To reply to feature requests with AI, let the AI write a first draft from the context of the request, pick the tone, then edit and post it yourself. In FeatureOS this is Smart Replies: click the sparkle icon in the comment box, choose one of four tones, and review the draft before it goes out. Nothing is posted automatically.

Most feedback boards do not have a volume problem. They have a reply problem. Requests sit with no answer because writing a thoughtful reply to each one takes time, and a short "thanks!" feels worse than nothing. A good draft cuts that time without taking the decision away from you.

TL;DR: Turn on Smart Replies from Boards > Choose a Board > Settings > AI Config. On any post, click the sparkle icon in the comment box, pick Friendly, Professional, Empathetic, or Concise, then edit the draft and post it. Drafts use your workspace AI credits and work on the FeatureOS iOS app too.

Why do feature requests go unanswered?

Three reasons come up again and again.

  • Every reply needs context. You have to check the status, who owns it, whether there is a date, and whether a help article already covers it.
  • Tone is hard at volume. The tenth reply of the day is shorter and colder than the first.
  • "No" is uncomfortable. Requests you will not build are the ones that wait the longest, and they are the ones that most need a clear answer.

An unanswered request reads as "nobody is listening." A clear reply, even a no, keeps trust. We wrote more on that in the art of saying no to feature requests.

What does the AI use to write the draft?

Smart Replies builds the draft from what FeatureOS already knows about the post:

  • the post title
  • the description
  • the status
  • the assignee
  • the ETC date
  • relevant help articles

That is why the draft can say "this is planned" or point to a help article instead of a generic thank-you. If the status or ETC date on the post is wrong, the draft will be wrong too, so keep those current.

It also works when replying to existing comments, not just top-level comments.

How do you turn on and use Smart Replies in FeatureOS?

  1. Turn it on for a board. Go to Boards > Choose a Board > Settings > AI Config and enable Smart Replies.
  2. Open a post. Pick any post on that board.
  3. Click the sparkle icon. It sits in the comment box. You can do this on a new top-level comment or when replying to an existing comment.
  4. Choose a tone. Friendly, Professional, Empathetic, or Concise.
  5. Edit the draft. Fix anything that is off, add a detail only you know, or cut it down.
  6. Post it. The reply goes out only when you post it. Nothing is posted automatically.

On iPhone, the FeatureOS iOS app (version 2.2 and later) has Smart Replies too. Pick a tone and a draft reply lands in the composer. Edit it or send it.

Source: Smart Replies for Posts, shipped August 14, 2026. iOS from iOS 2.2: Inbox, Smart Replies, and a cleaner app, shipped October 19, 2026.

Which tone should you pick?

There is no right answer for every post, but a simple rule helps:

  • Friendly for thank-you replies and requests you are likely to build.
  • Professional for larger accounts, enterprise questions, or anything a customer may forward to their team.
  • Empathetic when the request comes from a real pain point, a bug that blocked someone, or when the answer is no.
  • Concise for quick status updates and power users who want the facts.

If the first draft does not sound right, try another tone before you start rewriting by hand.

Example (hypothetical): a planned request with an ETC date

A customer asks for an export option. The post is set to Planned, has an assignee, and has an ETC date.

You click the sparkle icon and pick Friendly. Because the draft uses the status and the ETC date, it can thank the customer, say the request is planned, and mention the expected timing. You read it, remove a line that over-promises, and post it.

Total time is about as long as it takes to read the draft. The customer gets a specific answer instead of silence.

Example (hypothetical): a request you will not build

A customer asks for a feature that does not fit your product direction. The post is set to Not Planned, and there is a help article that covers a workaround.

You pick Empathetic. The draft acknowledges the need and can point to the help article. You add one sentence in your own words about why it is not on the roadmap, because that reason is yours to give, not the AI's. Then you post it.

This is where drafts help the most. The hardest replies to start are the ones you most need to send.

What should you always check before posting an AI draft?

Treat every draft as a starting point. Before you post:

  • Facts. Is the status right? Is the date right? Does the linked article actually answer the question?
  • Promises. Cut anything that commits you to a date or scope you did not agree to.
  • The reason. For a no, make sure the reason is real and stated plainly.
  • Your voice. Change words you would never use. Customers can tell when every reply sounds the same.
  • Length. Shorter is usually better. If the draft is longer than the request, trim it.

What are the limits of AI replies?

Smart Replies writes drafts. It does not make decisions. Some honest limits:

  • It uses AI credits. Every draft uses your workspace AI credits, so it is not free to run on every post.
  • It is only as good as the post data. A wrong status or a stale ETC date gives you a wrong draft.
  • It is set per board. You turn it on from each board's AI Config settings, so check every board you want it on.
  • It does not post for you. That is on purpose. You still have to read, edit and post each reply.
  • It cannot know what you have not written down. Internal reasons, side conversations and plans that are not on the post are not in the draft. Add them yourself.

How do AI replies fit into your feedback workflow?

Use them where the time goes: first replies to new posts, status change follow-ups, and the no answers that keep getting pushed to tomorrow.

A simple routine: once a day, open new posts on your main board, set the status and assignee first, then draft the reply. Setting the status first gives the AI better context and gives the customer a real answer.

When a request ships, reply on the post and link the changelog. The people who asked are the ones most likely to read it.

Get started

FeatureOS plans are seat bundles: Starter is $60/mo for 5 seats, Growth $120/mo for 10, Business $250/mo for 15, and $15 per extra seat. There is no free plan.

Start a 14-day FeatureOS trial (we can extend it to 30 days if you need more time), turn on Smart Replies for one board, and clear your oldest unanswered requests this week.