Name the work
Show us what arrives now, where it comes from, how quickly it matters, and the outcome you want.
Ten Thirty Two / Private transmitters
Every signal deserves its own door.
A customer request. An order. A note from the field. An update from your computer. We can set up a private transmitter for each approved source and give it a defined route into ChatGPT- and Codex-assisted work.
Start with one connection or plan several. We decide what each source may send, who may see it, and what happens after receipt before anything goes live.
Working reference: our owner-only operator relay is running with a reviewed request queue, status, and receipts. One authorized computer is connected to Codex today. Customer transmitters are individual setup projects, not an open self-service network.
THE ROUTE01 / MANY INPUTS, DEFINED PATHS
01Your websiteInquiry or order
02Your computerApproved workspace
03Field deviceChosen event
DEFINED CONNECTIONPrivate transmitterSource · access · limits · receipt
REVIEWED DESTINATIONAI-assisted work queueHuman-controlled next action
Example architecture. Each source and destination requires its own approved configuration and test.
A transmitter is a promise about what may cross.
It is a configured software connection for one named source and one approved purpose. Instead of forwarding everything into a chat, it carries only the request data you agree to share. The receiving workflow can record that it arrived, present it for review, and help prepare a response or task.
Our internal prototype accepts owner-approved requests, records status and receipts, and passes reviewed work to Codex through an authorized desktop connection. For clients, we design a separate route; the connector, permissions, capacity, and destination are chosen and tested for each business.
From signal to action
The request has a path. You keep the decision.
We design the handoff so a real person can tell where a request came from, what the assistant did with it, and whether anything still needs approval.
- 01
A source speaks
A device, site form, order system, or other approved workflow sends a request in the format we defined together.
- 02
The relay checks it
The connection is configured to recognize its sender, limit what it accepts, reject unexpected input, and record a useful receipt.
- 03
The work is reviewed
When the chosen tools and account permissions support it, ChatGPT or Codex can help sort, summarize, research, or draft the next move.
- 04
You approve the move
Publishing, purchasing, sending to customers, and other consequential actions follow the approval rules set for your project.
One or many
Add a door when the work earns one.
Some businesses need one intake route. Others need separate paths for orders, field updates, service questions, and internal operations. We can plan multiple transmitters, each with its own sender, purpose, access rules, and test record. The practical number depends on the accounts, provider limits, and support plan we agree on.
Tell us what needs to sendSet up with you
A useful relay begins before the connection.
Set the rules
Choose approved senders, permitted data, destination accounts, retention, limits, and who can act on a request.
Test one path
Use sample requests to check receipt, duplicates, failures, abuse resistance, recovery, and the human review step.
Expand deliberately
Add more connections only after the first route is understandable, supportable, and worth repeating.
Clear boundaries
A received message is information, not an order to obey.
The relay can carry a request; it does not give an unknown sender authority over your accounts. We set up sender checks, access limits, rate limits, retention, and a review path appropriate to the workflow before activating it.
ChatGPT and Codex are tools in the process. A connected assistant is not a guarantee of a continuously monitored inbox, an instant response, or autonomous fulfillment. We define monitoring and response expectations as part of the service. A device-based connection works only while the device and authorized account are online.
Approved request content may be processed by the chosen AI provider through the connected ChatGPT or Codex account. Before activation, we agree what may leave the source, who can access it, and how long records are kept.
This is separate from Ten Thirty Two’s Secure Relay private-messaging beta. That beta has its own release status and is not the customer transmitter network described here.
Start with one real signal
What do you want us to hear?
Tell Michael where the request begins, who should be allowed to send it, and what a useful response would look like. We will map the first transmitter with you, then decide whether you need more.
For this first conversation, describe the workflow. Do not email passwords, account tokens, payment-card details, or sensitive customer records.