Prepared for your AI-first wealth education platform — April 2026

A team of agents,
not a chain of workflows

You have six platforms and workflows stitching them together. Every new workflow is another rigid pipeline to maintain. The answer isn't more workflows — it's AI agents that understand your business and have the tools to operate across it.

Who I am

A quick intro before we get into the architecture.

Alexander Girardet

AI Engineer — Agent systems specialist

I build AI agent teams that run real business operations — task management, client interactions, CRM, research pipelines. Each agent has tools to access your systems and skills written in plain English that define what to do. Not chatbots. Not workflows. Agents that operate as part of your team.

I've built this before

A personal assistant agent on WhatsApp — manages tasks on a kanban board, interacts with clients, and centralises CRM and ticketing into one conversational interface.

Same architecture we'd build for you: an agent with tools (CRM, task management, messaging) and context about the business. Not a rigid workflow — an intelligent operator that navigates your systems the way an employee would.

Two ways to build this

Most proposals will consolidate your N8N and Make.com workflows. Here's why that's solving the wrong problem.

More workflows

  • Fixed sequences — Step 1 → 2 → 3. Every edge case is a new branch to build and maintain
  • Rigid integrations — one workflow per connection. GHL to Supabase is one pipeline. Email to GHL is another. Each breaks independently
  • No context — workflows don't know your business. They move data between systems without understanding why
  • Linear scaling — N workflows = N things to maintain. Every new process is another pipeline to build

Consolidating workflows makes the spaghetti neater. It doesn't make it intelligent. You'll still be building new pipelines for every new process.

The agent team

An executive assistant at the top, specialist agents underneath. Each has tools to access your systems and skills defining what to do.
Manager Agent

Executive Productivity Agent

The top-level agent. Has visibility across your projects, calendar, inbox, and member pipeline. Surfaces priorities based on what's actually moving, connects dots across systems, and delegates to specialist agents.

morning-briefing priority-synthesis cross-project-context delegation-routing
GHL CRM Supabase Calendar Email (read) Agent dispatch
↓   ↓   ↓ dispatches to specialist agents
Specialist Agent

Email Triage & Drafting Agent

Triages the executive inbox — classifies, prioritises, drafts responses in the founder's voice, and routes delegated items to the right team member.

email-triage response-drafting sender-profiling daily-digest
Email API GHL contacts Supabase logs
Future Specialists

Identified by the audit

The audit maps your workflows and identifies which processes should become specialist agents. Content delivery, member onboarding, advisory scheduling — each becomes an agent with its own skills and shared tools.

How we build it

Three projects, three weeks. Each builds on the last.
01

Capability audit Week 1

Map every active workflow across N8N and Make.com. Identify legacy integrations replaceable by modern agent tools, Frankenstein workflows that should become agent capabilities, and where custom tooling is needed. Build the context layer — what each system holds, how data flows, what matters.

02

Agent roadmap delivered Week 1

Prioritised plan showing which workflows become tools, which become skills, and what order to build. This is the blueprint for your agent team — it defines the tools, context, and skills each agent needs.

03

Email triage agent Week 2

The first specialist agent on your team. Equipped with tools for your inbox, GHL, and Supabase. Triages, classifies, drafts responses in your voice, and routes delegated items. Your inbox becomes review-and-approve instead of active management.

04

Executive assistant Week 3

The manager agent. Sits above the email triage agent with visibility across your projects, calendar, and member pipeline. Surfaces priorities, manages context across projects, and reduces cognitive load. Not a reminder tool — an agent that connects dots across your business.

05

Expand the team Ongoing

The audit identifies the next specialists. Content delivery agents, member onboarding agents, advisory pipeline agents. Each uses the same tools and context layer — every new agent is additive, not a new system to maintain.

What you get

Three projects, each a working deliverable. Not documentation — working agents.

Capability audit

Full mapping of active workflows across N8N and Make.com. Identifies legacy integrations, redundant workflows, and custom tooling needs. Produces a prioritised roadmap to build your agent team.

Context layer

The business knowledge your agents need — what each system holds, how data flows, what matters. This is what makes agents intelligent operators instead of blind pipelines.

Email triage agent

Your first specialist. Triages, classifies, drafts responses, routes delegated items. Learns from your corrections. Turns your inbox from active management to review-and-approve.

Executive assistant

The manager agent. Surfaces priorities, manages context across projects, reduces cognitive load. Knows your member pipeline, content schedule, and advisory commitments.

Agent infrastructure

Shared tools (GHL, Supabase, email, calendar), skill files, and context layer. Every future agent you build uses the same foundation. No new systems, no new maintenance.

Three weeks, three deliverables

Sequential projects — each builds on the last. Fixed price, no hourly surprises.
Project 2 — Email Agent
$1,000
1 week
  • 3-tier email classification
  • Response drafting
  • GHL + Supabase integration
  • Delegation routing
  • Learning from corrections
Project 3 — Assistant
$1,000
1 week
  • Cross-project visibility
  • Priority synthesis
  • Morning briefings
  • Member pipeline awareness
  • Agent-to-agent coordination

Want to get started?

Reply on Upwork and we'll walk through the architecture on a call.
Happy to discuss which workflows are causing the most friction right now.