Skip to main content
> Case Study

AI Operations System

How a 6-person team got 22 AI agents covering 7 departments — for ~$50/month.

22
AI Agents
7
Departments
~$50
/ Month
6→60
Force Multiplier
> The Client

A 6-Person Manufacturing Startup

A US-based manufacturing startup bringing production back onshore. Six people covering everything: sales, fundraising, marketing, partnerships, supply chain sourcing, product development, and daily operations. Seed-stage. Moving fast. Every person wearing three hats.

Industry
Manufacturing / Reshoring
Team Size
6 people
Stage
Seed / Pre-Series A
Departments
7 (covered by 6 people)
> The Problem

Six People, Seven Departments

When your entire company fits around one table, knowledge lives in people's heads. Every meeting, every Slack thread, every investor call generates decisions and context that disappear the moment the conversation ends.

GAP-01

Context Evaporates

Decisions made in meetings were forgotten by Friday. Strategy discussions lived in scattered docs nobody could find. New context overwrote old context instead of building on it.

GAP-02

Everyone Does Everything

The same person writing investor updates was also qualifying factories and drafting outbound emails. No specialization meant everything got 60% effort, nothing got 100%.

GAP-03

Outbound Was Manual and Slow

Prospecting, email sequences, LinkedIn outreach, partner research — all done by hand. The team could generate maybe 20 personalized touches per week across all channels.

GAP-04

No Institutional Knowledge

Competitive intel, pipeline notes, strategic decisions, risk assessments — none of it was captured in a structured, searchable way. If someone left, their context left with them.

> The Solution

22 Agents. Shared Memory. $50/Month.

Instead of hiring, we built an AI operations layer. 22 specialized agents  each with a defined role, domain context, workflows, guardrails, and output format  covering every department. Every agent inherits the company's full context and builds on a shared knowledge base that compounds with every interaction.

22
Specialized AI Agents
7
Departments Covered
96+
Knowledge Vault Entries
~$50
Total Monthly Cost
Outbound (4)

Buyer prospecting, email sequences, LinkedIn outreach, manufacturer recruiting

Fundraising (4)

Investor research, data room management, pitch prep, financial modeling

Marketing (3)

Content writing, social media, SEO strategy

Partnerships (2)

Partner research, deal tracking

Supply Chain (2)

Factory qualification, trade analysis

Operations (4)

Meeting summaries, weekly reports, process docs, task coordination

Product (2)

Feedback synthesis, roadmap advisory

Three Ways In
Slack Bot

Any team member @mentions the bot or uses a /command. A lightweight classifier routes to the right agent automatically. Threads persist with the same agent.

Terminal CLI

Developers reference agents by path or describe the task. Claude Code finds the right agent file and loads its full context.

Autonomous DashboardLive

All 22 agents running on Railway. Browser-based, accessible to the full team. Agents query the knowledge vault, memory, and past outputs at runtime.

> The Results

What Changed

The system didn't replace the team  it gave each person the operational depth of a small department behind them.

Knowledge Capture

Before

Decisions lived in Slack threads and meeting notes nobody revisited

After

96+ structured vault entries — decisions, strategy, risks, pipeline, competitors — all searchable and automatically injected into every agent response

Outbound Capacity

Before

~20 personalized touches per week across all channels, manually crafted

After

4 dedicated outbound agents draft, sequence, and personalize at scale. Team reviews and sends — 10x the volume at higher quality

Meeting ROI

Before

Action items from meetings were rarely captured, never systematically tracked

After

Meeting transcripts auto-mined into vault entries — decisions, risks, and action items extracted and committed to git within seconds of pasting

Operational Consistency

Before

Every output (emails, reports, research) had a different voice and format depending on who wrote it

After

Shared brand guidelines + company context auto-loaded into every agent. Consistent voice across 22 agents and all team members

Institutional Memory

Before

Knowledge walked out the door with every departing team member or forgotten conversation

After

Auto-memory persists across sessions. Vector embeddings surface relevant past conversations. The system gets smarter with every interaction

Infrastructure Cost

Before

Hiring 2-3 junior ops people to cover the gaps: ~$150K-$200K/year + management overhead

After

Full AI operations layer: ~$50/month. n8n Cloud ($24) + Claude API (~$25) + Supabase (free) + OpenAI embeddings (~$1)

> Under the Hood

How It Works

The system is built on five layers. Each one is production-live and costs nothing beyond the ~$50/month total.

The Context Chain

Every agent knows everything — not because of a massive prompt, but because context flows through a chain: company facts → brand guidelines → specialized agent prompt → organizational knowledge vault → cross-session memory. Each layer inherits from the one before it.

01
CLAUDE.md

248 lines of company context, auto-loaded in every session

02
Brand Guidelines

Colors, fonts, voice — agents follow these automatically

03
Agent Prompts

22 specialized prompts with role, tools, workflows, guardrails

04
Knowledge Vault

96+ structured entries synced to Supabase for cloud agents

05
Auto-Memory

Vector embeddings enable semantic recall across sessions

The Slack Bot58 nodes

The team-facing interface. 58 nodes in n8n Cloud handle the full loop: receive message → classify intent → load agent prompt → fetch conversation history → query vector memory → inject vault context → generate response via Claude Opus → store conversation → embed for future recall.

Smart Routing

Haiku classifies intent; explicit /commands bypass for speed

Thread Persistence

Threads stay with the same agent across the conversation

Scoped Memory

Your past conversations surface in your context, not others'

Full Company Context

CLAUDE.md + vault loaded from Supabase at runtime

The Data Layer12 tables

Supabase (PostgreSQL + pgvector) with 12 tables. Agent prompts, vault knowledge, conversations, messages, vector memory, usage metrics, outputs — all queryable. Sync scripts push local changes to cloud. Agent outputs push-sync to the team's project database on every insertion.

Deployment Status

Everything is production-live.

22 Agent PromptsComplete
CLAUDE.md + BrandComplete
Knowledge Vault + HooksComplete
Supabase (12 tables)Live
Slack Bot (58 nodes)Live
Agent Output PipelineLive
Project Database SyncLive
Meeting MinerLive
Autonomous DashboardLive
7 CLI SkillsComplete
+
This is how a 6-person team operates like a 60-person team.

22 AI agents across 7 departments. Shared context and memory. Every output saved and searchable. Knowledge compounds with every interaction. Total infrastructure cost: ~$50/month.

Build Yours

Every company's operational DNA is different. The system above was built for a manufacturing startup — yours would reflect your team, workflows, and growth stage. Same architecture, your context.

Book a Discovery Call →