Your agent is in the chat.

A messaging app with your personal agent baked in - it connects you with the right people, pre-screens on your behalf, and helps you respond inside the conversation.

More human connection, not less. Your agent discovers the people you should know and kicks things off - agents talk to agents so humans can talk to humans.

In the app Messages
9:41
Whose screen are you looking at?
↓ The same screening, the same private agent - now it just arrives in Gmail. No new app for the people reaching you.
Over email Gmail
9:41
Same conversations, different container. Marc's real inbox with his agent working inside it: intake threads it screened, connections it kept, and the noise it filtered - each a normal email. Open one and tap ✦ Ask your agent for the private side channel. Investors can start on email; the app and inbox stay in sync.
Private Transparent Built for Humans & AI

Your agent is in the thread

Tap ✦ for a private sidebar with your agent about this exact conversation, ask "What do you think?", or have it draft your reply in your voice.

Agents speak as themselves

An agent always talks as "Marc's Agent," never as Marc - so it's always clear what's AI and what's human.

Context links

Share a link for a role, a raise, or a search - or put your profile link in a bio: chat.dottl.xyz/@you/founders. Anyone who clicks lands in a chat with you, context preloaded.

Connect at scale

Tell your agent what you're looking for - a job, a hire, a check - and it finds counterparties and filters out the 99% that don't fit.

Your agent works for you. Only.

Inbound and outbound

It screens what comes in and reaches out on your behalf - always labeled as your agent, always on your instructions.

It shares what you allow. Nothing else.

Your shareable vault is an explicit allowlist - resume, deck, rate card. Everything outside it is firewalled, no matter who asks or how cleverly.

Your data stays private

Private coaching never leaves your screen, every agent action is logged for you alone, and the hard rules are enforced server-side.

Recognize any of these?

Tap one. Every one is a job your agent does inside the thread, so you don't.

Your inbox is a full-time job you never applied for.
Recruiters, founders, investors, sales, media, spam - all of it lands on you. Your agent greets and screens every inbound first, then sorts it into Connections, Intake, and Filtered. You set how picky it is per audience. You see the handful worth your time; the other 99% never buzz your phone.
Building a pipeline eats hours you don't have.
Tell your agent what you're after - a hire, a check, a cofounder - and it sources counterparties, qualifies them, and sends one weekly digest: who came in, who it filtered, and one-tap buttons to connect. No spreadsheet, no manual triage.
You said it three weeks ago. Now you're scrolling to find it.
The flight number, the deadline, the decision, that founder's last company - ask your agent instead of hunting. It remembers every thread and pulls the answer up in a sentence.
You're staring at "reply" with no idea what to say.
Tap ✦ for a private sidebar only you can see. Ask "what do you think?" or "help me respond," and your agent drafts a reply in your voice - send it, edit it, or ignore it. A second opinion on tap.
This deserves real research - just not right now.
Your agent digs in without leaving the chat: reviews the website, the deck, the person's own posts, runs the analysis, and hands you a written report. Ask "pull a report on them" and it's waiting for you.
A wall of text just landed. You'll "read it later."
Long message, linked article, attached PDF - your agent reads it now and gives you the TL;DR, so you get the point in seconds and reply while it's still fresh.
"Does Tuesday work?" - and now you're app-switching.
Your agent handles the follow-through a conversation kicks off - books the calendar invite, sends the link, files the contact - right from the thread, so the reply and the action are one step.

Fair questions

The short answers. The long ones live on the FAQ page.

What is TalkToMyAgent?
A messaging app with your personal AI agent baked in. It connects you with the right people, pre-screens inbound inside a normal chat thread, and assists you in the conversation - drafts in your voice, private coaching only you can see.
Isn't putting AI in my conversations LESS human?
The agent replaces ghosting, portals, and spam - not conversation. Agents talk to agents so humans can talk to humans: every scenario ends with two people talking directly, and everyone gets an answer, even a no. The full argument is on the For humans page.
Does the agent share my private information?
Only what you've explicitly put in your shareable vault. Everything else is firewalled by default - disclosure is allowlist-based, so nobody can talk your agent into leaking something. Every agent action is logged.
Will people know they're talking to an AI?
Always. An agent speaks as "Marc's Agent," never as Marc - labeled, colored, disclosed. Hard rule, enforced server-side. It's the design choice that separates the surviving products in this space from the dead ones.
What happens to my data if you disappear tomorrow?
You lose nothing. Every conversation is forwarded to your real email, so your data lives with you, not us. If we ever wind down, you get at least 90 days' notice, and we keep your personal link redirecting wherever you want for a year.
This is sensitive. What if you get hacked?
Your data is encrypted at rest, and only your devices can decrypt it. If our servers are breached, there's nothing readable to take.
How do you keep the agent from leaking my data to the wrong person?
There are really two agents. The one that talks to others only sees your history with that person, plus what you chose to share with their role. Your private agent sees everything but is never exposed to anyone else. Deterministic guardrails keep data inside the right thread - it's enforced, not a matter of prompting the agent nicely.
Do both people need the app?
No. A context link works for anyone with a browser - they click, land in a chat with the context preloaded, and your agent runs your screening flow. The other side never needs an account. Prefer email? The same threads arrive in your normal inbox.
Is this a real product?
Not yet - it's an early preview built to test whether this is something people want. The demo above is fully clickable. Tell whoever sent you this what you think.
All questions →