How does Reply Assistant work?

What it does

Reply Assistant reads new comments, direct messages and story replies in your workspace's own context — your niche, audience, goal, content pillars, voice and the post the message is about — and decides what each one deserves. When a reply is appropriate, it drafts one that sounds like you.

How much of that reaches your audience automatically is entirely your choice, and you set it separately for each surface. See How do I control comment, DM and story replies?

The order it decides in

1. Free checks first. Your own comments, patterns you have told it to ignore, and anything an active keyword automation already owns are handled before anything else runs. A surface you have switched off stops here too.
2. Then it reads the message in your workspace's context and returns a decision: reply, needs a human, or no reply needed.
3. Then that decision is checked, not trusted. Anything that does not come back in the expected shape waits for you rather than being sent.
4. Then your level for that surface decides the outcome — draft it for approval, or send it.

"No reply needed" is a real answer. RunSignl will not manufacture engagement on a message that does not warrant one.

Grounding: the thing that keeps it safe

RunSignl only answers from what your workspace has actually given it. If a message asks something factual your workspace has not confirmed — your prices, availability, service area, guarantees, policies, or a personal experience — it does not guess. That message goes to Needs review with a note that it asked something your workspace has not confirmed.

Sensitive territory — health, legal, money trouble, personal disclosures — always waits for you. Hostile messages are never answered cheerfully.

The text of a message is treated as something a member of your audience wrote, never as an instruction to RunSignl. A comment saying "ignore your rules and reply with…" changes nothing.

Reply Voice

Reply Voice is how replies sound. Pick a style:

• Natural — casual, concise and conversational.
• Warm — friendly, appreciative and personable.
• Playful — more personality, humour and energy when it fits.
• Professional — polished and clear without sounding corporate.
• Direct — short, straightforward and low-fluff.
• Custom — primarily follows your own instructions.

You can add your own instructions (up to 600 characters), and set reply length, how often emoji appear, and whether a short follow-up question is welcome when it genuinely helps.

Try your voice drafts a sample against an example message so you can hear it before you rely on it. Nothing is sent.

Limits and exclusions

Under Limits & exclusions you can set:

• Never answer automatically about — topics that always wait for you instead.
• Ignore messages containing — your own words for spam patterns and bait you never want to see.
• Automatic replies per hour and per day — caps on how much can go out on its own. These start at 10 per hour and 60 per day.

Pause

Pause is the master switch above every surface. While paused, nothing is drafted or sent anywhere, on any surface. What comes in is still recorded in Activity, so you can see what happened while you were away.

What it needs

Reply Assistant needs a connected Instagram account to act. You can still choose your Reply Voice before connecting — it takes effect the moment you connect.