon. Start with this quick guide ↓
2
10
Rowpin, day 1.
My crochet app has been live for 3 weeks.
Paying customers: 0
Active trials: 0
MRR: $0
I'll post one number a day until that changes.
5
6
155
Everyone's asking whether to drop the free trial. I checked when mine cancel first.
Rowpin, my crochet app:
6 trials started
6 cancelled within 2 days
2 within 21 minutes
0 paid
One was a 7-day trial. Gone within 2 days too.
A shorter trial wouldn't have saved any of them.
4
5
179
Everyone's writing AGENTS.md files this week.
Skip what the code already shows. The most useful lines in ours are what the agent must not do yet.
Like: this refactor is staged, leave these writes alone until stage 2.
The code can't tell it what we decided to wait on.
4
3
124
Spent the last two nights in PaperCheck.ai instead of another chat tab.
Not using it as a grammar pass. Using it as a source-locked workbench on one ugly job: a methods + citations draft where three sources disagreed on the same retrieval claim, and I needed a paper I could defend in review — and later in a defense.
Started with the instant report. That layer is the cold-start fix. You upload the thesis, it comes back in minutes: structure, clarity, citations, logic flags. Fine for “what’s loud.” Dangerous if you stop there, because a clean score will treat your own draft as ground truth.
This is the part that felt different.
The product is not a solver. The product is a box you fill: the manuscript + the rules you attach.
I dumped the paper, then uploaded the journal guide and the citation spec as custom checks. Then I told the review to only judge against those files plus the text I submitted. When a “significant improvement” claim survived the first pass with no primary number behind it, it was because I had left a blog citation in the set. I removed it. The next report changed. That is the whole product if you actually curate the pile.
I also refused to let it “do the math.” Stats stayed in the table I already had. PaperCheck’s job was whether the sentence matched the table, not to invent a result. Mid-task I killed a branch: no consumer chatbot evals in a production RAG methods section. I did not restart. I just dropped that subsection and re-ran. The rest of the manuscript held.
What I asked, almost verbatim:
“Check this draft only. Compare hybrid search vs late-interaction rerank for a 40M-chunk enterprise corpus. Constraint: on-prem, no data leaving the VPC. Ignore marketing language. Every numeric claim needs a primary source already in this paper or the requirement files. If two sources conflict, flag the conflict. Do not recommend a vendor until the evidence graph is complete.”
What came back was not a vibe rewrite. It was a punch list tied to pages I could open. I still rejected the polished wording on the vendor lean. The structure notes were good. The cost paragraph was thin until I rebuilt the table. Second pass was something I would actually send to a supervisor. Defense Studio slides were leftovers. The locked checks were the deliverable.
What I would not give it yet: an open scientific claim I cannot check, then treating the AI report as a library. If I cannot falsify the sources, a grounded supervisor is just a confident wrong room.
Working rules from this week
• Upload to gather. Custom checks to decide what is allowed to exist.
• Read the flagged line before the summary score.
• Delete any citation that cannot show a primary number, even if the prose is good.
• Re-run mid-task instead of starting over. The requirement set is the feature.
• Run numbers in the sheet. Do not let the editor invent a spreadsheet.
PaperCheck is not “chat, but for theses.” It is a library you are responsible for. The model only survives an audit if the pile does.
Anyone else running a real paper inside locked guidelines, or still pasting chapters into a general model and hoping the citations are real?
@papercheck_ai
We just launched PickyProf, an AI essay checker built for US students.
Page-by-page feedback on structure, arguments and citations, plus a plagiarism check, before you submit.
Your first paper review is free: pickyprof.com
3
4
180
Jordan retweeted
I can’t stress enough how little an idea matters compared to the agency of the people executing the idea.
I have had the privilege of knowing and sometimes even working with some of the most successful people (by various metrics).
The difference between mediocre and excellent work and outcomes is predominantly one of agency.
In practice this means: they dont wait for things to happen to them they go out and make things happen for them.
They don’t wait for someone else to do something, for someone to teach them, for someone to give them the path, etc. They just go out and find a way to do it.
I think the single biggest superpower these people have is the realization/belief that the world around them is completely mutable. Most everything that happens is because a person made it happen.
I used to tell people to look around the room you’re sitting in. Look at everything. Every noun. It almost all exists because a person willed it into existence. Nothing is stopping you from doing the same.
I see people online all the time dismissing someone else’s success because “I had that idea first” or whatever. I mean… yeah? If so then the difference is… you. So a bit of a self own whenever I hear that.
Number one tip: act with agency.
Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber @google blog.google/innovation-and-a…
2
90
Jordan retweeted
The Solutions team @MistralAI is hiring globally!
If you’re entrepreneurial, hands-on, and want to shape how enterprises adopt AI, let’s talk.
Apply: mistral.ai/careers/?utm_sour…
Jordan retweeted
day 1 of AppOrbit
i am officially the top earner on the platform.
let's see which fellow astronaut comes to dethrone me tomorrow.
list your app: app-orbit.net
Jordan retweeted
For a country that holds prayers every opportunity it gets, and demonizes its own cultures as evil, and gives glory to God while doing very evil things, I find our obsession with witchcraft very ironic.
Are you saying the weapons formed against you are prospering?
Monthly subscription for people writing continually. Credits roll over up to 12 months, and you choose which checks to run, so you stay in control of usage.
1
1
23
Word support is more robust now (no PDF export first), and there's a referral program: share PaperCheck, earn usable credits.
More on the new pricing here: papercheck.ai/en/blog/new-cr…
1
7





















