I connect the tools your team already uses, so the busywork runs itself.
I'm Abu Hanif MD Jakaria, an AI automation engineer. I design and build systems that trigger on real events, hand the thinking to AI where it belongs, and take the action — no dashboard-checking required.
I build the plumbing between your apps and your AI.
Most businesses don't need more software — they need the software they already have to talk to each other. I specialize in wiring up automation platforms and large language models so that leads get qualified, tickets get routed, reports get written, and follow-ups get sent, without anyone opening a spreadsheet at 11pm.
My work sits at the intersection of workflow automation (n8n, Make, Zapier) and applied AI (LLM prompting, RAG, agentic tool-use) — the pipes and the judgment layer that runs through them.
If a task is repetitive, rule-based, or "someone has to read this and decide" — it's very likely automatable. I find that boundary and build to it.
WITHOUT SUPERVISION
LEFT IN THE LOOP
FIRST WORKING DRAFT
Three ways I plug in
Each engagement starts at one of these nodes and expands to fit the workflow around it.
Workflow automation
End-to-end pipelines that move data between the tools you already run — no more copy-pasting between tabs.
AI agents & assistants
LLM-powered agents that read, decide, and act — qualifying leads, drafting replies, summarizing calls.
Systems integration
Connecting CRMs, sheets, chat apps and APIs into one working system, so data lives in one place, not five.
How a build actually happens
A fixed sequence, every time — so you always know what's next.
Map the manual work
We walk through the task as it's done today — every click, every copy-paste — and mark exactly where time is lost.
Design the workflow
I sketch the trigger, the decision points, and the destinations, and flag anywhere AI judgment replaces a manual call.
Build & connect
The automation gets built against your real tools and real data, not a demo environment.
Test on live edge cases
I run it against the messy, real examples that break most automations, and tighten it until it doesn't.
Hand off & monitor
You get a plain-language explanation of what runs and when, plus a short window of monitoring after launch.
Systems in production
A sample of the shape of work — details adapt per client under NDA.
Lead-qualification agent
● liveInbound form submissions are read by an LLM, scored against ICP criteria, enriched with public data, and routed straight into the CRM with a drafted first reply.
Support-ticket triage
● liveIncoming tickets are classified by urgency and topic, matched against a knowledge base, and either auto-answered or handed to the right teammate.
WhatsApp order assistant
● liveCustomers order over WhatsApp; an assistant confirms items, checks stock via API, and logs the order to a shared sheet in real time.
Weekly reporting pipeline
● liveData is pulled from three separate tools every Monday, summarized by an LLM into plain language, and delivered to leadership before standup.
What runs under the hood
Have a workflow worth automating?
Send a message with what you'd like to stop doing by hand. I'll reply with a quick read on whether it's a good fit.