Abdullah Tariq
I build AI voice agents & automations that run 24/7.
New to Contra
Abdullah is ready for their next project!
I built a system that writes, images, and schedules my LinkedIn + Contra posts every day automatically.
Here's exactly how it works:
Every morning at 8am it:
→ Searches the web for what's trending in AI and automation (Tavily)
→ Picks the best topic based on my past posts (no duplicates)
→ Researches it with a live deep-dive
→ Drafts two versions — one for LinkedIn, one for Contra — using Claude Sonnet 4.6
→ Generates a branded image card
→ Emails me a preview to review
→ Posts automatically once I approve
Total running cost: $1.57/month
The stack I used:
n8n (self-hosted workflow engine)
Claude Sonnet 4.6 — post drafting and rewriting
Tavily API — real-time trend research
Google Sheets — content calendar, idea bank, post log
HCTI — branded image cards
Kie AI — AI-generated visuals
Gmail OAuth2 — daily preview emails
The whole thing runs on Docker on my local machine. Zero monthly SaaS fees. I still review every post before it goes live — the system handles the research, writing, and scheduling.
If you're a founder or business spending hours on content every week, this is the kind of automation I build.
DM me if you want one for your brand.
2
1
70
Had a fun weekend building Lumina ⚡
Designed with Google Stitch.
I'm a developer who's been putting off rebuilding my portfolio for months. My question going into this challenge: could Stitch help me ship a real product — not just a mockup — in a single sitting?
So I built Lumina. Paste your GitHub URL, watch a fully animated portfolio stream onto the canvas in 60 seconds. Real repos, real data, live and shareable immediately.
How Stitch fit in:
I designed all five screens in Stitch before touching any code. The streaming canvas generation was the moment that changed everything — watching the portfolio build itself section by section made me realise: that's exactly what I want users to feel. The feature inspired the feature.
From there I worked almost entirely through in-place edits. Clicked the project cards, typed "stronger 3D hover with a glow matching the language color." Clicked the hero, typed "mesh gradient with a grid pattern." Changes in seconds. No re-prompting from scratch.
The motion work was where Stitch surprised me most — particle backgrounds, animated skill rings, 3D card hovers, all authored inside Stitch. Then one push to Netlify.
Features I used:
Streaming canvas generation
In-place AI edits via prompts + point-and-click
Import from existing website (used Vercel.com (http://Vercel.com) as style reference)
Native motion and hover states on HTML canvas
Netlify export
Feedback:
The streaming canvas collapses the describe → wait → judge → re-prompt loop into something that actually feels like designing. In-place editing is where I spent most of my time — clicking and describing beats tweaking properties in a panel every time.
Stitch didn't feel like a tool I was prompting. It felt like a canvas that understood what I meant.
Check it out:
Live site → https://grand-pixie-01f370.netlify.app
(https://grand-pixie-01f370.netlify.app)GitHub → https://github.com/AbdullahTariq74/Lumina-Stitch
3
1
87
Missed calls cost dental clinics thousands in lost appointments every month. A voice receptionist that picks up every time solves that.
The system answers incoming calls, confirms patient details, checks calendar availability in real time, and books the appointment through natural conversation. The patient receives an SMS confirmation automatically. Everything runs 24/7 with no staff involvement.
Built using Retell AI for voice interaction, Google Calendar API for scheduling, and n8n to connect the workflow. Setup takes about 6 hours. Cost per call runs under 50 cents.
A simpler version starts with appointment reminders, the system calls tomorrow's patients automatically to confirm their bookings. Same tools, easier build, still eliminates hours of manual calling.
Works for any service business that schedules by phone, dental clinics, med spas, home services, salons.
If you need this built, message me on Contra.
0
18
Most small businesses lose 62% of their incoming calls. Each missed call costs around $1,200 in lost revenue. That adds up to $126,000 per year.
An AI voice agent answers every call — after hours, during lunch, when your team is busy. No hold music. No voicemail. Just a conversation that books appointments or captures leads.
The system connects to your calendar and CRM. Cost runs under $200 a month. Setup takes about a day. Most businesses see 70-90% cost reduction compared to hiring a receptionist for the same coverage.
Companies winning right now are not the ones with the best service. They are the ones that answer the phone.
If you need this built, message me on Contra.
0
25
Voice AI platforms like Vapi and Retell AI solve different problems. Choosing the wrong one wastes time and budget.
Vapi works best for fast deployment — appointment setters, lead qualification bots, workflows that need to go live in under a week. Clean API. Plugs into n8n and Make without friction.
Retell AI handles high-volume outbound better. Voice quality and latency stay consistent at scale. Matters when conversion rates depend on how natural the bot sounds.
Most businesses pick based on pricing. That leads to rebuilds later. The smarter move is mapping out the conversation flow first, then choosing the platform that fits the logic you need.
Voice agents only work when the flow is designed right. If you need this built — appointment booking systems, lead qualification bots, or outbound voice campaigns — message me on Contra.
0
21
AI Workflow Builder — Requirements Doc to n8n in Minutes
Upload a requirements document. Get production-ready n8n workflow files back.
That's the short version. Here's what actually happens under the hood.
The system takes any PDF or DOCX requirements doc, extracts the text, and sends it to Claude Opus 4.7 which generates a structured production plan broken down by workflows, triggers, integrations, and logic. You review the plan in the dashboard and approve it. Claude Sonnet 4.6 then generates the actual n8n workflow JSON files from that plan. Download them individually or as a ZIP, import into n8n, add your credentials, done.
Workflows come out 80 to 90% production ready. No more building from scratch.
What I built:
Document parsing pipeline (PDF + DOCX)
AI planning agent using Claude Opus 4.7
Workflow generation agent using Claude Sonnet 4.6
Full approval flow with plan review UI
Single and bulk ZIP download for workflow files
Auth, project management, file storage via Supabase
Stack: FastAPI, Next.js, Supabase, Railway, Vercel, Anthropic API
Built with: Claude Code
Interested in something like this? Let's talk.
0
25
Professional Website built with just one prompt in Claude Code.
0
29
This is a video auto generated by python agents.
0
32
I automated the Blog generation for SEO for a company
0
37
Building AI Agents that work 24/7
0
38