The Core Identity Protocol
(Who, What, & Why)
A story about spite, karak chai, and building a digital twin to handle your content while you touch grass.
Connect & Give Feedback
FeedBack, Bug Tracker & Feature Requests
Found something broken? Have a wild idea? Submit it directly, every report makes this better.
Fuel the Late-Night Coding Sessions
Every chai keeps the chai-powered developer alive and the all-nighters going.
Let's Chat :)
15 minutes, no agenda pressure. Grab a slot and let's sync on what you need.
The Origin Story
It was 2 PM on a random Tuesday, and I was staring at a blank terminal screen with the distinct realization that my screen time was a straight-up biological hazard. I was spending four hours a day scheduling content across three different platforms, switching chrome tabs like a caffeinated day trader who is deeply in debt, and getting absolutely zero dopamine in return. That's when the lightning bolt hit. Why am I manually managing social media like a medieval peasant when I could write a digital twin to do the heavy lifting for me while I sit back and drink a karak chai?
This platform didn't start in some fancy VC corporate incubator with beanbags and ping pong tables. It started right here at my cluttered desk, fueled by sheer, unadulterated spite against corporate API limitations and genuinely atrocious UI designs. I wanted to build something with absolute Main Character Energy. Not another boring corporate dashboard, but a hyper-competent, slightly chaotic social media copilot.
I built “Soshl” to listen to your messy, late-night WhatsApp voice notes, transcribe your raw thoughts using Gemini, and instantly distill them into high-converting content bundles while you're out touching grass. It's built for creators who want to dominate their niche without losing their sanity to the algorithm.
The Sentimental Vision
But if we're being completely real, this is about more than just automating a post queue. It's about creative sovereignty. We've all felt that crushing anxiety of staring at a blinking cursor, wondering if the algorithm is going to shadowban our hard work because we didn't use a trending audio track or some magical mix of thirty hashtags. I wanted to build an engine that acts as a protective shield for your creativity. It handles the soul-crushing, repetitive distribution pipelines so you can focus entirely on your art, your vision, and your voice.
My ultimate mission is to build a full-scale digital twin ecosystem. I want to level the playing field so that an independent creator working out of their bedroom has the exact same tactical output, data intelligence, and operational speed as a multi-million dollar media agency with an internal staff of twenty.
After spending three straight weeks pulling all-nighters, fighting cascading TypeScript compilation errors, and drinking enough chai and coffee to give a small horse a heart attack, I finally got the production build to compile successfully. I leaned back in my chair, feeling like absolute genius, took my phone out to run the first live WhatsApp production integration test message, and texted my AI bot:
“Pilot, are you alive?”
My phone buzzed instantly. Pilot replied:
“Yes, Haseeb. I am alive, fully functional, and ready to dominate the internet. By the way, your posture looks like a cooked shrimp right now. Sit up straight.”
I have never felt so deeply validated and personally attacked by my own creation at the exact same time.
The Bare-Metal Architecture
(For the Tech Nerds, though anyone is welcome to pull up a chair)
[ SoshlPylt Web UI ] ──────┐
(Clerk Auth, React) │ ┌──────> [ Gemini AI (Drafts) ]
▼ │
[ Next.js Route Handlers ] ────┼──────> [ Cloudinary (Media) ]
▲ │ │
[ WhatsApp Remote ] ───────┘ ▼ └──────> [ Sentry (Telemetry) ]
(Twilio Webhooks) [ Database (Neon + Drizzle) ]
│
▼
[ Redis Queue ]
│
▼
[ Background Workers (BullMQ) ]
│
├──> [ Official APIs (YouTube, X) ]
└──> [ Ghost Sync Chamber (Puppeteer) ]The Production Stack
The Framework
Next.js 15 (App Router) on full static-type compliance via TypeScript. Core routing runs on edge-optimized API route handlers for minimal latency when webhooks hit our cluster.
The Database Layer
Neon Postgres paired with Drizzle ORM fully type-safe SQL query generation with zero overhead, allowing the platform to run lightning-fast data lookups.
The Queue Engine
BullMQ running on top of an independent Vercel/Upstash Redis data cache. Events are immediately offloaded into a resilient background job queue so your WhatsApp webhook never times out.
The Intelligence Core
Google Gemini 2.0 Flash. Handles everything from native raw audio ogg-stream transcriptions to complex context-aware polymorphic social media drafting.
Integrated Third-Party Infrastructure
Twilio API
Secure SMS/WhatsApp carrier gateway handles incoming form data streams and formats outgoing conversational messages dynamically.
Clerk Authentication
Manages secure session state structures, user token footprints, and identity management across dashboard routing layouts.
The Ghost Engine (Puppeteer / OpenClaw)
When standard APIs lock us out or limit velocity, the system provisions an automated digital proxy window using headless Chromium to execute native actions securely on behalf of your session profile.
Future Roadmap Strategy
The long-term engineering goal is to fully move past standard API dependency maps. Over the next few quarters, I am expanding our proprietary Ghost Agent Lab to build complete multi-platform browsing automations that work natively directly from your local browser profile context. This means automated community management, deep-scraping inbox discovery, and direct shadowban monitoring across all major social networks without relying on brittle, restricted official partner apps.
The Real Cost Breakdown
Testing Phase: completely self-funded to collect high-value feedback from the first 100 alpha users
| Infrastructure Layer | Service Provider | Monthly Cost |
|---|---|---|
| Serverless Compute | Vercel Pro Tier | $20 / mo |
| Resilient Messaging Engine | Twilio Dedicated WhatsApp Sandbox | ~$35 / mo |
| Background Processing Memory | Redis Cloud Cluster | $10 / mo |
| Database Compute Engine | Neon Postgres Scale Instance | $19 / mo |
| AI Content Inference | Google AI Studio (Gemini Enterprise API Keys) | ~$25 / mo |
| Total Test Sandbox Burn Rate,Keeping the lights on for feedback | ~$109 / mo | |
This system might have absolutely zero bugs, or it might have a couple of hilarious, catastrophic system exceptions that completely crash the loop on execution. We are about to find out together! I would deeply appreciate your raw, unfiltered, completely honest feedback. Test it to its absolute limits, break things, and let me know exactly where it falls short so we can make it bulletproof.