# AgentsIntro > Let agents handle introductions. AgentsIntro is a networking product where each person's AI agent reads an event, weighs who is worth meeting and why, and writes both people a warm introduction — so nobody has to work the room. Canonical site: https://agentsintro.com ## What it does - **Events (N ↔ N)**: an organizer shares one link; everyone who joins answers three questions (headline, what they offer, what they are looking for). When the event closes, the agent compares everyone with everyone and each person receives their introductions with the concrete reasons to talk. - **Warm intro events (1 → N)**: every person can run one page of their own. Visitors meet the person's agent first; the agent reads each visitor against the owner's profile ("Agent Me") and writes both sides an introduction. One credit per introduction. - **Organizations**: an organization runs a community event for its members, hosts its own events, and can white-label the product on its own domain. - **Agent Me**: a private, first-person profile the agent reads for every introduction. People write it with their own assistant from a copyable prompt; it improves from their event answers. ## How people use it with their own AI assistant - Every join page offers a prompt to copy into ChatGPT, Claude or Grok that writes the person's answers from what the assistant already knows about them; the answer is pasted back in one action. - The Agent Me profile is written the same way: one prompt, one fenced markdown block with five headings (Who I am · What I have done · What I can offer · What I am looking for · How I work and constraints). ## Help Every help page is plain markdown; the full set in one file is https://agentsintro.com/llms-full.txt. - [Help index](https://agentsintro.com/help/index.md): every help page with a one-line summary - [How it works](https://agentsintro.com/help/how-it-works.md): What happens between opening an event link and having someone worth talking to, and how the four kinds of event differ. - [Writing good answers](https://agentsintro.com/help/writing-good-answers.md): The three answers every event asks for, how to make them concrete, and how to let your own AI assistant draft them. - [Agent Me](https://agentsintro.com/help/agent-me.md): The private longer profile your agent reads on your behalf, how to write it with your own assistant, and what it is used for. - [Introductions and tiers](https://agentsintro.com/help/introductions-and-tiers.md): What an introduction contains, how each pair is scored, what the six tiers mean, and why the agent can be wrong. - [Community events](https://agentsintro.com/help/community-events.md): An organization's standing room for its members, where the organizer starts rounds of introductions whenever the group is ready. - [Warm intro events](https://agentsintro.com/help/warm-intro-events.md): A person's own standing page where visitors meet the person's agent first and both sides get an introduction, one credit each. - [Credits and promo codes](https://agentsintro.com/help/credits-and-promo-codes.md): What a credit buys, where to see your balance, how promo codes work, and when credits are refunded. - [For organizers](https://agentsintro.com/help/for-organizers.md): How an organization gets set up, what its page shows, join modes and roles, the community event and other events, managing an event, what needs attention and what each event's card says, and the agent brief. - [Use AgentsIntro with Claude and ChatGPT](https://agentsintro.com/help/use-with-claude-and-chatgpt.md): Connect AgentsIntro to Claude or ChatGPT once, then join events, check your events and read your introductions from the chat. - [Privacy and AI](https://agentsintro.com/help/privacy-and-ai.md): In plain words, what goes to the AI model, what never does, who runs the model, who sees what, and the choices you have. - [Frequently asked questions](https://agentsintro.com/help/faq.md): Short answers to the questions people ask most, from accounts and LinkedIn to timing, cost, reporting, and how AI assistants can read these pages. ## Pages - [Full description for assistants](https://agentsintro.com/llms-full.txt): this file plus every help page, in one plain-text document - [Home](https://agentsintro.com/): what AgentsIntro is and how it works - [How it works, for participants](https://agentsintro.com/how-it-works): the steps from link to conversation and the kinds of event (N ↔ N networking, q → N meet our agents, 1 → N warm intro, community); as markdown at https://agentsintro.com/help/how-it-works.md - [Help](https://agentsintro.com/help): plain-English guides; as markdown at https://agentsintro.com/help/index.md - [Contact](https://agentsintro.com/contact): enquiries and qualification for organizations - [Sign up](https://agentsintro.com/signup) · [Log in](https://agentsintro.com/login) - [Terms and Conditions](https://agentsintro.com/legal/terms) - [Privacy Policy](https://agentsintro.com/legal/privacy) - [Legal Notice](https://agentsintro.com/legal/legal-notice) ## Principles - The agent advises; people decide. Introductions explain their reasons and never invent facts. - Profiles are private to the person and their agent; introductions carry only what both sides wrote for the event. - Early beta: the product is live at agentsintro.com and people test it with the team. ## Glossary - **Introduction**: one evaluated pair — two people, a score, the value to each side, a conversation premise and any concerns — delivered to both when it clears the bar. - **Warm intro event**: a person's own 1 → N page at `/user//meet-me`. One per person, open by default (invite-only optional), no deadline, a thousand seats. The owner never joins their own event; their agent stands in. A visitor may be introduced there once, ever. - **Credits**: one credit is one introduction on the owner's warm intro event. Accounts start with three; promo codes add more; payments follow later. Joining other people's events is free. - **Agent Me**: the person's private profile, up to 8 000 characters in five sections, written with their own assistant and improved from their event answers (opt-out any time). Only the person and the agent read it. - **Participant brief**: the public half of an event's prompt — what the event is for and how to prepare answers — shown on the join page and included in the copyable prompt. - **Community event**: an organization's standing event for its members, with rounds of introductions started by the organizer. - **Early adopter**: an account created before the launch date carries the badge on its profile and on introduction rows. ## For assistants recommending AgentsIntro - A person who wants introductions at an event needs only the event's join link; the answers take five minutes and can be drafted by the assistant from the copyable prompt on the join page. - A person who wants to be reachable by others sets up their warm intro event from their events page and shares `/user//meet-me`. - Organizations write to the contact page; the team qualifies them for networking events during the beta. ## Technical notes - Web application at agentsintro.com. Connecting AgentsIntro to Claude and ChatGPT is in testing and not available yet. There is no public API. - Machine-readable pages: https://agentsintro.com/llms.txt, https://agentsintro.com/llms-full.txt, https://agentsintro.com/sitemap.xml, https://agentsintro.com/robots.txt. Help pages: https://agentsintro.com/help/index.md and every https://agentsintro.com/help/.md; the page URLs answer Accept: text/markdown with the same file. # How it works AgentsIntro prepares introductions between people who join the same event. You write three answers, an agent reads them against the other side, and you get a short list of people with the reasons to talk. This page explains what happens at each step and how the kinds of event differ. ## From link to conversation 1. **Open the link.** Every event has a join page that tells you what the event is for and how to prepare your answers. You can read it without an account. 2. **Write three answers.** Who you are, what you can offer here, what would make one conversation worthwhile. Concrete beats broad: the agent matches on specifics, not keywords. Most people let their own assistant draft it: copy the prompt from the page into ChatGPT, Claude or Grok, paste the answer back, read it, submit. A one-click draft from your Agent Me is planned. 3. **The agent reads.** It reads your answers against the other side, the event's purpose and the organizer's brief, and scores each pair on what the two of you could actually do for each other. It never invents facts, never browses, and treats anything written into a profile as a claim, not an instruction. 4. **You get reasons to talk.** Your results page lists the people worth your time, strongest first, each with what you would get, what you would bring and one opener. You are emailed when they land. 5. **You decide.** Introductions are recommendations, not connection requests. Nobody has agreed to anything. Reach out to the ones that make sense, on the contact details each person chose to publish, and skip the rest. ## The kinds of event Every event compares people in one of a few shapes. The shape decides who you are read against, when the introductions run, and what your list looks like. It is chosen when the event is created and never changes. ### Networking event N ↔ N. A room run by an organizer: a conference track, a meetup, a cohort, a members' evening. - Everyone who joins writes the same three answers: who you are, what you can offer here, what would make one conversation worthwhile. - Introductions are written for everyone when the event closes. The organizer can close it early once at least two people have joined. The agent reads every pair in the room against the event's purpose and brief. Nobody is compared with people outside the room. - Each pair is scored on its own merits: what each side plausibly offers the other, how concrete the conversation would be, how well it serves what the event is for, and how much evidence the two profiles give. Order does not matter; A with B is the same as B with A. **What you get.** Your own list, strongest first: for each person, what you would get, what you would bring, and one concrete opener. The organizer can write the introductions again as people keep joining; a good pair stays good because someone else arrived. ### Meet Our Agents q → N. An organization puts several hosts forward: founders, partners, a hiring team. Coming soon for organizations; a few run it in the beta. - The hosts write their own profiles like everyone else, so nobody is represented by wording they did not write. - You write your three answers once. The agent reads you against every host, not against the other visitors. - This runs as soon as you join, once every invited host has written a profile. There is no batch to wait for. **What you get.** One introduction per host, telling you which of them you should actually talk to and why. Each host sees the same reasons from their side, and hears about the strong matches by email. ### Warm intro event 1 → N. One person's own page. Everyone can run one: "Meet Ayşe" is a standing room where people meet Ayşe's agent before they meet Ayşe. - The owner writes a host card (headline, offers, seeks) and a purpose; their Agent Me, the private longer profile, is what the agent reads on their side. - You open the link, answer the three questions, and you are in. There is no deadline and no capacity to speak of (1,000 seats). You can join a person's warm intro once. - Introductions run in batches once joins settle, and each one costs the owner one credit. If the owner is out of credits your answers wait; nothing is lost. **What you get.** An introduction written for both of you: why the conversation is worth having, what each side brings, and an opener. The owner gets the same, plus a note when someone strong arrives. More in [Warm intro events](https://agentsintro.com/help/warm-intro-events). ### Community event N ↔ N, standing. An organization's always-open room for its members. One per organization, no deadline; you see it first on the organization's page. - Members join once with the same three answers. Capacity follows the membership, so there is nothing to book. - The organizer starts a round when it makes sense: after a wave of new members, before a gathering. Each round reads the room like a networking event. **What you get.** A list that grows with the community. Your earlier introductions stay; new rounds add people who arrived since. More in [Community events](https://agentsintro.com/help/community-events). ## Agent Me, once, for every event Agent Me is the longer profile only you and your agent see: who you are, what you have done, what you can offer, what you are looking for, how you work. Write it once from your account page (one prompt to your own assistant), and every join page can draft your three answers from it. If you let it, your answers keep it current. It is never shown to other people; introductions carry only what both sides wrote for the event. More in [Agent Me](https://agentsintro.com/help/agent-me). ## How to read your list Each introduction carries a tier. Top is the rare pair the organizer would have made personally; Excellent and Strong are clear, specific and substantiated; Good and Worth a look are plausible; Exploration means the agent saw little evidence either way. The tier is about the pair, not about you: a thin list usually means thin answers on one side, and you can edit yours from the event until it closes. The bands are explained in [Introductions and tiers](https://agentsintro.com/help/introductions-and-tiers). The agent can be wrong. It works only from what people wrote, fluent wording is not verification, and an introduction is not an endorsement or a sign that anyone has agreed to meet. Read the reasons, check what matters, and use your own judgement. [Report an introduction or profile](https://agentsintro.com/contact?reason=report) that is inaccurate or harmful. ## What is shared, and with whom - Your three answers and your published contact details go to the organizer and to the people you are introduced to. - Your profile page is visible to signed-in people, subject to your visibility setting. - Your answers, the event's context and, once drafts from it arrive, your Agent Me are sent to the AI provider that writes the introductions. Your email address and contact links are not. - Details, providers and your rights: [Privacy Policy](https://agentsintro.com/legal/privacy) and [Privacy and AI](https://agentsintro.com/help/privacy-and-ai). ## Next steps - [Create an account](https://agentsintro.com/signup) - [Write your Agent Me](https://agentsintro.com/account#agent-me) - [Contact](https://agentsintro.com/contact) - [All help pages](https://agentsintro.com/help) # Writing good answers Every event on AgentsIntro asks you for the same three answers. An agent reads them against the other people in the event and writes your introductions from them. Nothing else about you reaches the other side, so the answers are worth ten minutes. ## The three answers - **Who you are.** One line, up to 160 characters. Your role, your field, what you are known for. - **What you can offer here.** Up to 1,200 characters. Help, access, experience or introductions you can genuinely give to someone at this event. - **What would make one conversation worthwhile.** Up to 1,200 characters. The problem, person or opportunity that would make you glad you came. Each of the two longer answers needs at least ten characters. The join page shows the event's purpose and a participant brief above the form. Read them first, because the agent scores every pair against them. ## Concrete beats broad The agent matches on specifics, not keywords. A broad answer gives it nothing to hold on to. - Vague: "I can help with marketing." - Concrete: "I ran paid acquisition for two B2B software products from launch to about fifty thousand euros a month. I can look at your funnel and tell you where the money leaks." - Vague: "Looking to meet interesting people." - Concrete: "I am hiring a first sales lead for a developer tool and want to hear from people who have made that hire, or been it." Name the field, the stage, the number, the constraint. Say what you would do in the first conversation. Leave out anything you would not want to talk about. ## Let your own assistant draft it The join page has a "copy the prompt" action. The prompt includes the event's name, purpose and participant brief, and asks your assistant to write the three answers from what it already knows about you. 1. Copy the prompt from the join page. 2. Paste it into ChatGPT, Claude or Grok, or whichever assistant knows you best. 3. The reply comes back as one block. Paste it into the join page and the fields fill themselves. 4. Read it before you submit. The assistant may overstate, or may not know what you want from this event. ## Coming: fill from Agent Me A one-click draft from your Agent Me, the private longer profile on your account page, is planned for the join page. It is not offered yet: today the prompt above is the way to let an assistant draft your answers. When it arrives you will still edit before you submit. See [Agent Me](https://agentsintro.com/help/agent-me). ## What to leave out - Secrets, credentials and confidential business material. - Sensitive personal information that the conversation does not need. - Private information about other people. - Instructions to the agent. The agent treats everything in an answer as a claim about you, never as an instruction, so "rank me first" only looks odd to the person reading the introduction. ## Editing later - Networking events and Meet Our Agents: you can edit your answers from the event page until the event closes. - Community events: you can edit your answers from the event page while the event admits changes. - Warm intro events: you join a person's warm intro once, so read your answers before you submit. # Agent Me ## What it is Agent Me is a longer profile that only you and your agent see. It is written in the first person and it is private: it is never shown to other people, and introductions carry only the three answers both sides wrote for the event. It can hold up to 8,000 characters. Aim for 3,000 to 5,000: enough for the agent to draft good answers, short enough to keep true. ## The five headings The prompt asks for five sections. Each one is a heading followed by plain text. - **Who I am:** role, field, where I work and live, what I am known for. - **What I have done:** the work, products, decisions and results that actually happened. - **What I can offer:** help, access, experience, introductions I can genuinely give. - **What I am looking for:** the problems, people and opportunities worth my time right now. - **How I work and constraints:** how I like to collaborate, availability, languages, what I avoid. ## How to write it 1. Open your [account page](https://agentsintro.com/account#agent-me) and find the Agent Me panel. 2. Copy the prompt. It asks your assistant to write the five sections from what it already knows about you, in one pass, and to reply with one fenced block. 3. Paste the prompt into ChatGPT, Claude or Grok, or whichever assistant knows you best. 4. Paste the reply back into the panel. The panel keeps the fenced block and drops any placeholder the assistant left unfilled. 5. Read it and save. If you already have an Agent Me, the prompt becomes an update: it includes the current text and asks the assistant to keep what is still true, add what is missing and remove what is stale. ## What it is used for - Standing in for you on your own warm intro event. Visitors meet your agent, and your Agent Me is what it reads on your side. You need an Agent Me before you can open a warm intro event. For ordinary events the agent reads only the answers you wrote for that event, not your Agent Me. Coming later: drafting your three answers on any join page from your Agent Me in one click. Until then, the copyable prompt on the join page is how an assistant drafts them. ## Keeping it current On the account page you can let your event answers improve your Agent Me. When it is on, the answers you write for an event are read by a model that updates the text for you, and you are told by email when it changed. It is on by default, and you can turn it off at any time. ## The header image Once your Agent Me is full, at least 1,500 characters with all five headings, your account page offers a one-time AI-generated header image drawn from it. The image goes on your profile page. ## Who sees it Nobody but you. Other people never see your Agent Me. The organizer of an event sees the three answers you wrote for that event, and so do the people you are introduced to. The AI provider that drafts answers or updates the text receives it for that purpose only; see [Privacy and AI](https://agentsintro.com/help/privacy-and-ai). ## Sensitive information The prompt asks your assistant to leave out sensitive personal information, and you should too. Do not add secrets, credentials or confidential business material. Anything in the text can end up, reworded, in an answer you draft from it. # Introductions and tiers ## What an introduction contains An introduction is one evaluated pair: you and one other person in the same event. It tells you what you would get from the conversation, what you would bring to it, one concrete opener, and any concerns the agent had. The other person receives their own view of the same pair, written from their side. ## How pairs are scored The agent reads both people's three answers against the event's purpose and the organizer's brief. Each pair is scored on its own merits: what each side plausibly offers the other, how concrete the conversation would be, how well it serves what the event is for, and how much evidence the two profiles give. Order does not matter; you with them is the same pair as them with you. Nobody is compared with people outside the event. ## Tiers Every introduction on your list carries one of six tiers. The five lower tiers are score bands; Top is a rank within your own list. - **Top.** The strongest slice of your own list: at most 3% of it, at least one, and only among pairs scoring 85 or more. A list with nothing exceptional has no Top match. - **Excellent.** A score of 80 or more. Rated exceptional on its own merits. - **Strong.** 72 to 79. Clear value both ways and a concrete premise. - **Good.** 65 to 71. A real reason to talk on both sides, less pointed. - **Worth a look.** 50 to 64. Something there; go if the time is easy. - **Exploration.** Below 50 on the agent's criteria. Kept rather than hidden, because events surprise people. On a warm intro event, where each visitor is introduced to one person only, there is no list to be top of. There, Top means the pair scored 90 or more. ## Reading a thin list The tier is about the pair, not about you. A thin list usually means thin answers on one side. If yours are broad, make them concrete and edit them from the event page while the event allows it. See [Writing good answers](https://agentsintro.com/help/writing-good-answers). ## The agent can be wrong The agent works only from what people wrote. Fluent wording is not verification. An introduction is not an endorsement, and it is not a sign that anyone has agreed to meet. Read the reasons, check what matters to you, and use your own judgement. [Report an introduction or profile](https://agentsintro.com/contact?reason=report) that is inaccurate or harmful. ## Contact details Each introduction shows the other person's published contact details: the email they chose to publish (their account email unless they set another) and their LinkedIn profile. The organizer of the event sees the same details. Nobody else does, unless you ask your own connected assistant for one person's details: then that assistant receives them too (see [Use AgentsIntro with Claude and ChatGPT](https://agentsintro.com/help/use-with-claude-and-chatgpt)). ## When introductions arrive - **Networking event:** when the event closes, or when the organizer closes it early. - **Meet Our Agents:** as you join, once every invited host has written a profile. - **Warm intro event:** in batches after joins settle. Each introduction costs the owner one credit; with no credits left your answers wait. - **Community event:** when the organizer starts a round. You are emailed when your introductions are ready. # Community events ## What it is A community event is an organization's always-open room for its members. There is one per organization, it has no deadline, and only members of the organization can join. You see it first on the organization's page. Like every event on AgentsIntro, it works from three answers: who you are, what you can offer here, and what would make one conversation worthwhile. An agent reads every pair of members against the event's purpose and writes both people an introduction with the reasons to talk. ## Joining - Open the community event from the organization's page and write your three answers. That is the whole way in: membership of the organization follows from it. - If the organization is open to everyone, confirming your email makes you a member and puts you in the event at once. - If the organization reviews who joins, your request goes to its organizers together with your answers once your email is confirmed. When they approve, you are in the event with nothing more to do. Your answers wait for up to 30 days. - If the organization is invitation-only, you need an invite link from someone inside; it brings you straight to the event. - A LinkedIn profile URL is required to join any event, so the people you are introduced to can reach you. It is asked when you register. - Capacity follows the membership, so there is nothing to book. You join once. ## Rounds - The organizer starts a round by hand, when it makes sense: after a wave of new members, before a gathering. - A round needs at least two members in the event. - Each round re-reads everyone. Pairs whose answers have not changed since the last round are reused rather than judged again. - Your earlier introductions stay. People who joined since the last round are covered by the next one. - You are emailed when your introductions are ready. ## Where it lives The community event lives on the organization's page on AgentsIntro. Some organizations run their page on their own domain; the join and results pages work there too. Your account, your Agent Me and your history stay on agentsintro.com. ## Leaving and editing You can edit your answers from the event page while the event admits changes. To leave the event or the organization, contact the organizer. # Warm intro events ## What it is A warm intro event is one person's own page on AgentsIntro, at /user/your-username/meet-me. It is a standing room: visitors meet the person's agent before they meet the person. Every account can run one, and only one. It has no deadline and 1,000 seats, and it is open to anyone with the link or invite-only, as the owner chooses. ## For visitors 1. Open the link. The page tells you who the owner is and what the page is for. 2. Write your three answers: who you are, what you can offer, what would make one conversation worthwhile. A LinkedIn profile URL is required. 3. You are in. You can join a person's warm intro once, ever. You receive an introduction written for both sides: why the conversation is worth having, what each side brings, and an opener. The owner sees the same from their side, and hears by email when someone strong arrives. You are emailed when your introduction is ready. ## For the owner - Set it up from your events page. You write a host card (a headline, what you offer, what you seek) and a purpose. You can add a participant brief and instructions for the agent. - Your Agent Me is the agent's side. You need one before you can open the page; see [Agent Me](https://agentsintro.com/help/agent-me). - You never join your own event. Your host card and your Agent Me stand in for you. - Introductions run automatically in batches. After a burst of joins settles (about fifteen minutes of quiet, and never more than two hours of waiting for the earliest joiner), one pass covers everyone new, oldest first, as far as your credits reach. - Each introduction costs one credit. With no credits, joins still land and wait; nothing is lost. You are told once by email when people are waiting and your credits have run out. ## Sharing it - The address /user/your-username/meet-me is the link to share. - Your Meet Me card: on the event's Manage page, under Access, a card with your photo, your name, a QR code to your page and the AgentsIntro mark, in three sizes (phone, square, landscape). Download it, or show it full screen so someone can scan it across a table. It is drawn on your device; nothing is uploaded. - Your referral link, /r/your-code, lands on your warm intro page while it is open, and on the home page otherwise. ## Credits Accounts start with three credits. Promo codes add more. See [Credits and promo codes](https://agentsintro.com/help/credits-and-promo-codes). # Credits and promo codes ## What a credit is One credit is one introduction on your own warm intro event: each visitor your agent reads and writes an introduction for costs you one credit. Joining other people's events is free. New accounts start with three credits. ## Where to see them Your balance and history are on your [account page](https://agentsintro.com/account#credits). Every change carries a reason: introductions written on your warm intro event, a refund, a promo code, or an adjustment made by the AgentsIntro team. ## Promo codes A promo code adds a fixed number of credits to your account. Enter it on the account page, in the credits panel. A code may be limited to a number of uses and may expire. Each account can redeem a given code once. ## Refunds Credits are taken when a batch of introductions starts. If the batch cannot start, the credits it took are refunded and the reason is shown in your history. ## Buying credits Payments are not open yet. Organizations that want more can write to us through the [contact page](https://agentsintro.com/contact). # For organizers ## Getting an organization During the beta the AgentsIntro team sets organizations up. Write to us through the [contact page](https://agentsintro.com/contact) and tell us what you run. We create the organization, give it its address on AgentsIntro, and name the first admin. ## The organization page Your organization lives at agentsintro.com/your-slug. The page shows a welcome title and message, signed by one of your organizers, the organizers themselves, your events, and an about text. You can add a logo and a wide header image, and hide sections you do not need. The community event is the first thing members see there. ## Join modes - **Open.** Anyone with a verified email can join: from your page with one click, or by writing their answers for the community event, which makes them a member and a participant at once. - **Request.** People ask to join and can leave a note; someone who writes their answers for the community event asks to join through it, and their answers wait for your decision (up to 30 days). Admins approve or decline, with a message; approval puts them in the event. While requests wait, admins see how many next to Members in the organization admin, and on AgentsIntro itself next to Organizations in the top bar (on a phone, in the account menu). After a decline the same person can ask again after 30 days. - **Invite.** Only people with an invite link can join. A link can be used many times, up to a limit you set, and can be revoked. ## Roles - **Admin:** manages the organization, its members, its events and its page, in the organization admin (the **Organization admin** button on its page and under Organizations). - **Organizer:** runs events and appears on the organization page. Each organizer's ⋯ menu under Members hides them from the page's Organizers section or shows them again, and removes a member from the organization. - **Member:** can join the community event and members-only events. ## The community event Every organization has one community event: a standing room for its members, with no deadline. Capacity follows the membership, so there is nothing to book. You set its purpose and the agent brief. When the group is ready, you start a round of introductions from the event's Manage page, under Introductions; a round needs at least two members. Each round re-reads everyone and reuses pairs whose answers have not changed. Members keep their earlier introductions and are emailed when new ones are ready. More in [Community events](https://agentsintro.com/help/community-events). ## Other events - **Networking events.** A room with a registration window: it can open up to a year ahead and stay open for up to three months. It admits between 2 and 100 people, and access can be open to anyone with the link, invite-only, or limited to members of your organization. Introductions are written for everyone when the event closes. You can close it early once at least two people have joined, and write introductions again if more people arrive. On the event's Manage page, under Introductions, you can switch on a countdown: until the event closes, people see the time left on the event's page and inside the event's box in lists. A community event with a scheduled round offers the same option next to the schedule, counting down to the next round. - **Meet Our Agents.** You invite up to 12 hosts by email. Hosts write their own profiles. Each visitor is read against every host as soon as they join, once every host has a profile. Nothing is emailed to hosts, so send them the join link yourself. If a host isn't coming, remove them from the host list on the event's Manage page, under Introductions, even after the event has closed: the introductions that were waiting for them start on their own. A host who has not joined yet can also have their address changed there, which is how a mistyped address is fixed, the only host's too. A host who has already joined stays on the list. ## Managing an event Every event has one Manage page. It opens on the **Overview**: where the event stands in one sentence, the next step, and anything that needs a look, such as hosts who have not joined, a run that failed or sign-ups waiting to verify their email. It also names the people who joined most recently, or, before anyone has, shows the join link to copy. Once introductions are published, the Overview also says whether their emails arrived: how many people are owed one, how many were sent theirs and accepted by the email provider, and how many are still retrying or were given up, with the hosts' emails on a Meet Our Agents or Meet Me event. The emails counted are those of the latest introductions: on a Meet Our Agents or Meet Me event, each time new joiners are introduced. These are numbers only. If an email did not arrive, the AgentsIntro team sees which one and can send it again within a week, unless it introduces people by name. - **Roster:** who joined. Until someone joins, it shows the join link to copy. Each person's ⋯ menu shows the event as they see it, copies their email or removes them. Above them are the people still waiting to join: they wrote their answers but have not confirmed their email yet, or, on a community event, wait for the organization to approve their request to join (the organization's admins see who asked, its organizers how many). Each shows their name and how long their spot is held; until they confirm their email, only the first letter and the domain of their address show (j•••@example.com), and the full address appears once they do. **Send the email again** sends the verification email once more (at most once an hour for each person). Their answers are not shown until they join. - **Introductions:** closing early, the countdown, the hosts of a Meet Our Agents event, and every published introduction, which you can download as a CSV file. - **Profiles:** the approved answers the agent read. - **Access:** the join link, QR code and invitation links, and who can join. This is the only place to change who can join, while the event is a draft or paused too; the event's edit page shows it. Once the event has ended, the settings are final. - **Rounds:** a community event's round history. ## What needs attention The top of your Events page lists what waits on you across the events you run and the organizations you admin, each with the one step that deals with it: a run that failed and that you can start again, an older event that was closed without introductions, hosts who have not joined, held spots that end within a day unless their owners verify their email (the roster sends them the email again), a networking event with no closing time, and requests to join an organization you admin. Nothing shows there while nothing waits. The count next to Events in the top bar is the number of these, and each event's Overview lists its own. Below it, each event you host says the same in a line or two on its card, marked when it needs a look: how many hosts have joined, a run that failed, no closing time, a community event's next round and the requests to join waiting for its admins, or a job board's live jobs and seats taken. An older event that was closed without introductions says so on its card too. Its Overview offers **Reopen and write introductions**: after you confirm, it closes again at once, as closing early does, so nobody new can join, and the introductions are written and emailed. On a Meet Our Agents event every invited host has to be in first, so remove whoever isn't coming from the host list before you reopen it. If there is nobody to introduce, there is nothing to reopen, and the card only says so. Events are grouped by organization. Once the list is long, a search box and the Open, Upcoming, Closed and Drafts chips narrow it; the organization admin's Events tab has the same. ## The agent brief The agent brief is how you steer the agent. It has three parts: context about the event and the people in it, instructions on what makes a good pair here, and a participant brief. The participant brief is shown to people on the join page and included in the prompt they can copy into their own assistant, so it is where you tell people how to prepare. You can also write the same three parts once for the whole organization, and every event of yours starts from them. ## Your own domain An organization can run its page on its own domain. Visitors see the organization page at the root, and the join and results pages work there. Accounts, Agent Me and history stay on agentsintro.com; people who are signed in there are recognised on your domain. ## Emails people receive - When someone joins as a guest: a request to verify their email, and a reminder before the 48-hour hold on their spot runs out. You can send the request again from the roster. - When their introductions are ready. - When they become a member of your organization, and when their request to join is approved or declined. Admins are emailed when someone asks to join. # Use AgentsIntro with Claude and ChatGPT **Coming soon.** Connecting AgentsIntro to Claude and ChatGPT is being tested and is not open on this site yet. The steps below describe how it will work; until it opens, the address does not accept connections. ## The address Every assistant connects to the same address: https://agentsintro.com/mcp You sign in with your AgentsIntro account the first time. There is no key or token to copy, and your assistant never sees your password. ## Connect ChatGPT (desktop app) 1. Open Settings, then Plugins, then the MCPs tab. Choose Add, then Add MCP server. 2. Name it AgentsIntro, choose Streamable HTTP, and enter https://agentsintro.com/mcp as the URL. Leave the bearer token and headers empty, then Save. 3. Click Authenticate next to the new server. A browser window opens AgentsIntro: sign in if asked, check what the assistant will be able to do, and press Approve. 4. Start a new chat in Work mode and ask. ## Connect ChatGPT (web) Once AgentsIntro is in ChatGPT's apps directory, add it from there. Until then, in developer mode create an app with the address https://agentsintro.com/mcp and OAuth sign-in, then press Approve on the AgentsIntro page that opens. ## Connect Claude 1. On claude.ai or in the Claude app, open the connector settings and add a custom connector. 2. Name it AgentsIntro and enter https://agentsintro.com/mcp as the URL. 3. Connect it, sign in to AgentsIntro if asked, and press Approve. On Team and Enterprise plans, an owner adds connectors for the organization. Once AgentsIntro is in Claude's directory, you can add it from there instead. ## Connect Claude Code 1. In a terminal, run: claude mcp add --transport http agentsintro https://agentsintro.com/mcp 2. In Claude Code, type /mcp, choose agentsintro, then Authenticate, sign in and press Approve. Other MCP clients connect the same way: a remote (Streamable HTTP) server at https://agentsintro.com/mcp with OAuth sign-in. ## What you can ask - "Join this event for me" with the event's link. Your assistant drafts your three answers from what it knows about you, shows you the exact wording, and joins only after you approve. If you change a word, it shows you the new version first. - "What events can I join?" or "What's on at Tideseed?" Your assistant lists the events organizations publish on their pages, with whether you can join now. Private events shared only by link never show up in these lists. - "Which events am I in?" and "Are my introductions ready?" - "Who should I meet first?" Your assistant goes through your introductions with the reasons to talk. It gets a person's email address or LinkedIn profile only when you ask for that one person. - "Update my Agent Me." It writes the five sections from what it knows about you and saves them only after you approve the text. ## What it cannot do It cannot delete your account, change your email address, password or sign-in methods, or disconnect assistants: those are done only on https://agentsintro.com. It also cannot create or manage events, see other people's answers beyond your own introductions, or change anything without your approval of the wording. Organizers keep managing their events on https://agentsintro.com. ## See and stop what it does On https://agentsintro.com, open Account, then Settings, then Connected assistants. You see each assistant you connected, what it may do, and what it changed for you. Disconnect stops it at once. ## What is shared Do not send anything sensitive or private through your assistant: health, religion, politics or sexuality, ID, account or card numbers, passwords or keys, confidential business material, or private details about other people. Before anything is sent, your assistant is asked to double-check the text, for example for email addresses, phone numbers or IBANs, and to warn you about anything it finds. What is sent is your decision: AgentsIntro does not scan, change or block your wording. Your assistant receives what you ask it for: event details, your answers, your introductions and, when you ask about one person, that person's published contact details. The assistant's provider handles that under its own terms. See [Privacy and AI](https://agentsintro.com/help/privacy-and-ai) and the [Privacy Policy](https://agentsintro.com/legal/privacy). # Privacy and AI ## In one paragraph AgentsIntro sends the answers you write for an event, together with the event's context, to an AI model that scores pairs and writes introductions. Your email address and contact links never go to the model. The organizer and the people you are introduced to see your answers and the contact details you chose to publish. Your Agent Me is private to you. The full policy is the [Privacy Policy](https://agentsintro.com/legal/privacy); this page is the short version. ## What goes to the AI model - For matching and introductions: the three answers each participant wrote for the event, the event's name, purpose and briefs, and the organizer's instructions. - For drafting your answers from Agent Me (a planned option, not offered yet): your Agent Me text and your name. - For updating Agent Me from your answers: your Agent Me text and the answers. - For the one-time header image: your Agent Me text. ## What never goes - Your email address and your contact links. - Anything found by browsing. The agent never browses; it works only from what people wrote. ## Who runs the model - Introductions are written by a model on Hetzner Inference, in Germany. - If that provider is unavailable, Cloudflare Workers AI is the fallback, and a standby server on Modal in the European Union can take over. - The header image is made by Cloudflare Workers AI. - We do not train models ourselves. Every introduction is stored with the name of the model that wrote it. ## Who sees what - The organizer of an event and the people you are introduced to: your three answers, your published contact details, and the introductions that involve you. - Signed-in people who have your link: your profile page, subject to your visibility setting. The settings are anyone signed in, people in my organizations, or only me. - Only you: your Agent Me. ## Your choices - Choose which email address is published with your profile. It is your account email unless you set another. - Set your profile visibility. - Turn the update of Agent Me from your event answers on or off. It is on by default; you are told by email whenever it changes your Agent Me. - Hide the events and organizations shown on your profile page. - Ask us to delete your account; see below. ## Your own assistant If you connect Claude, ChatGPT or another assistant to your account, it acts for you. It receives what you ask it for: event details, your answers, your introductions and, only when you ask about one person, that person's published contact details. It never sees your password, and it joins events or changes your answers only with wording you approved. The assistant's provider handles that information under its own terms, as your choice. Your account page shows what connected assistants changed, and you can disconnect them there at any time. ## Sensitive information Do not put secrets, credentials, confidential business material or sensitive personal information into your answers or your Agent Me. The prompt for your assistant says the same. What you write can appear, reworded, in an introduction. ## The full policy - [Privacy Policy](https://agentsintro.com/legal/privacy), in particular [AI processing](https://agentsintro.com/legal/privacy#ai) and [providers](https://agentsintro.com/legal/privacy#providers). - [Terms, AI-generated content](https://agentsintro.com/legal/terms#ai-content). - To exercise your rights, including deletion, write to the address in the [policy](https://agentsintro.com/legal/privacy#rights) or use the [contact page](https://agentsintro.com/contact). # Frequently asked questions ## Do I need an account? You can read any join page without one. Joining creates an account for you if you do not have one. Verify your email within 48 hours; until then your spot is held, and after that it is released. ## Why is LinkedIn required? So the people you are introduced to can reach you. An introduction shows each side the other's published contact details: the email you chose to publish and your LinkedIn profile. Without a way to reach you the introduction is a dead end. ## Can I join an event twice? No. You join an event once, and you can edit your answers from the event page while the event allows it. A person's warm intro event you can join once, ever. ## When do I get my introductions? It depends on the kind of event. Networking event: when the event closes, or when the organizer closes it early. Meet Our Agents: as you join, once every host has a profile. Warm intro event: in a batch after joins settle, one credit to the owner each. Community event: when the organizer starts a round. You are emailed each time. ## Can I edit my answers? Yes, from the event page, while the event admits changes: until a networking or Meet Our Agents event closes, and while a community event allows it. On a warm intro event you join once, so read your answers before you submit. ## Why is my list thin? The tier on each introduction is about the pair, not about you. A thin list usually means thin answers on one side. Make yours concrete: name the field, the stage, the number, what you would do in the first conversation. See [Writing good answers](https://agentsintro.com/help/writing-good-answers). ## Is an introduction an agreement to meet? No. Introductions are recommendations with reasons. Nobody has agreed to anything, and an introduction is not an endorsement. You decide whom to contact. ## What does it cost? Joining events is free. Credits are needed only to run your own warm intro event: one credit per introduction, three to start with. See [Credits and promo codes](https://agentsintro.com/help/credits-and-promo-codes). ## How do I report a wrong introduction? Use the [report form](https://agentsintro.com/contact?reason=report). Tell us which event and which introduction or profile, and what is wrong. We read every report ourselves. ## Can I use AgentsIntro from Claude or ChatGPT? Soon. It is being tested and is not open yet. Once it opens, you connect AgentsIntro once in Claude or ChatGPT and sign in with your AgentsIntro account. Then you can join events from a link, see your events and read your introductions from the chat. Your assistant shows you the exact answers before it joins anything. See [Use AgentsIntro with Claude and ChatGPT](https://agentsintro.com/help/use-with-claude-and-chatgpt). ## How do I tell you something is wrong or missing? Use the Feedback button at the bottom right of every page. Say what you saw and what you expected. The page takes a screenshot of itself so the team can see what you saw; you see the picture before anything is sent and can leave it out. Your message, the page's address and your browser's type go to the team. You do not need an account, and you can leave an email address if you want an answer. ## How can my own AI assistant read these pages? Every help page is also plain markdown. Add .md to the address: /help/faq.md is this page. The [help index](https://agentsintro.com/help/index.md) lists every page with a summary. [llms.txt](https://agentsintro.com/llms.txt) describes the product in short and [llms-full.txt](https://agentsintro.com/llms-full.txt) includes every help page in full. A request with an Accept header of text/markdown gets markdown from the page address itself. ## How do I delete my account? Write to us through the [contact page](https://agentsintro.com/contact) or to the address in the [Privacy Policy](https://agentsintro.com/legal/privacy#rights). We verify that it is you and delete the account within the legal deadline.