TEA AIOS

An AI Operating System built for The Ecommerce Accelerator. A series of Claude agents the team uses to automate daily workloads, plus a custom application that scrapes data, displays live dashboards, and lets the team build public shareable agents. 40+ hours saved per week across the business.

The Ecommerce Accelerator (TEA) coaches and operates ecommerce businesses across the ANZ market. Their team was spending the bulk of every week on repetitive analysis, data pulls, reporting, and content production. Smart people doing the same manual work over and over.

We built TEA an AI Operating System with two pieces working together: a fleet of Claude agents the team uses every day, and a custom application sitting on top that handles scraping, dashboards, and public shareable agents.

Result: 40+ hours saved per week across the team, more time on real strategy, and a single source of truth for the whole business.

The TEA AIOS

At the foundation is a unified knowledgebase modelling TEA's actual business: the ICP, the playbooks, the brand voice, client data, and ongoing campaigns. Every Claude agent and every dashboard reads from the same brain, so the work is grounded in real context instead of generic AI output.

1. A Series of Claude Agents

The TEA team uses a library of purpose-built Claude agents to automate the workloads that used to eat their days. Each agent has its own role, prompts, and tool access, and all of them read from the unified knowledgebase.

• Growth strategy agent — drafts campaign strategies and positioning for new client engagements
• Creative brief agent — turns raw inputs into structured briefs for ads, content, and email
• Client onboarding agent — handles the kickoff workflow from intake to first deliverables
• Data and reporting agents — pull metrics, summarise performance, and surface anomalies
• Content and copy agents — write in TEA's brand voice for socials, email, and landing pages

The agents do the heavy lift. The team focuses on review, judgement, and client relationships.

2. The Custom Application

Sitting above the agents is a custom platform built for TEA. It is the place the team actually works: the agents run inside it, the data flows through it, and clients interact with parts of it.

The problem we solved

• Manual reporting — the team was rebuilding the same dashboards and pulls week after week
• Scattered context — client data, brand voice, and playbooks lived across tools, never in one place
• Capacity ceiling — growth was bottlenecked by how many manual hours the team could throw at it
• Inconsistent output — different team members produced subtly different work, hard to QA at scale
• No shareable layer — clients had no way to interact with the AI work TEA was producing

The outcome

• 40+ hours saved per week across the team — the equivalent of a full-time hire returned to strategic work
• One source of truth — every agent, dashboard, and client touchpoint reads from the same brain
• Faster client outcomes — work that used to take days now ships in hours
• A new product surface — public shareable agents have become a way TEA delivers and showcases value

Who this AIOS serves

Why this approach works

Agents grounded in real context

Every Claude agent reads from the unified knowledgebase, so output matches TEA's brand voice, ICP, and playbooks instead of generic AI text.

A platform, not point tools

Scraping, dashboards, agents, and shareable surfaces all live in one custom app. The team works in one place instead of stitching ten tools together.

Compounding leverage

Every new agent and every new dashboard plugs into the same knowledgebase, so the system gets sharper the longer it runs.

Capacity unlocked, not replaced

The 40+ hours per week saved went back into strategy, client work, and growth. The AIOS scaled the team without scaling headcount.

Technical architecture

• Reasoning layer — Claude as the model behind every agent in the system
• Knowledgebase — unified company context (ICP, brand voice, playbooks, client data) that every agent reads from
• Custom application — Next.js front end with auth, role-based access, and an internal admin layer
• Data layer — scrapers, APIs, and scheduled jobs feeding live dashboards
• Public surface — shareable agent endpoints with their own context windows and access control

Get in touch

If your team is bottlenecked on manual workloads or stuck stitching point tools together, we can build you the same kind of AIOS — agents, custom platform, dashboards, and shareable surfaces — tuned to how your business actually runs.

More from this sector