The framework, the product and the marketing are 80% reusable in any situation

Every new industry takes 20% customization. In that 20% the data, the policies and the promise change.

80%

20%

Redacy is AI first, so agents will run most of the GTM

Through email marketing or similar channels. We're evaluating a specialist hire.

Contacts start at trade shows and trade associations, then professional bodies, funds and software vendors who already have those customers.

Outreach leads the professional to the free plugin, their way in. Companies start with FDE consulting.

The GTM strategy is iterative. A machine this adaptable lets us test many ICPs fast, like running a hundred startups in one

We position in 2 industries a month, each with a vertical solution sold on output: a promise with its own unit of result. Customers keep using the Claude or ChatGPT interface they know. What they buy is the outcome of the work done.

Each ICP is a hypothesis with its own promise, materials and revenue. The first ones are:

Workplace safety and acoustics consulting professionals (two already in the pipeline)
ESCOs (in the pipeline)
Public-contract service companies
Radiation protection experts for dental practices

We pick new ICPs on three things, in this order:

1Speed of acquisition
2Pain already paid for
3Easy to verify and repeat

We keep what brings revenue. Everything else changes target or promise and goes back into the cycle.

The customer journey

The steps every customer takes:

Discovery
Pilot
Transformation

1. Make the process data accessible

Done by

Professional: alone, with the Codex or Claude plugin

Company: with the FDEs

What the customer sees

Claude Code or Codex

Commercial phase

Entry point: the plugin is free, outreach leads to it

For companies, a service

What the customer thinks

Can ChatGPT really handle a job this specific?
How does my team do it today, and what does it cost?
My data is a mess. Is it too early?

Discovery
Pilot
Transformation

2. Write the policies and the automation target (the eval)

Done by

The customer, guided by the plugin or the FDEs

What the customer sees

Claude Code or Codex

Commercial phase

Same as step 1

What the customer thinks

What I know isn't written down. Who has time to write it?
How much work does it have to take off me to be worth it?
If I write it down, where does it end up?

Discovery
Pilot
Transformation

3. Move data and policies into the VM and get used to accessing it

Done by

Redacy moves in data and policies

The customer learns to log in

What the customer sees

The VM purchase and management screen, and the Claude Code or Codex desktop app connected to the VM (ssh)

Commercial phase

Setup, paid once

What the customer thinks

My data lives there now. Who can see it?
Will I be able to get in on my own?
If I stop or switch, do I start over from zero?

Discovery
Pilot
Transformation

4. Follow the agent as it learns, up to the eval target

Done by

The orchestrator learns

The customer corrects

In companies, the FDEs

What the customer sees

The orchestrator proposing and asking for a yes, the logs, the eval progress

Commercial phase

Pilot subscription, with support

What the customer thinks

It still gets things wrong. When will it stop?
Does correcting it cost me more than doing it myself?
I pay every month. Have I actually seen finished work?

Discovery
Pilot
Transformation

5. Trust the agent and extend it to other processes

Done by

The customer raises the autonomy

Each new process restarts from step 2

What the customer sees

Fewer confirmation requests, new agents on other processes

Commercial phase

Recurring subscription and expansion

For an industry, the joint venture

What the customer thinks

If I stop double-checking and it's wrong, do I pay for it?
What other work do I give it now?
Now I know how to train it. Do I still need the subscription?

Can you help with any of this?