PranjaliPatil

Building

[PranjaliPatil @PranjaliBuilds]

Day 25 🚀
Added likes to video requests 👍 + shipped UI improvements.
Creators can now spot high-demand ideas faster.

#BuildInPublic #IndieHacker #SaaS #CreatorEconomy

[Vala Afshar @ValaAfshar]

Steve Jobs on how to give feedback to high performers when their work is simply not good enough; giving difficult feedback without causing resentments is a superpower.

[PranjaliPatil @PranjaliBuilds]

@saen_dev That’s helpful validation, thanks 🙌
At this stage, optimizing for speed of iteration and low ops overhead felt more important than maximum flexibility.

[PranjaliPatil @PranjaliBuilds]

Day 28:🚀
No code today,just architecture decisions.

Researched on AWS vs GCP, serverless & cost trade-offs .Choose GCP for compute because:
• Generous Cloud Functions free tier
• Less setup overhead than Lambda + API Gateway
• Cloud Run is a clean path for LLM services

[PranjaliPatil @PranjaliBuilds]

@helderbuilds An early pattern we’re seeing is that creator replies are highest right after launch and taper off over time, despite ongoing comment activity.

[PranjaliPatil @PranjaliBuilds]

Day 25 🚀
Added likes to video requests 👍 + shipped UI improvements.
Creators can now spot high-demand ideas faster.

#BuildInPublic #IndieHacker #SaaS #CreatorEconomy

[PranjaliPatil @PranjaliBuilds]

@helderbuilds Thanks a lot for the support !!Helps a lot to keep building

[PranjaliPatil @PranjaliBuilds]

@helderbuilds Thanks! We haven’t rolled this out to creators yet, so feedback is still incoming. What’s been interesting so far is how much internal testing already highlights clearer demand signals,excited to see how creators react once they start using it.

[PranjaliPatil @PranjaliBuilds]

@helderbuilds Thanks! we’re still in early rollout. Right now the biggest learnings are from our own usage, especially how quickly high-demand ideas surface when likes and intent frequency are combined. Creator feedback is the next big milestone.

[PranjaliPatil @PranjaliBuilds]

@getnachonacho Thanks for the quick insight!

[PranjaliPatil @PranjaliBuilds]

Building an AI SaaS MVP , any recommendations for cloud providers or tools with generous free tiers?

Currently exploring
-Vercel / Render / Railway free tiers
- AWS Free Account
- Oracle OCI Free Tier

[PranjaliPatil @PranjaliBuilds]

@helderbuilds Appreciate that! At the moment, demand is derived from intent frequency and likes, aggregated at the topic level. Creators see ranked ideas based on how many unique users requested them

[PranjaliPatil @PranjaliBuilds]

@PMV_InferX @saen_dev Thanks for the insight!! .Idea of not tying model lifetime to GPU lifetime.
Feels like a much cleaner way to keep costs predictable as traffic fluctuates.

[PranjaliPatil @PranjaliBuilds]

@PMV_InferX Totally agree. Cloud Run shines for control planes, routing, and bursty/light inference. Once GPUs and larger models come in, cold starts, GPU warm-up, and idle cost become real constraints.

[PranjaliPatil @PranjaliBuilds]

Claude is making good moves in healthcare !!

[Claude @claudeai]

Claude can now securely connect to your health data.

Four new integrations are now available in beta: Apple Health (iOS), Health Connect (Android), HealthEx, and Function Health.

[Haider. @haider1]

Anthropic CEO, Dario Amodei:

"we might be 6-12 months away from models doing all of what software engineers do end-to-end"

We're approaching a feedback loop where AI builds better AI

But the loop isn't fully closed yet, chip manufacturing and training time still limit speed

[PranjaliPatil @PranjaliBuilds]

Day 31 :
Found a backend issue during testing
Similar video requests were getting split into multiple topics (same idea, different wording), which made recommendations messy

Now they’re merged into one topic with a request count
Cleaner, more accurate for creators

[PranjaliPatil @PranjaliBuilds]

@erlalucas Embeddings, transformer

[PranjaliPatil @PranjaliBuilds]

Day 29: Split my backend into 2 repos following microservices:

Backend API: auth, logic, orchestrates ML calls
ML Service: FastAPI, 1 model, 1 endpoint, stateless, Cloud Run–ready
Both services containerized & ready for GCP Cloud Run

#BuildInPublic #FastAPI #Microservices #GCP

[PranjaliPatil @PranjaliBuilds]

Was looking for this 😃

[aditya @adxtyahq]

Someone curated 925 failed VC-backed startups, broke down why they failed, and how to make it work with today’s tech -

https://t.co/NFUhrhe7P2

Cool fr🙌

[PranjaliPatil @PranjaliBuilds]

AI Rocks !!

[Joma Tech @jomatech]

Top webtoon artists make $9M a year and creating anime like this is easy with AI now.

However, I quickly learned that my drawing skills were never the limiting factor to creating a successful webtoon.

This goes to show how all the "HOLLYWOOD IS DEAD" posts don't know what they're talking about.

Image-gen is way better than Video today. There's a webtoon market for it. Why are we not making millions a year yet with AI?

[PranjaliPatil @PranjaliBuilds]

@StevBuilds Thank you,glad it resonates!!

[PranjaliPatil @PranjaliBuilds]

Day 30🚀
Testing on more videos revealed some inaccuracies.
So we’re launching with next video idea requests and pausing intent/sentiment for now.
Clear value first. Intent & sentiment will follow once it meets quality bar.

[Prajwal Tomar @PrajwalTomar_]

People saying “AI can’t design” haven’t updated their workflows.

I replicated this landing page in under 10 minutes using AI.

If your landing pages still look mid, stop blaming AI and fix your workflow.

Here’s the exact workflow I used ↓

[PranjaliPatil @PranjaliBuilds]

@DanutPralea Thank you!! Ml service consist of Intent analyzer : preprocessing → caching → embeddings → intent matching → locking → persistence