Your business, run by a 24/7 AI team.
I design self-hosted, multi-agent AI ecosystems that automate customer communication, lead capture, booking, voice interaction, moderation and reporting — across Messenger, WhatsApp, Telegram, Email, Voice and e-commerce.
If you can describe the job, I can build the agent.
Not limited to one platform, one channel or one tool. I design, build, deploy and maintain custom AI agents for any workflow — from a single chatbot to a coordinated team of agents. n8n is one of the tools I use; the model, framework and channels are chosen for your job.
Customer-support & sales agents
Answer questions, qualify leads, book calls and hand over to a human — on WhatsApp, Messenger, Telegram, web chat and email, in Bangla or English.
Voice & call agents
Agents that listen and speak: voice notes, phone-style booking calls, speech-to-text summaries and spoken replies.
Knowledge & RAG assistants
Assistants grounded in your own documents, products, SOPs and databases — with sources, and a safe “I don’t know” fallback.
Research & analyst agents
Scheduled agents that browse, compare, extract and report: market, competitor, price and lead research delivered as a daily brief.
Document & data agents
Read invoices, forms, IDs, PDFs, spreadsheets and emails; extract, validate and push clean data into Sheets, CRM or accounting.
Coding & DevOps agents
Code review, bug triage, test generation, uptime and log monitoring — agents that diagnose problems and open the ticket.
Marketing, ads & SEO agents
Content pipelines, social posting, comment moderation, ad-account audits and SEO reports that run on their own every day.
Back-office & operations agents
Orders, invoices, inventory, HR screening, approvals and reminders — the repetitive work your team does by hand today.
Tool-using agents
Agents that call APIs, read and write inside your apps, fill forms and trigger other systems (REST, webhooks, MCP-style tools).
Multi-agent teams
An orchestrator with specialist agents — researcher, writer, checker, publisher — that plan, debate and merge their results.
Education, healthcare & NGO assistants
Multilingual information, appointment and follow-up agents with strict data control and a human always in the loop.
Something not on this list?
Describe the job in plain words. If it follows a process, I’ll scope the agent, the stack and the cost with you.
Live on self-hosted n8n — real automations, real results.
Pick a system to see how it works, step by step.
💬 Saizul Personal AI
An end-to-end Messenger AI agent that replies in fluent Bangla, captures leads and hands over to a human when needed.
What it does
- Fluent Bangla replies
- Automatic lead capture into Google Sheets
- Conversation-state memory
- Human-handoff routing
Stack
🏛️ Multi-Agent Council
Three specialised agents with distinct personas — working alone, or together in Council Mode where outputs are merged into one structured answer.
What it does
- Force (Gemini), Nexus (OpenRouter), Victor (Groq)
- Name-based routing via Switch node
- Parallel “Council Mode” with merged output
- Gmail integration and shared memory
Stack
📱 WhatsApp AI
A conversational booking and inquiry agent that checks availability, answers pricing and confirms bookings.
What it does
- Availability checks in chat
- Pricing and service answers
- Booking confirmations
- Lead capture routed to mailbox or sheet
Stack
👁️ Facebook AI
Page automation for customer messages and booking requests with multi-modal understanding.
What it does
- Understands text, images and voice
- Comment context analysis
- Booking request handling
- Google Sheets logging
Stack
🛍️ Order Automation
An AI sales assistant that answers product questions, captures orders and follows up with customers.
What it does
- Product inquiries answered instantly
- Chat-to-order conversion
- Structured CRM entries
- Customer follow-up
Stack
👟 KiK Messenger Bot
A branded Messenger agent for a commercial shoe brand, with Meta Ads campaign structure managed via API / MCP.
What it does
- Branded conversational experience
- Meta Graph API integration
- API / MCP-managed Meta Ads structure
Stack
📧 Daily AI Email Digest
An email-ops assistant — a scheduled inbox digest every morning plus on-demand email queries.
What it does
- Scheduled 10 AM digest
- On-demand email queries
- Secure OAuth 2.0 access
Stack
From a single chatbot to a coordinated team of agents.
I design, build, deploy and maintain every category of AI agent and automation system.
Conversational AI Agents
24/7 support, sales and inquiry agents for Messenger, WhatsApp, Telegram, web chat and email — multilingual (incl. Bangla), conversation memory, human-handoff logic and unified-inbox routing.
Multi-Agent Systems
Orchestrator-led agent teams: specialist sub-agents (email, tasks, calendar, research) coordinated by a master agent; parallel “council” architectures where multiple LLMs deliberate and merge outputs.
Voice AI Agents
Voice-first automation: speech understanding of customer voice messages (Gemini Audio), voice-note-driven workflows and voice-controlled agent interfaces.
Business Process Automation
Lead capture → intent classification → personalised reply → CRM entry → follow-up sequences; order processing, booking management, invoice and report generation, approval chains.
RAG & Knowledge Assistants
Company-specific assistants grounded in your own data — Pinecone vector databases, embeddings, retrieval pipelines and context-aware answers for support teams, internal docs and product catalogues.
Marketing & Growth Automation
AI content pipelines, automated posting and comment moderation, ad-campaign structures via Meta Graph API, daily AI analytics reports and email digest automation.
AI Strategy & Integration
Practical adoption roadmaps for SMEs and agencies: the right LLM stack, self-hosted vs API trade-offs, cost engineering, data privacy and measurable-ROI rollout plans.
Active Roadmap
Google Drive / Docs automation, PDF pipelines, advanced voice-agent interfaces and deeper multi-agent orchestration.
How three LLMs think as one.
In Council Mode a single message fans out to specialist agents in parallel — their answers are merged into one structured reply.
Real automations, end to end.
Three more production workflows — one that audits ad accounts, one that turns lead forms into qualified conversations, and a WhatsApp agent that understands voice, images and video.
Meta Ads Daily Audit Bot — AI reads your ad account every morning
Eight Meta Graph API pulls are merged, analysed by Gemini and delivered as a clear daily & weekly report in your inbox — no dashboards to check.
Lead Ads → CRM → WhatsApp AI Agent — from lead form to qualified opportunity
A new Facebook lead lands in the CRM and gets an instant WhatsApp welcome; the AI agent then chats (text, voice, images, video), qualifies the lead and moves it through the pipeline.
Multimodal WhatsApp AI Agent — understands text, voice, images and video
Every message type is converted into text by Gemini, then a business-trained agent answers using a Google Docs knowledge base, remembers each customer’s conversation and logs every chat.
Complex workflows, designed end to end.
The systems above are live in production. Below are seven more advanced agent architectures — multi-step, multi-agent and human-in-the-loop — that I can design, build and run on the same self-hosted stack.
✍️ AI Content Factory
A team of agents that researches a topic, writes it, fact-checks, designs the visuals and publishes across your channels — with you approving before anything goes live.
What it does
- Parallel research from the web and your own site data
- Brand-voice and style memory so every piece sounds like you
- One-tap approve / reject on Telegram or WhatsApp
- Auto-publishing to WordPress, Facebook and YouTube
- Performance feedback loop that improves the next batch
Stack
🎯 Lead-to-Close Sales Agent
Every new lead is enriched, scored, answered within seconds and followed up until it books a call or says no.
What it does
- Replies in seconds in Bangla or English on WhatsApp / Messenger
- Scores intent, budget and timeline with an LLM
- Checks the calendar and books the slot itself
- Follow-up sequence that stops the moment the lead replies
- Weekly pipeline report to the owner
Stack
🛟 Omnichannel Support Agent with RAG
One agent for every channel that answers only from your approved knowledge — and hands over to a human, with a full summary, when it isn’t sure.
What it does
- Answers strictly from your documents, prices and policies
- Low confidence triggers a handoff with the conversation summary
- Same memory across channels in one unified inbox
- SLA timers and sentiment alerts for unhappy customers
Stack
📞 Voice Appointment Agent
An agent that takes the call, understands what the caller needs, books the appointment and confirms it by message — day or night.
What it does
- Understands Bangla and English speech
- Books, reschedules and cancels against real availability
- Sends confirmation and reminder messages automatically
- Full transcript and summary logged for your team
Stack
🧾 Document & Invoice Processing Agent
Documents arrive by email or chat; clean, validated data leaves — and only the real exceptions reach a person.
What it does
- Reads invoices, receipts, IDs, contracts and forms in Bangla and English
- Checks totals, duplicates and missing fields against your rules
- Routes exceptions to a human with the exact reason
- Structured output in Google Sheets, CRM or your ERP
Stack
🔭 Market & Competitor Research Agent
Scout agents gather the data every morning, an analyst agent compares it, and you get a one-page decision brief instead of 40 open tabs.
What it does
- Tracks competitors, prices, reviews, keywords and news
- Cross-checks sources and highlights only what changed
- Turns raw data into a clear recommendation, not a dump
- Instant alert when something moves sharply
Stack
🛠️ Self-Healing DevOps & Monitoring Agent
Watches your servers and sites, diagnoses what broke, proposes the fix and — after your approval — runs the safe action and writes the incident report.
What it does
- Separates real incidents from noise
- Reads logs and config to find the likely root cause
- Only pre-approved, reversible actions can run automatically
- Every incident documented for the team
Stack
…and many more, for example
Own your AI workforce — no SaaS lock-in.
Cut response time, reduce headcount cost and scale without paying per seat or per message.
Built for your industry.
Education & consultancies
Answer course and visa questions, qualify students, book counselling calls and capture leads around the clock.
E-commerce & retail
Handle product inquiries, capture orders in chat and follow up — straight into your CRM.
Booking-based services
Check availability, quote prices and confirm bookings on WhatsApp, Messenger or your website.
Real estate & property
Qualify enquiries, share listings, schedule viewings and route hot leads to agents.
Healthcare & NGOs
Multilingual information assistants, appointment requests and internal knowledge bots with strict data control.
Agencies & marketers
Automate reporting, comment moderation, content pipelines and ad-campaign setup.
From first chat to live agent.
Discover
Map your channels, questions, leads and bottlenecks.
Design
Conversation flows, personas, LLM stack and data privacy plan.
Build
n8n workflows, API and CRM integrations, memory and handoff.
Test
Real conversations, edge cases, languages and escalation paths.
Deploy & improve
Self-hosted launch, monitoring and continuous optimisation.
The tools behind the agents.
Questions, answered.
Can you build any kind of AI agent?
Yes. I build custom AI agents for any process that can be described — customer support, sales, voice, research, document processing, back-office, marketing, coding and DevOps, and multi-agent teams. n8n is one of the tools I use, not a limit: I choose the model, framework and channels that fit the job.
I don’t know exactly which agent I need — can you help?
Yes. Tell me the task your team repeats most. In the free discovery call we turn it into an agent design, the right stack, a timeline and a clear price.
Do you only work with n8n?
No. n8n is my main tool for fast, self-hosted automation, but I also build custom agents in code (Python / FastAPI), connect any API or webhook, and combine several LLMs when the job needs it.
What does “self-hosted” mean for me?
Your agents run on your own server (a VPS with Docker + n8n) instead of a monthly SaaS platform. You keep control of your data, pay only for hosting and API usage, and avoid per-seat or per-message fees.
Which channels can the agents work on?
Facebook Messenger, WhatsApp, Telegram, web chat, email and voice-note workflows — with a unified inbox and human handoff when a person needs to step in.
Can the agent reply in Bangla and other languages?
Yes. My Messenger agent replies in fluent Bangla, and the agents can work in English and other languages, depending on the LLM stack chosen.
Which AI models do you use?
Google Gemini, GPT-4 / 4o, Claude, Groq and OpenRouter — chosen per use case for quality, speed and cost. Company knowledge can be added through Pinecone-based RAG.
How long does it take to build an agent?
A focused single-channel agent can take about one to three weeks. Multi-channel or multi-agent systems take longer — you get a clear timeline after the discovery call.
Do you offer support after launch?
Yes — monitoring, optimisation and new workflows as an ongoing retainer, plus hosting and server management if you need it.
Ready to automate your business?
Let’s build your automated future — starting with a free discovery call about the work you want your AI team to take over.
No commitment · Free discovery call · WorldwideTell me what you want your agents to do.
Describe the channel, the questions and the process — I’ll come back with a plan and an estimate.
Message sent!
Thanks for reaching out — I’ll reply within a few hours.