Pinned Tweet
I built Shorteefy after trying every AI clipper out there.
They are fast. But they decide everything for you. The clip starts mid sentence and you cannot fix it. And you pay every month whether you clip or not.
Paste a long video → get short captioned clips. You pick how involved you want to be:
→ Quick Mode: AI does the whole thing. Finds the moments, crops to 9:16, adds captions, hands you finished clips.
→ Advanced Mode: AI drafts the clips. You move the start and end word by word so nothing cuts mid sentence, fix any wrong caption, then render.
The pricing part:
→ No subscription
→ ₹2.50 a minute in India, $0.03 a minute everywhere else
→ Start with ₹150 or $6
→ Full HD, landscape and portrait
shorteefy.in
1
1
3,163
Shorteefy is being supported by claude for startups as well as grant from eleven labs.
Thank you.
@ElevenLabs @claudeai
2
33
Actually I thought it made sense to move to a local transcription model-based platform like Handy but the point of Wispr Flow is the flow.
The transcription happens immediately and we do not notice the amount of time wasted when we use Handy. It breaks the flow for someone who is trying to achieve that.
The other part is, if you have a good laptop maybe try with the different models.
The only place I use opensource models on everyday basis is local transcription. I don't use any subscription of products like wispr flow. I use this opensource app called handy.
I am curious, are you using any opensource models on day to day basis, if yes - what's the use case?
70
Figuring out the SEO for a SAAS: Post 1
I’m figuring out SEO for Shorteefy, and the first lesson was humbling: most keywords in my space aren’t mine to win.
Every keyword falls into two piles.
Pile one: terms the big players have owned for years. A new site fighting there doesn’t land on page 5; it doesn’t land at all.
Pile two: terms they don’t care about. That’s where a small product can actually show up.
So how do you find pile two? I’m using three filters.
Does it describe what makes us different?
Opus Clip can’t rank for “AI clipper without subscription,” because they’re a subscription. We’re pay-as-you-go from ₹150. Our differences are our keywords.
Does it lead to the product?
A blog post that ranks but never sends anyone to clip a video is vanity traffic. Every keyword has to map to a feature, a free tool, or a comparison.
Will an AI Overview eat the click? Google’s AI summaries now sit above results, and studies show clicks drop sharply when one appears, mostly on “what is” questions. Searches where someone has to do something (convert, trim, compare, buy) still get clicked. That’s where I want to be.
The test I now run before writing anything: Google the keyword myself. If page one is all giant brands, I skip it. If it’s Reddit threads and outdated listicles, Google is telling me it has no good answer yet.
If you’ve cracked SEO for a small SaaS, I’d love to hear what worked.
8
4
175
I’ve been working on Shorteefy for a while now and kept running into the same thing. AI clipping tools are fast, but the clips they make often don’t make sense to someone who hasn’t seen the full video.
Stuff like: the clip starts with the answer but the question was two minutes earlier. Or it cuts off mid-sentence because the “score” dropped. Or it spells the guest’s name wrong in the captions, which is the one word people need to find you.
After looking at a lot of clips, most good ones have 4 parts:
Hook: Does the first line make someone stop scrolling?
Setup: What’s the smallest bit of context a new viewer needs?
Payoff: Does it deliver what the hook promised?
Ending: Does it end cleanly, or loop back so people rewatch?
AI can find candidate moments well. It can’t really judge these four. So in Shorteefy’s Advance Mode, the AI shows you its picks as text before anything gets rendered. You pick the ones you want, move the start/end, fix captions, and only then the clips are made.
If you just want speed, Quick Mode still lets the AI do everything.
Also no subscription, you just pay per minute of video you use.
Wrote the whole thing up here, including a short checklist you can use with any tool: shorteefy.in/blog/human-in-t…
Curious how you all handle this right now. Do you check every clip before posting, or just trust the AI and post?
1
86
Has any one worked on Claude.md file recently to make it sync with newer models? There has been lot of advise from Anthropic Top People
2
50
Update from a solo built product.
This morning I had 6 speed fixes from a Lighthouse report.
Browser waited for:
> 283 KB of JavaScript started before the first paint
> My error tracker and analytics only loaded after 6 s
> The one stylesheet finished last, at 3.2 s
> Nothing could paint until it arrived
Few things I tried:
> 2 of the 6 fixes would not have helped. One would have made it slower
> New fix: each page now carries the CSS it needs inside its HTML
Lets see.
4
70
Do not miss watching this.
As AI becomes accessible to everyone, there is less chance to stand out.
Also, we are all moving towards the average (a better average), and there is nothing to distinguish us.
Prof shared a funny observation: all the recent presentations looked the same.
So what is the way?
Old school method. Take a problem. Struggle.
Where does AI come in then?
To make the process smoother, by coming in when you’re stuck.
2
56
Thanks Claude for Startups.
This flow is awesome to work on HTML mockups before moving to actually implement it in product.
3
1
78
From the head of design, stripe.
If we only use AI to make the things that we already make, just faster, then we're definitely missing out on the most interesting part.
From my talk at @lennysan’s Summit, here are four ideas for raising the ceiling, not just the floor:
- Have a point of view [5:38]
- Encode your standards into the machine [9:07]
- Don't confuse done with good [12:21]
- Unleash creativity and artistry [16:19]
Full talk here: youtu.be/GLvFTMtw4Jk?si=-5Ts…
1
1
51
Take time and read!
There is a skill worked out to use it.
github.com/pronskiy/skills/b…
44
I asked Claude 5.5 to write notes for Lecture 3 (RAG) from Stanford CS 329Z: Engineering AI Agents.
Lately, I’ve found spoken Hindi useful for explaining technical concepts. So I made these notes in two languages and at two levels:
→ English + spoken Hindi
→ Easy + Dense
Easy builds intuition. Dense aims to match the depth of the course PDF.
The approach to writing the notes draws on the research explained here:
stanford-cs329z-notes.vercel…
Read the notes:
stanford-cs329z-notes.vercel…
Take them, adapt them, or improve them:
github.com/shorteefy/stanfor…
Hope someone finds these useful :)
This fall @michaelryan207 @jyangballin and I are teaching a new course CS329Z "Engineering AI Agents" on how to build AI Agents from scratch. Come join us and learn how to build them 🤖
2
95
ChatGPT recommendations for most affordable tools to repurpose long form video into shorts.
Shorteefy is one among them..
Happy.
1
69
Has anyone successfully received ElevenLabs credits after being accepted into Claude for Startups?
1
4
181
I ask AI Agent to make changes and as I reflect on them, I switch on Handy (Open Source Speech to Text Wispr Flow) and keep on saying things I align with and with on.
Basically a dump of how I see the changes.
Agent takes the baton then.
This approach has worked well for things where I am not thinking from first order principles to design a feature. There, I am good with a notebook and a pen.
1
87
I made the same mistake while interviewing a candidate. Some times great programmer who remembers all syntax are not great thinker (problem solver).
In the era of AI, we need to give weightage to it too.
Decent talent is very rare in the market these days.
I was taking an interview today for a person who has 3 YOE. I asked the most basic question someone can ask - "Write a code to make an API call and if it falls then retry it for a maximum of 3 times".
Even for this the candidate took 15 minutes and still getting syntax errors.
This is also an outcome of using AI for everything and then not knowing the very basics. Just because you are using AI, doesn't mean that you can't make an API call.
And all this after the candidate claimed that he is not good at DSA but is good at development🤦🏻
2
317




















