For e-commerce agencies · Context · Agents

The operating system for your agency.

Wagon is an AI operating system for e-commerce agencies. It captures every client conversation, meeting and document, turns it into trusted context, and turns requests into tickets that AI agents research and complete, with your team reviewing every result.

Waitlist open
Abstract white data lines forming a flowing landscape on black
01

How Wagon works

Everything your clients tell you, remembered.

Client work is scattered across Slack, email, WhatsApp, calls and docs. The details that matter often live only in people's heads, and they go missing when someone is busy, away or moves on.

Wagon connects to the tools your team already uses, keeps every piece of evidence with its source and owner, and builds one current picture of each client that people and AI agents can rely on.

The context layer

One trusted memory for every client account.

Wagon preserves where every fact came from, connects people and requests across tools, and keeps client context current for your team and your agents.

Evidence lifecycle Capture → context
01Capture

Connect your tools

Messages, emails, meetings and docs flow in from the places client work already happens.

02Preserve

Keep the original truth

Every message and file is stored with its source, owner, timing and lineage attached.

03Resolve

Connect the dots

Duplicates merge, and every contact maps to the right client, request and decision.

04Remember

Build usable context

Source-independent client context stays current and ready for search and retrieval.

Put the context to work

The same client context powers every part of Wagon.

  • 01Tickets
  • 02AI agents
  • 03Reviewed drafts
  • 04AI tools via MCP
02

What Wagon does

From client request to invoice, in one place.

Wagon brings together the inboxes, ticket boards and timesheets agencies use to run client work, with AI agents doing the legwork and people making the calls.

01
An abstract routing system combining several inputs into one resolved output

Automatic triage

Wagon reads new client activity, spots actionable requests and creates tickets with the right people assigned, so nothing gets lost in a thread.

02
A chain-reaction machine representing dependable automated work

Agents that do the work

Claude or Codex agents research each ticket, check whether it is feasible, then carry it out in an isolated workspace, from pull requests to drafted replies.

03
A sculptural data system sorting scattered inputs into structured layers

Context for every AI tool

Search any client's history in Wagon, or connect Claude and other MCP-compatible tools so your team's AI answers from real client context.

04
A tactile composition of charts and performance indicators

Time and billing built in

Team timers and agent run time are logged against each ticket and roll up into per-client invoices, so the work agents do is visible and billable.

03

From request to review

Agents do the legwork. You stay in charge.

Every ticket follows the same visible path. Agents prepare and carry out the work; a person decides when it starts and reviews it before anything reaches a client.

  1. 01

    Capture

    A client asks for something in Slack, email, WhatsApp or on a call. Wagon picks it up with its source attached.

    Automatic
  2. 02

    Triage

    Wagon creates a ticket, links it to the right client and assigns the people who should own it.

    Automatic
  3. 03

    Prepare

    An agent researches the ticket, writes a brief and checks whether the work is something it can do.

    Agent
  4. 04

    Start

    When you move the ticket to In Progress, a Claude or Codex agent starts the work in its own isolated container.

    Your call
  5. 05

    Review

    Pull requests and drafted emails, Slack messages or events wait for your review. Nothing merges or sends by itself.

    Human review
  6. 06

    Bill

    Your team's time and the agent's run time stay logged against the client, ready to invoice.

    Per client
04

Questions

Clear answers before you join.

What Wagon does, who it is for and how it keeps your team in control of client work.

What is Wagon?

Wagon is an AI operating system for e-commerce agencies. It captures client conversations, meetings and documents, turns them into trusted context, and turns client requests into tickets that AI agents research and carry out. Your team decides when work starts and reviews every result.

Who is Wagon for?

Wagon is built for e-commerce agencies that manage many client accounts across Slack, email, WhatsApp and calls. It suits teams that want every client request captured, the context behind it kept in one place, and repeatable work handled by AI agents under human review.

Which tools does Wagon connect to?

Wagon builds client context from Slack, Gmail, Google Calendar, WhatsApp, Notion, Granola and Plaud, and connects to GitHub for repository work. Agents run on your team's own Claude or Codex subscription, and Claude and other MCP-compatible tools can read Wagon's context.

How does Wagon keep people in control of AI agents?

A person decides when an agent starts a ticket. Agents work in isolated containers with scoped tools, and repository credentials never reach the model. Finished work moves to In Review: pull requests never merge automatically, and drafted messages are only sent when a person sends them.

How is our client data kept separate?

Each agency runs on its own independent deployment rather than sharing a database with other agencies. Every record keeps its source, owner and timing, so the context your team and agents rely on can always be traced back to where it came from.

When can our agency start using Wagon?

Wagon is not yet generally available. Join the waitlist by emailing hi@thewagoncompany.com with your agency's name and website, and we will be in touch as places open.

Early access

Run your agency on what it knows.

Be one of the first e-commerce agencies to run client work on Wagon.

Join the waitlist

Or email hi@thewagoncompany.com

The Wagon Company

© The Wagon Company.

Daniel Goodwin

Co-founder · AI & engineering

Daniel Goodwin

RurocDemandlyProfitably

Daniel was VP of Engineering at Demandly, where he built the AI and data platform behind its demand-planning and profitability products - including dozens of integrations across Shopify, Amazon, TikTok and Xero.

Before that he led web development at Ruroc and Enginehawk, working across the Magento replatform, ERP implementation and custom data stack. He now focuses on turning complex business context into reliable AI systems and workflows.

Robbie Thompson

Co-founder · Technology & operations

Robbie Thompson

RurocStudio Underd0gDemandly

Robbie was co-founder and CTO at Demandly, an AI demand-planning startup. Before that he was Technical Director at Ruroc, where he helped scale the business from $1m to $50m in revenue.

He has led transformation across tech, web, data, and mobile, with deep experience redesigning operational workflows, implementing ERP systems and creating resilient technical foundations for fast-growing brands.