Proposal · June 2, 2026 Prepared by Jason · Axis Labs For: Aviation Finance Advisory Firm · Miami

Five autonomous agents.
One lean team.
Built to maintain.

Your post asked the right question first — which agent platform fits your use case best, and why. This proposal answers it directly, then walks the architecture, the multi-agent orchestration, and the phased rollout from one trial agent to five working in coordination.

5
Autonomous agents
12+
Lender relationships
30+
Broker partnerships
6
Regions · US, LatAm, EU, ME, Asia
The Architectural Reality

Three constraints that shape every decision.

Three things drive the platform choice and the build architecture for a 3-founder firm running a multi-region pipeline. Get any of them wrong and the agents become a liability instead of leverage.

🛬

Ops don't scale with headcount

12+ lender relationships, 30+ brokers, 6 regions. The bottleneck isn't strategy — it's the orchestration of repetitive, high-attention-to-detail tasks. That's exactly the work agents do well, and exactly why your scope is correct.

🧰

Platform choice locks in the cost curve

Pick the wrong agent framework and you're either paying for orchestration you don't need, or fighting visual-tool limits when you outgrow them. This is the most consequential decision in the project — which is why you flagged it as question one.

🛠️

Founders need to maintain it

A 3-person team can't run a black-box agent system. Every workflow needs to be readable, debuggable, and editable without calling the developer back. That requirement rules out several of the platforms you researched — and points cleanly at one.

Platform Recommendation

Click each platform. The recommended pick has a star.

You named seven platforms in your post. I've shipped production agents against most of them. Here's the honest comparison — strengths, weaknesses, and the verdict for your specific use case at your specific scale.

Scorecard At A Glance

Same criteria. All seven platforms.

Platform
Visual
workflow
Native
Gmail/Notion
Multi-agent
orchestration
Production
reliability
Cost
predictability
Non-dev
maintainability
Fit for
5-agent scale
Score
n8n + Claude API
13 / 14
LangGraph
8 / 14
CrewAI
7 / 14
OpenAI Agents SDK
7 / 14
Claude Managed Agents
7 / 14
Hermes
8 / 14
AutoGen
5 / 14
Recommended Architecture

n8n as orchestrator. Claude as the brain.

Hover any node in the diagram to inspect what it does and why it's there. The architecture is intentionally simple — fewer moving parts means fewer things to maintain.

INPUTS Broker List · Excel TRIGGERS Schedule · Webhook · Reply ORCHESTRATOR · LAYER 1 n8n Workflows Scheduling · Retries · Logging · Branching BRAIN · LAYER 2 Claude API Personalization · Ranking · Reasoning · Content GMAIL Send · Open · Reply NOTION CRM · Logs CALENDLY Scheduling DRIVE Docs Storage FOUNDERS · APPROVE · REVIEW · EDIT
Layer 1 · Orchestrator
n8n Workflows
The spine. Handles scheduling (7/14/30-day follow-ups), retries on API failures, branching logic, activity logging. Founders can read every workflow visually, edit prompts in-line, and debug without code. Hover any layer to inspect.
Live Demo · Trial Agent

Same broker outreach template. Three regions. Zero rewrites.

This is what Claude does inside the Broker Outreach agent: take your master template plus the broker's region, company, and personalized outreach angle, and produce an email that reads like you wrote it. Click between brokers to see it shift live.

Render For:
To:carlos@globalair.com.br
Subject:Financing partnership for São Paulo / Miami jet flows

Hi Carlos,

I lead an aviation finance advisory firm in Miami — we specialize in cross-border business jet acquisitions across the US and Latin America, with 12 active lender relationships.

I've been watching GlobalAir's deal flow into the São Paulo market this year — particularly the mid-cabin transactions back to US-based owners. That's exactly the corridor where our LatAm cross-border financing expertise tends to add the most value for brokers.

Would you have 20 minutes for an intro call next week? I'd love to understand what your typical buyer profile looks like and where financing tends to slow your deals down.

Best,
[Your name]

Region angle: LATAM cross-border financing Tone: Direct, business-forward, Portuguese-aware Generated by: Claude · 1.4s
The 5 Agents

One trial. Four follow-ons. One orchestrator.

Build the trial first. Validate against real broker outreach for one to two weeks. If it earns its place, agents two through five roll out one per week — each pulling reusable pieces from the trial's n8n workflows and Claude prompt library.

02

Client Document Chaser

Financing closes faster when docs land sooner

  • Document checklist auto-sent to financing clients per deal
  • 3-day follow-up cadence on outstanding items, reply detection halts cadence
  • Notion tracking against a master checklist per deal
  • Auto-flag when a package is complete — handoff signal to downstream agents
Phase 2 · Week 4 Reuses ~60% of trial logic
03

Lender Routing Assistant

12 lenders, ranked per deal, with reasoning

  • Deal parameters ingested via Notion form — model, year, price, buyer profile, region
  • Claude ranks all 12 lenders against your eligibility criteria with written reasoning per ranking
  • Draft outreach emails generated per eligible lender — paste-ready or auto-send with approval
  • All outputs reviewable in Notion before any send goes out
Phase 3 · Week 5 Reasoning-heavy use case
04

Marketing & Content Engine

LinkedIn, newsletter, conference prep

  • LinkedIn post generation from deal news and industry signals in your founder voice
  • Weekly email newsletter drafts pulling from active pipeline + market commentary
  • Conference prep materials — briefing docs, target lists, talking points per event
  • Always reviewed before publish — quality stays in your hands, volume goes up
Phase 4 · Week 6 Content + voice profile
05

Operations Coordinator

The orchestrator — agents talk to agents, founders see one view

  • Cross-agent handoff orchestration — broker positive response triggers deal creation, which triggers Doc Chaser + Lender Router automatically
  • Status monitoring across all 4 prior agents with failure alerts routed to founders in real time
  • Consolidated weekly operations report — pipeline state across the entire agent fleet on one page
  • Idempotent handoffs — duplicate triggers are detected and merged, no double-firing on the same broker reply
Phase 5 · Week 7 Multi-agent orchestration · the spine
Agent 5 · Multi-Agent Orchestration

How the agents talk to each other.

Agent 5 isn't a separate intelligence — it's the connective tissue. When Agent 1 detects a positive broker response, the orchestrator creates a deal record, triggers Agent 2 to start chasing documents, and triggers Agent 3 to rank lenders. Founders get one consolidated view of everything moving.

AGENT 5 Orchestrator Handoffs · Status · Alerts AGENT 1 Broker Outreach AGENT 2 Doc Chaser AGENT 3 Lender Router AGENT 4 Marketing Engine FOUNDERS Notion dashboard · weekly report positive reply → deal created → deal params → ← deal news
Ongoing Cost Profile

What the agents cost to run.

You asked for the ongoing maintenance and cost profile. Move the sliders to model your specific outreach volume — and watch how it shifts.

Active brokers in outreach
30 brokers
Follow-up emails per broker per month
3 per broker
Emails sent per month90
Claude API · personalization tokens~270K
Claude API monthly~$8
n8n (cloud or self-hosted)$0 – 50
Total monthly run cost~$58
Phased Rollout

Trial first. Validate. Then scale.

The trial agent runs the gauntlet for 1–2 weeks of real broker outreach before any other agent gets built. Once validated, the remaining four roll out at one per week. Total project window: 7 weeks end-to-end.

Phase
Wk 1
Wk 2
Wk 3
Wk 4
Wk 5
Wk 6
Wk 7
Platform doc + spec lock
Agent 1 · Broker Outreach (Trial)
Validation period (your eyes)
Agent 2 · Doc Chaser
Agent 3 · Lender Router
Agent 4 · Marketing Engine
Agent 5 · Operations Coordinator
Investment

Fixed-fee by phase. No surprises.

Trial fixed at the top of your stated range — the platform recommendation doc is part of the deliverable and that thinking compounds across every subsequent agent. Phases 2–5 quoted on the back end so you only commit after the trial earns its place.

Phases 2–5 · Roll-Out
$1,400 per agent · 1 week each

Top of your stated $800–1,500 range. Each agent reuses ~60% of the trial's n8n workflows and Claude prompt patterns.

  • Agent 2 · Client Document Chaser
  • Agent 3 · Lender Routing Assistant
  • Agent 4 · Marketing & Content Engine
  • Agent 5 · Operations Coordinator
  • Documentation per agent + handoff Loom
  • Optional retainer for tuning + new use cases
Total Engagement
$2,500 + ($1,400 × 4) = $8,100 total
Within your $5K–8K stated range
7-week timeline · phased commitment
Next Step

Let's walk the platform recommendation together.

A 30-minute call where I share my screen, walk through the platform tradeoffs against the 6 you researched, and confirm trial agent scope. I'll bring n8n open in one tab and a sample broker outreach workflow ready to demo. Ready to start immediately on selection.