AI automation

AI automation consultant: agentic workflows, custom GPTs and n8n-style systems.

I build agentic workflows that connect Claude, ChatGPT, MCP and the tools you already use. Custom GPTs for a team sit in the same engagement. Your team keeps the logins. A person approves anything that spends money, emails a customer, or publishes.

Illustration of an automation flow from a trigger through review steps into email, calendar and an approval list
What does an AI automation consultant do? An AI automation consultant designs the steps a repeated job should take, then builds them so the tools do the fetching and the drafting. I do that with Claude, ChatGPT, MCP and n8n-style automation. Agentic AI here means the workflow can take the next step and still stop for a person. I run a 19-agent Claude and MCP stack with three people for my own marketing. Most client builds start smaller.

TL;DR

I build agentic workflows for marketing and operations: custom GPTs, n8n-style automation, and Claude agents connected through MCP.

Your team keeps the accounts and the rules. Anything that spends, publishes, or emails a customer waits for a person.

The pattern is the 19-agent stack I run with three people. Customer facing bots are on the chatbot page. The marketing agent stack is on the AI agents page.

In 30 seconds

Vicky Lalwani is an AI automation consultant and custom GPT builder for teams in Australia, the UK, the US and the UAE. He connects Claude, ChatGPT and MCP to the tools a marketing team already uses, and builds n8n-style flows for the jobs in between. A person approves anything that spends or sends. His own stack is 19 agents and 3 people. Claude and MCP setup · AI agents.

Anthropic Claude Code 101: verify ↗

At a glance

Stack I run19 agents + 3 people
ModelsClaude and ChatGPT
ConnectionsMCP and n8n-style flows
Also buildsCustom GPTs for teams
ApprovalA person before spend or send
MarketsAU, UK, US, UAE
How I scope it

Your team keeps the keys.

I map the week first. The workflows worth building are the ones with a clear input and a clear output: a Monday report, a new lead routed to the right person, a search term list waiting for approval. I do not automate a decision you have not written down.

Custom GPTs are for a team that lives in ChatGPT and needs the same instructions every time. n8n-style automation is for the glue: form to CRM, alert to Slack, sheet to inbox. Claude with MCP is for jobs that have to read a live ad account, Search Console or GA4.

Human in the loop is a rule, not a slogan. If a step can spend money, email a customer, or publish, it stops. I run this way on a 19-agent stack with three people. Most teams start with two or three workflows and add the next one when the first is dull and reliable.

  • Custom GPTs for the SOPs your team repeats
  • n8n-style automation between CRM, inbox, sheets and ads alerts
  • Claude and MCP agents on Search Console, GA4 and ad accounts
  • ChatGPT workflows for agencies that already pay for the seats
  • Approval gates before anything spends, sends or publishes
  • Handover so your team can run the system without me
Three ways to automate

Custom GPT, n8n-style automation, or a Claude agent?

They solve different jobs. A lot of teams need two of them, not a new platform.

Custom GPTn8n-style automationClaude and MCP agent
Who it is forA team that already works in ChatGPTRepeatable steps across the apps you useMarketing work that needs live account data
What it doesFollows one set of instructions and filesMoves data and triggers the next stepReads Search Console, ads or GA4 and drafts the output
Human gateYou review the chat before you use itThe step that sends or spends waitsRequired before a publish or a bid change
Best forSOPs you want everyone to follow the same wayLead routing, alerts and CRM updatesAudits, search term reviews and reporting
Where it goes deeperThis pageThis pageClaude consultant
What's included

What an automation engagement includes

Job map

The repeated tasks, the hours they take, and which ones are safe to hand to a workflow.

Custom GPTs

Instructions, files and a review step, documented for the people who will use them.

App glue

n8n-style flows for routing, alerts and the updates your CRM should have received yesterday.

MCP connections

Claude reading the marketing tools you already pay for, with the least access that still works.

Approval design

The exact step that waits for a person, written down.

Handover

A short operating note and a session so the system is not stuck in my account.

Process

How we'll work together

Map

List the week and pick the two or three jobs with a clean input and output.

Prototype

Build them on your real tools, with the approval step in place.

Prove

Run them beside the old way until the output is something you would send.

Hand over

Document the logins, the rules and how to add the next workflow.

How I work

Clear terms, no surprises

What working together looks like before you sign anything.

You own everything

Ad accounts, analytics, data, content and assets stay in your name, always, including if we stop working together.

No minimum budget

No minimum ad spend. I'll tell you honestly what your budget can achieve on the free call.

3-month minimum, then monthly

Three months gives the work time to show results. After that it's month to month.

Senior work, direct access

I lead the strategy and I am your contact. A small team handles delivery. My AI agents cover the repetitive audits and reports.

Scoped and quoted up front

Pricing depends on channels and markets. You get a written scope and quote after the free 30-minute call.

Plain-English reporting

Regular reports tied to leads, bookings and revenue, not vanity metrics.

Proof

Results from similar work

Client names and domains withheld for confidentiality.

AI systems · In-house

A 19-agent AI marketing team

Claude + MCP agents connected to Search Console, GTM, Semrush and ad data, automating reporting, audits and research.

19+3
AI agents + humans
~15
Person-agency output
Healthcare · Dental

A dental practice's AI turnaround

Rebuilt the patient-acquisition funnel with paid search, AI booking and lifecycle automation.

+57%
Revenue in 90 days
−59%
No-shows
143
AI-booked appointments
FAQ

Questions people ask

What does an AI automation consultant do?

I design and build the repeated jobs your team should not do by hand: routing a lead, drafting a report, checking a tracking setup, preparing a search term review. The tools do the fetching. A person still approves anything that spends or sends.

Are you an agentic AI consultant?

Yes, in the practical sense. The workflow can take the next step on its own, and it stops where you said it must stop. I am not selling a model. I am wiring Claude, ChatGPT, MCP and your existing tools into a job you can supervise.

Do you work as a custom GPT builder?

Yes. I build custom GPTs for teams that want one set of instructions for a report, a brief, or a QA pass. The GPT includes the files and the review step. If the job needs live ad data, I use a Claude agent with MCP instead, because a GPT cannot see the account.

Do you work as an n8n automation consultant?

I build n8n-style automation: triggers, branches and handoffs between the apps you already use. The point is the job, whether the lead is routed or the alert fires. I do not sell a badge. I ship the workflow and leave you the keys.

What is human in the loop AI automation?

A person stands at the step that matters. The workflow may draft the email, the bid change or the page. It does not send, spend or publish until someone approves that step. That is how I run a 19-agent stack with three people.

Can you connect MCP servers to marketing tools?

Yes. Search Console, GA4, Google Tag Manager, ad accounts and similar tools, read only unless we have agreed a write step with approval. The Claude setup itself is on the Claude consultant page. This page is the workflow around it.

Do you build ChatGPT automation for agencies?

Yes. Agencies get the most from one GPT or one agent that runs the same way on every account. I document it so a new hire can follow it. Client data stays in the agency's own accounts.

How is this different from hiring an AI agent developer?

An AI agent developer engagement is the marketing stack: audits, reports, search terms. This page is the wider ops glue, including custom GPTs and n8n-style flows. A lot of teams book both. Chatbots that talk to customers are on the chatbot page.

Do you take freelance AI automation work in Australia, the UK and the US?

Yes, and in the UAE. The work is remote. I quote after a free 30 minute call, once I have seen the tools and the job. There is no price on this page, and you keep the accounts.

Let's talk

Let's build something that grows.

Tell me where you want to grow: search, ads, social or AI. I'll reply within one business day with honest next steps.