Replies in your voice, everywhere you sell. Sent only when you say.
Programmable Social drafts replies, comments and first touches in each sender's own voice, on your own AI key. A person approves every one, and only the extension in your own browser ever sends. It never sends on its own.
Every send switch starts off. Your AI key, your browser, your call.
Each one off until you turn it on.
The extension acts on each platform's own website, in the tab where you're already signed in. A new platform starts in dry run with low caps, and pauses itself the moment that platform shows a warning.
Conversations, connection notes and comments, from your own signed-in tab.
Starts at 25/hour · 100/day
DMs and comments, sent through old.reddit's own forms, in a plain peer voice with no links.
Starts at 5/hour · 20/day
X
DMs and replies, held to 280 characters by X's own count.
Starts at 3/hour · 10/day
DMs and comments on posts and reels, short and in the thread's own tone.
Starts at 3/hour · 10/day
TikTok
DMs to people you can message, and comments on a video, up to 150 characters.
Starts at 3/hour · 10/day
Your offer, your people, your conversations.
Describe your offer
A workspace says what you sell, the motions you run (replies, first touches, follow-ups), the segments you sell to and the rules for each platform.
Add the people who send
Each sender is a real person. Upload their LinkedIn profile PDF, or paste a bio or a few things they've written, and their voice is drafted from it: how they write, what they've done, what they'd never say.
Capture, draft, approve
The Chrome extension captures the conversations you open on LinkedIn, Reddit, X, Instagram and TikTok. Drafts appear in the sender's voice. An owner or approver signs off before anything can go out.
Your voice, from your profile. Not a template.
Each sender is a real person, and drafts sound like them because they start from what they've written about themselves. Workspaces keep the offer straight; senders keep the voice; each platform keeps its own manners.
- From their own profile: the PDF LinkedIn gives you, read on your AI key. Edit anything it gets wrong.
- Each platform gets its own register and length: a plain peer voice on Reddit, under 280 characters on an X reply, 150 on a TikTok comment.
- Proof you supply (cases, numbers, quotes) is the only proof a draft can use. Nothing is invented to sound good.
Twelve years running field crews before moving into software. Writes short, plain sentences, asks one question at a time, and talks about dispatch the way a dispatcher would.
Under 80 words. One concrete number from the proof library, or none. End with one easy question.
Plain and peer-to-peer. Add a real point to the thread. No pitch, no links.
Examples. Your senders and workspaces are your own.
Nothing goes out until a person says so.
Every way a message could leave is behind a switch, and every switch starts off. The extension checks them all, the account's and its own, before each send, and holds if any one is off or can't be read.
Never sends on its own
The server has no way to send a message on any platform. The extension, in a person's own signed-in browser, is the only thing that acts, and only when the account's switches and its own both say so.
Approval before anything goes out
Every draft starts pending. Owners and approvers approve; editors draft. Skipping approval is a switch the owner has to turn on, and it starts off.
Do-not-contact, everywhere
One list for the whole install, checked before every draft and every send. If the check can't run, nothing goes out. Closing an account doesn't take anyone off it.
Autopilot drafts; it never sends
Each person opts in for themselves. It writes drafts for new replies, under hourly caps and a warm-up of hand-reviewed drafts, and the caps live in the database.
Clear the queue from the keyboard.
Approval is the one step that can't be skipped by accident, so it's built to be quick.
- j and k to move, a to approve, e to edit, c to send one back with a note, x to select.
- Approve or reject up to 50 at once. Each one still goes through every check on its own.
- Every edit is kept: who changed what, word by word, down to the text the extension actually sent.
- Works on a phone, with the approve bar under your thumb. Install it like an app.
Built to hold, not to blast.
Every check fails closed: if it can't tell, it doesn't send. Your accounts are worth more than any one message.
A browser agent on your key, on a short leash.
Platforms redesign their pages. When the extension's own code can't find the box, an optional agent can, using your AI key on our server. The extension checks its work before the one click that matters.
See what people answer.
Reply rates by platform, workspace and segment, meetings booked, and how much your team edits the drafts before they go. Test two openers and keep the one people answer. A weekly results email if you want one.
A cockpit for the conversations, not a blaster.
Everything a team needs to keep up with every platform without losing the thread, or its manners.
Inbox
Every captured thread on every platform, newest reply first, with a draft beside it, snooze and follow-up reminders.
Drafts
Replies, comments and first touches from pending to approved to sent, with attachments for context and saved replies to insert.
Contacts
Add by hand, import a CSV, or let the extension capture them. Tags, saved views and a merge for duplicates.
Follow-up sequences
Spaced touches that stop the moment someone replies. A step moves on when it's sent, not when it's drafted.
Meetings
Booked calls with a brief from the thread, and your real free times offered in replies when you connect Google Calendar.
Insights and experiments
Reply rates by platform, workspace and segment, how much people edit before sending, and which opener wins.
Scout
Finds reporters and podcast hosts posting about your topic, on a budget you set. Everything it finds waits for your review; it contacts no one.
Alerts
Email or Slack when replies land or drafts wait for you, a daily digest, and a weekly results email. All opt-in.
Search
Across every contact, message and draft, from the bar at the top.
Activity and usage
Who approved what and when, and what your AI key has spent this month, by feature and by person.
Webhooks and HubSpot
Replies, sends and booked meetings to your CRM or your own endpoint, signed, with retries. Off until you turn it on.
Browser agent fallback
When a platform changes its page, an agent on your own AI key can find the box and the button. It can only type the approved draft.
Your own AI key. Your own signed-in sessions.
We hold neither your model bill nor any platform password. The AI key is encrypted on our server and used only there; the extension works in the browser where you're already signed in.
- Anthropic or OpenAI, on your key, at your provider's price
- Keys encrypted at rest, never shown again or logged
- One extension key per person, paired with a one-time code, revocable any time
- Webhooks, Slack and HubSpot only to addresses that pass our checks
- The Chrome extension is the only part that touches any platform, and it's yours to read: it ships as plain source you load yourself.
Free while we're getting started.
Review, approval and the do-not-contact list are never limited on any plan. Model calls are billed to you by your AI provider, not by us.
- Senders, workspaces and your team
- Capture, drafts, approval and sequences
- Autopilot, opt-in per person and capped
- Every platform, Scout, webhooks and HubSpot
If paid plans come, you'll hear at least 30 days before anything you use stops being free.
Start draftingQuestions
Does it ever send a message on its own?
No. The server can't send anything on any platform. The Chrome extension is the only thing that acts, in your browser, as you, and only when every switch allows it: the draft is approved (or the owner has turned approval off), the platform is on, live sending is on for the account, and all of it is on in the extension too. Every one of those starts off. If any of them can't be read, the extension holds.
Which platforms does it work on?
LinkedIn, Reddit, X, Instagram and TikTok, through the websites, in your own signed-in browser. Each platform is off until the owner turns it on and you turn it on in your extension, starts in dry run (the extension fills the box but doesn't press Send), and has its own low caps.
Is this allowed on these platforms?
You're responsible for following each platform's terms and the rules on unsolicited messages where you work. X, Instagram and TikTok in particular restrict automated messages and act on accounts quickly. The extension works at a person's pace, under caps, and pauses the moment a platform shows a warning, but any automation can put an account at risk. That's why sending starts off, approval starts on, and caps start low.
Whose AI key does it use?
Yours. Connect an Anthropic or OpenAI key in Settings and every draft, reply, research step, meeting brief and browser-agent step runs on it, on our server, so the key never reaches a browser or the extension. We pay for no model calls and add nothing on top of them. Without a key, you can still capture, write and approve by hand.
What's the browser agent?
A fallback for when a platform changes its page and the extension's own code can't find the box. It's off until both the owner and your extension turn it on. It reads a description of the page's buttons and boxes on your AI key, on our server, and can only put the approved draft into a box: it can't type anything else, leave the platform's site, or touch settings, payments or sign-in. Before any click that could send, the extension itself checks the box holds exactly the approved text and the button belongs to that box. Every run is logged step by step.
How does it get my voice?
From your LinkedIn profile PDF (More → Save to PDF on your profile), or from a bio or writing sample you paste in if you're not on LinkedIn. It's read on your AI key into a short description of how you write and what you've done, which you can edit. Drafts follow it, in each platform's register, and a draft with nothing to say asks you instead of guessing.
What's the do-not-contact list?
Anyone on it is never drafted for or sent to, from any account on this install. It's matched on every spelling of their profile link, on every platform, and covers their other profiles once you link them as the same person. It's checked before every draft and every send, and a check that fails holds the message rather than letting it through.
Who can approve?
The owner and approvers. Editors capture conversations and write drafts; viewers read. Each person has their own extension key, so what's sent is attributed to them, and removing someone from the team turns off their extension.
What does it cost?
Nothing, for now: every feature is on, with no limits. If paid plans come, you'll hear at least 30 days before anything you use stops being free. Model calls are billed by your AI provider, to you.
How do I install the extension?
From Settings → Chrome extension: download it, open chrome://extensions, turn on Developer mode, press Load unpacked and pick the folder. Then get a one-time pairing code in Settings and type it into the extension's options page.
Can I take my data with me?
Any time: download everything in your account as one file from Settings. If you close your account, every switch goes off at once and you have 30 days to export before it's deleted.
Keep up with every conversation, in your own words.
Set up a workspace and a sender in a few minutes. Nothing is sent until you decide it should be.
Start drafting