346 active days.
386 problems solved.
112-day max streak.
The biggest lesson wasn't DSA.
It was learning to show up even on days when I didn't feel like solving anything.
Consistency compounds.
#leetcode#100DaysOfCode#SoftwareEngineer
If you're building an AI product today...
What's the first feature you'd add?
• Document chat
• Memory
• Citations
• Multi-agent workflows
I'm curious what other builders are prioritizing.
One thing RAG taught me:
LLMs don't become "smart" because of better prompts.
They become useful when they can retrieve the right context.
Retrieval quality > prompt quality.
Agree or disagree?
#AI#RAG#SoftwareEngineering
Big milestone for FollowUp AI 🚀
Started with AI-powered follow-up generation.
Now it supports:
📄 Document uploads
🧠 RAG
🔎 Semantic search
📌 Source citations
Now it's all about polishing the experience.
#buildinpublic#RAG
@Anthony_K81 I haven't reached streaming yet—my MVP uses synchronous requests with structured JSON validation before persisting data. Once I add document processing and RAG, retries and idempotency will become much more important. Curious what solution worked best for you.
Hot take:
Most AI demos fail in production for reasons that have nothing to do with the LLM.
It's usually:
• state management
• retries
• validation
• bad architecture
What's the biggest engineering challenge you've faced building AI apps?
@Sventekvo1 That's a great point. I'm trying to design FollowUp AI with a provider abstraction from the start for the same reason. Out of curiosity, what failure ended up being the hardest to track down in your agents?
Curious how other developers approach this.
When building AI apps, what's more important?
A) Better prompts
B) Better system design
I'm finding that architecture matters more than the model itself.
What's your experience?
#AI#buildinpublic
A small engineering lesson:
Don't tie your app directly to one AI model.
I kept my AI logic separate from the business logic, so switching from Gemini to another provider later becomes much easier.
Future-proofing > rewriting.
#SoftwareEngineering#AI
Building FollowUp AI one step at a time.
Phase 1 is complete:
✅ Action items
✅ Open questions
✅ AI-generated follow-up emails
Next up: Document uploads + RAG so the AI can answer using your own files.
#buildinpublic#AI#NextJS
Who is FollowUp AI for?
Freelancers.
Agencies.
Small teams.
Anyone who spends more time writing follow-up emails than doing actual work.
The goal isn't just AI-generated summaries.
It's making sure nothing discussed with a client gets forgotten.
#buildinpublic#AI
Why I'm building FollowUp AI 👇
After every client meeting, the real work begins—notes, action items, and follow-up emails.
I'm building AI to automate that.
🚀 Phase 1 MVP is complete.
#buildinpublic#AI#NextJS
Been away from X for a while.
Spent the last few months building instead of posting.
Currently:
• Building FollowUp AI 🚀
• Learning Next.js, Node.js & PostgreSQL
• Grinding DSA & SQL
• Preparing for SDE-1 roles
Back to building. Time to share the journey.
Just received my first payment from a freelance client.
A few months ago, I was just learning, building projects, doubting myself, and wondering if any of this would ever work. Today, someone paid for my skills.
Small win maybe, but for me this is proof that consistency compounds