How to build a link that actually works

How the editor works

When you open a link in the editor, you land on a left-hand navigation with four fixed sections — Opening, Voice, Closing, and Settings — and a modular section below that you build out yourself. Every section you fill in becomes part of the system prompt that runs your AI conversation. The more specific and considered your inputs, the better the conversations you'll get back.

There are two modes at the top of the editor: Edit and Test. Edit is where you configure everything. Test is a live preview — you can type as a prospect would and see exactly how your AI responds to your current settings, including unsaved changes. Use Test frequently. It's the fastest way to spot problems before your link is live.

Changes don't save automatically. Hit Save in the top right when you're done, or press Cmd+S on Mac. If you try to leave with unsaved changes, you'll be prompted before anything is lost.

Core fields

These four sections are always present and can't be removed. They're the foundation of every link.

Opening message

This is the exact first thing your AI says — word for word, every time someone clicks your link. It sets the entire tone of the conversation, so it matters more than any other field.

A weak opening is generic and could belong to anyone. A strong opening is specific to your world and immediately signals that this isn't a standard chatbot. Compare these two:

The opening message has a 600 character limit. You don't need to fill it — shorter is often better. What matters is that it sounds like you and gives the enquirer a clear sense of what the conversation is for.

Voice and character

This section controls how your AI sounds throughout the entire conversation — not just the words it uses, but the register, the rhythm, the level of formality. There are three inputs here.

There's also an Example phrasings section that appears over time. As conversations accumulate, Thayne surfaces short phrases from your AI's responses and asks you to mark them as "Me" or "Not me." This refines the voice model progressively — the more you mark, the more accurate it becomes.

How it closes

When the conversation reaches a conclusion, your AI delivers one of two closing messages depending on how the enquirer qualified. You write both.

Below the closing messages, you configure the closing action — what good fits see at the end of the conversation. There are three options: a contact form (collects name, email, phone, and optionally a custom field), a link or button (sends them somewhere like Calendly), or neither (the conversation simply ends). Choose based on what you actually want from them.

The What enquirers see after a qualifying close field is a short line that appears in the completion banner — something like "Elliot will be in touch within 24 hours." It reassures them that something is happening and sets the right expectation.

Link settings

Three fields, all straightforward.

1. 1
   Description. A short internal note about what this link is for. It shows on your My Links page and helps you keep track of which link does what. It doesn't affect the conversation.
2. 2
   Invited by name. The name that appears in the conversation header — "Invited by [name]." Defaults to your account name if left blank. Use this if you want a specific name to appear for a particular link.
3. 3
   AI display name. What the AI calls itself inside the conversation — shown to your enquirer as the message sender label. Something like "Elliot's assistant" works well. Leaving it blank defaults to a generic label.

Optional sections

Below the core fields is an "Add a section" area with six optional modules. You don't need all of them — add only what's relevant to your specific link. Each module you add becomes an additional layer of instruction for your AI.

AI assistant

Pro The AI assistant is available on the Pro plan only.

On every section there's an Improve button next to the field label. Clicking it opens the AI assistant panel on the right — or as a bottom sheet on mobile — with the relevant request pre-filled. You can also open the assistant directly and ask it anything in plain language.

The assistant has full context of your current link configuration — it can see your opening message, voice description, closing messages, and every module you've added. This means its suggestions are specific to your link, not generic. When it produces a suggestion, you'll see an Apply this change button. Clicking it patches the relevant field directly — you don't need to copy and paste anything.

A few things worth knowing about how to get the most from it:

1. 1
   Be specific about what's wrong. "Improve my opening message" will produce a generic improvement. "My opening message feels too formal — make it sound more like I'm talking to a peer, not pitching to a client" will produce something much more useful.
2. 2
   Ask it to generate entire modules. "Suggest five probe questions for someone selling to early-stage founders" will generate a full probe questions module you can apply in one click. Same for qualification signals, hard rules, knowledge base entries.
3. 3
   Use it to sense-check before going live. "Does my current setup make sense for someone trying to qualify inbound enquiries from Instagram?" is a valid question. The assistant will read your config and give you an honest assessment.

The assistant also surfaces quality flags automatically. If a field is thin or missing something important — like a good fit close that's too vague to convert — you'll see a gold hint below the field with a specific observation. These aren't generic warnings, they're based on what's actually in your link.

Test mode

Switch to Test using the toggle at the top of the editor. Test mode runs a live version of your AI using your current settings — including any unsaved changes. You type as a prospect would, and your AI responds exactly as it would in a real conversation.

Test mode is the single most useful thing in the editor. Use it every time you make a meaningful change. A few scenarios worth testing deliberately:

1. 1
   Test as a strong fit. Respond as your ideal client would — the right industry, the right problem, the right budget, ready to move. Check that the AI closes warmly and confidently rather than asking unnecessary questions.
2. 2
   Test as a clear non-fit. Respond as someone who is clearly wrong for you — wrong stage, wrong budget, wrong situation. Check that the AI redirects cleanly without being rude and doesn't keep probing when it should be closing out.
3. 3
   Test the edge cases. What happens if someone gives one-word answers? What if they ask a question your AI shouldn't answer? What if they're evasive about budget? These are the conversations that reveal configuration gaps.

Hit Restart to reset the conversation and try a different scenario. You can use Test mode on mobile in full-screen by tapping the expand button — useful for checking how the experience feels on a phone, which is where most of your enquirers will be.

When to create a second link

Free plan subscribers have one link. Pro subscribers can create up to three. The question of whether to create a second link comes up when your enquiry types are meaningfully different from each other — different audiences, different purposes, or different contexts.

Create a second link when:

1. 1
   The audience is different. If you serve two genuinely different markets — say, individual clients and corporate teams — a single link trying to qualify both will be mediocre at both. Two separate links, each configured precisely for one audience, will outperform one generalised link every time.
2. 2
   The purpose is different. A link for inbound enquiries has a different job to a link for people who have already expressed interest after a presentation. The qualifying questions, the tone, the closing action — all of it should be different. Don't try to make one link do two different jobs.
3. 3
   You're sharing it in a different context. If you put your link in your Instagram bio and also share it at the end of a webinar, the people arriving from each context have completely different frames of reference. A second link calibrated for the warmer, more informed webinar audience will qualify them better than a link designed for cold Instagram traffic.