Using Studio Recommendations

Studio Recommendations suggest concrete changes to an existing flow. Flow suggestions cover copy, structure, and A/B variants. Targeting suggestions are a separate list: an audience update on a web flow, or a LinkedIn segment change on a message flow. You review each list, apply what you want, and publish when you are ready.

This is different from Compose, which builds a new flow from a brief. Recommendations improve the last saved flowchart — unsaved canvas edits are not included until you save.

Open the panel

1. Open the flow in Studio (web or LinkedIn message flow).
2. In the top bar, click the sparkles AI improvement button.
3. The badge counts open flow suggestions. Targeting suggestions are in the panel even when that badge does not include them.

The sparkles icon animates while a job is running. The panel shows live status until results arrive. The panel title is AI improvement. The button inside it is Recommend improvements with AI.

On Flow, flow suggestions are open. Targeting suggestions start collapsed behind Targeting suggestions · N, unless they were created while flow suggestions were already open — those stay open at the top. On Targeting, targeting suggestions are open, and flow suggestions sit behind Flow · N.

Generate recommendations

Click Recommend improvements with AI. You can run it again from that panel.

If flow suggestions are already open, those stay as they are. The panel says so. Targeting suggestions are a different list, so that run still creates them. They open at the top of the panel, above the flow cards. Review, apply, or reject the open flow suggestions before asking for a new flow set.

Recent applies and rejection comments are used the next time you ask for improvements. A rejection can include a short comment.

You can also start a job from Console Dashboard → Improve Agents when a flow first reaches 100+ interactions. After that job finishes, Improve Agents shows Open Studio instead of Generate. See Improving agents from the Console dashboard.

Generation always uses the last saved version of the flow, not unsaved canvas edits. Save first if you want the job to see your latest copy.

What gets recommended

Flow suggestions

• Web flows — copy, structure, locales, forms, thank-you / survey patterns, and A/B-style variants where they help conversion.
• LinkedIn message flows — connection notes, DMs, follow-ups, InMail, and suggested-reply copy, plus A/B (bandit arm) variants. Web-only ideas (forms, locale maps, thank-you pages) are skipped because the conversation happens in LinkedIn.

Each flow card targets a specific message. Hover or click it to focus that node on the canvas before you apply.

Targeting suggestions

• Web — Update audience replaces the campaign-brief audience. Apply writes that text. It does not publish the flow.
• LinkedIn — Add segment, Update segment, or Retire segment. Add segment creates the segment and a prospect list. It does not start an import. Retire segment pauses the segment. It does not delete it. Apply updates the portfolio, not the message canvas.

Actions

• Apply on a flow card updates the canvas. It leaves the pending list and appears under History as Applied.
• Apply as variant adds the suggestion as a variant instead of replacing the current copy. Flow cards only.
• Apply on a targeting card updates the audience or the segment portfolio, as above.
• Reject drops the item from the pending list. A comment is optional. It appears under History as Rejected. You cannot re-apply from History.

History

History at the bottom is an audit trail of Applied and Rejected items. It is closed until you expand a section. Sections are grouped by date. A day with more than 10 decisions is split, for example 5 Nov 2026 · 1–10 and 5 Nov 2026 · 11–20. Decisions without a date sit under Earlier. There is no re-apply from History.

Recommended review order

1. Open AI improvement and wait if a job is still running.
2. If flow suggestions are already open, review those, or leave them and read the targeting suggestions at the top.
3. On Targeting, expand Flow · N when you want the copy suggestions.
4. Apply high-confidence flow changes one by one. Use Apply as variant for promising but uncertain copy.
5. Apply or reject targeting suggestions on their own cards.
6. Reject anything off-brief. A short comment steers the next run.
7. Walk the canvas, then publish when you are satisfied.

Troubleshooting

The panel says flow suggestions stay as they are
Open flow suggestions were not replaced. Targeting suggestions from that run are at the top of the panel. Apply or reject the flow cards before asking for a new flow set.

No targeting suggestions
On Flow, look for Targeting suggestions · N. Expand it. On Targeting, the cards are the open list. A run with nothing useful to change can still come back empty.

“No improvements yet”
Nothing is pending. Click Recommend improvements with AI, or start a job from Dashboard Improve Agents after the flow has 100+ interactions. Keep the panel open while the job runs.

Wrong node highlighted
Click the flow card again. Confirm the message before applying. Targeting cards do not move a canvas node.

Apply fails or looks partial
Apply the remaining flow items one by one. Reject a malformed card, then generate again if needed.

Job ignored unsaved edits
Save the flow, then run Recommend improvements with AI again.

Related

• Improving agents from the Console dashboard
• Generating a flow with AI in SMOC Studio
• Web flow vs message flow