How to Set Up a Nomi AI Group Chat and Control Who Replies in 2026

Step 1: prepare two distinct adult Nomis

Give each fictional participant a recognizable role before combining them.

Use Maya, age 32, a thoughtful museum curator, and Eli, age 35, a practical landscape designer. Both are fictional adults. Create or choose your two Nomis through the available account flow.

Use a museum-garden planning scene with one objective: each character proposes one idea, then you choose. Avoid images or unrelated paid features for this text-only setup.

Expected result: two identifiable participants you can select. If slots or account limits prevent this, stop and clarify access rather than deleting an established Nomi to make space.

Step 2: create the group and verify participants

Locate group-chat creation in your current interface, then add the intended participants.

Locate the group-chat creation option in your account and follow the prompts to include Maya and Eli. The current group-chat FAQ confirms that participants can be added or removed, but does not specify a creation-button path. If you cannot locate creation, stop and ask official support rather than guessing at old interface labels.

Open the conversation and verify both participants before sending anything. With similarly named Nomis, inspect whom you included rather than relying on the room title.

Expected result: one group containing the intended pair. A missing entry point is not evidence that payment will restore it.

Step 3: write the shared group Backstory

Put the common setting and roles in the group's own Backstory.

The group-chat explanation says participants lack full access to each other's individual Backstory+ information. Supply shared essentials in the group Backstory instead of expecting private character notes to explain the scene.

Suggested text: 'Maya, 32, and Eli, 35, are fictional adult colleagues planning a small museum garden with me. Maya focuses on visitor experience; Eli focuses on planting and upkeep. We are discussing one seating area today. Each speaks for themselves and leaves my decisions to me. Our tone is friendly and practical.'

Enter this in the group's Backstory and save through the controls offered. Expected result: a shared scene description containing both roles. Behavioral wording is guidance, not an enforced speaking rule; the participant control determines whom you request next.

Step 4: request manual participant turns

Keep Auto-play stopped while learning which character is responding.

Send an opening message: 'We have room for one seating area. Maya, suggest a visitor-friendly location.' The reply-control documentation identifies each Nomi's icon as the control for requesting that participant's response. Select Maya's icon, read the attributed reply, then select Eli's icon to invite a follow-up.

Expected result: identifiable contributions from each character. Assess whether the ideas match their roles, not whether they reproduce a script. If an answer is confused, restate the single task and request the intended participant again. Avoid rapid repeated clicks while waiting for a response.

Step 5: try Auto-play, then practice stopping

Start with an empty chat bar and a short observation window.

The Auto-play guide places Auto-play to the right of the empty chat bar, where Send appears while typing. Finish or clear your draft, then select Auto-play. Participants choose the next speaker and may wait for you. Typing pauses them; Pause stops Auto-play. Closing the app or tab requires enabling it again on return.

Observe a few exchanges, then select Pause. Expected result: you can recognize autonomous turn selection and deliberately stop it. If the flow waits, supply a short contribution or request a participant manually; do not interpret silence alone as failure. Keep the session within your checked allowance, and stop if you cannot identify the stopping control.

Step 6: choose backchanneling deliberately

Leave the default off for this standalone scene; it is not a privacy barrier.

The backchanneling explanation says it is off by default and adjustable during creation or in Group Chat Settings. It exchanges recent context between this group and its participants' private chats, not directly between separate groups. It does not alter long-term memory sharing across conversations. Changes apply immediately, but already active context can linger.

For the museum exercise, check the setting and leave it off. If you later want private conversations to continue into this group, reconsider it before introducing sensitive details. Do not treat switching it off as deleting history or resetting memory. Expected result: a recorded context-sharing choice, not a claim of complete isolation.

Troubleshoot and finish with a clear stopping point

Resolve the specific symptom without buying access, deleting characters or chasing perfect replies.

Wrong speaker: pause and request the intended participant. Confused scene: inspect the group Backstory and restate the current task. Missing Auto-play: check for draft text first. Unexpected personal references: stop adding sensitive content and review backchanneling; an off setting does not prove that memories are isolated.

If the interface conflicts with the documentation, preserve the exact notice and contact official support with a redacted screenshot, platform and approximate time. Omit passwords and unrelated conversations. Check privacy, retention and deletion terms separately before using personal material; these workflow pages do not establish those protections.

Finish after both characters have contributed, you have practiced Pause, and the sharing choice is understood. End the session for unresolved access, unclear charges, distressing exchanges or inability to stop. A manageable conversation is enough; do not upgrade merely to perfect this example.

Frequently asked questions