Creators and personal brands face a specific problem: people want to talk to you, not a generic assistant, but you can't be in every DM at once. A Digital Twin is an AI Agent trained on how you think, write, and respond, so your audience gets a genuinely personal conversation even when you're busy, asleep, or creating.
What you'll build
An AI Agent that mirrors your voice, knowledge, and perspective. It engages your audience, answers questions the way you would, and keeps conversations feeling personal at scale. When something needs the real you, it flags it and steps aside.
Step 1: Draft your personality from your profile
In AI Agent → Voice and personality, use Let Inrō draft a personality for you to generate a starting profile from your Instagram account. Then refine it: paste in examples of how you actually write (DMs you've sent, captions, email replies) and adjust anything that doesn't land. Set the message length to Short or Normal so replies feel like real DMs. See Configuring Your AI Agent's Personality and Tone.
Step 2: Build the knowledge base around your expertise
In AI Agent → Knowledge access and behaviour, this is less about FAQs and more about your ideas. Add your core frameworks and methods, your actual answers to the questions you get asked most, links to your key content with a note on what each covers, your offers in full, and your genuine point of view on the topics your audience brings you. The more specific this is, the more distinctly you the agent sounds. See Knowledge Base: Teaching Your AI Agent What to Know.
Step 3: Define your audience and your moves
In AI Agent → Goals and instructions, use Audience type to describe who you're talking to by how they think ("beginner creators overwhelmed by content strategy" beats "creators aged 25 to 35"). In Other instructions, write out the situations your audience raises most and how you'd handle each: "If someone asks about 1:1 coaching, tell them spots are limited and share the waitlist link. If someone doubts their niche, lead with validation before advice." The agent won't guess your approach; you have to show it. See Goals and Instructions: Telling Your AI Agent What to Do.
Step 4: Decide who it talks to
In AI Agent → Activation rules, All pending messages suits a twin that responds to your whole audience. If a team handles some conversations, use Specific intents only so the agent covers defined topics and leaves the rest for people.
Step 5: Keep the personal conversations for you
In AI Agent → Human intervention, configure the agent to detect requests for a personal conversation, a major brand inquiry, or a sensitive situation, and flag them immediately. Route these to a "Needs me personally" folder so nothing slips through. See Human Intervention: Handing Off to a Human.
How it works in practice
A follower DMs after your latest post: "This changed how I think about content. How do you decide what to post?"
The twin, trained on your frameworks and voice, replies the way you would: "Honestly, I start with what I'm actually curious about, not what the algorithm wants. If I'd send it to one specific person, it's worth posting. What made that post land for you?" The conversation continues, the agent shares relevant content, and if the follower asks about your program, it describes it accurately and offers a next step.
Variations
Team-assisted twin. Use Specific intents only so the twin handles general engagement and program questions while your team takes brand deals and partnerships.
Twin behind a funnel. Open conversations with a Comment-to-DM flow, then hand over to the twin so the structured first step gives way to a personal conversation.
Results to expect
A good twin gives every follower a real, in-voice conversation at any hour, which is impossible to do by hand past a certain size. Many creators find that being upfront about it ("an AI version of me handles DMs") builds trust rather than reducing it, especially once the replies are genuinely good.
🐾 Netsuke's Tips
Prioritise message examples over written style rules. Ten real messages teach the agent more about your voice than any description. Add at least five before going live.
Paste in messages where you handled a tricky moment well (a hard question, a critic). Edge cases teach the agent more than the easy exchanges.
Test it yourself first. DM your own account, have a full conversation, and adjust anywhere it doesn't sound like you.
What's next?
With the twin configured, dial in its quality and scope. See AI Agent Best Practices for getting consistently good output, and Human Intervention: Handing Off to a Human to make sure the right conversations still reach you.

