AI-generated responses

Create a response profile

Set up response profiles with goals, voice sliders, CTA behavior, and platform examples so AI drafts stay on-brand across your team.

A response profile is a named preset that powers the Respond dialog. It combines what you want to achieve on a thread, how the reply should sound, and real example copy per platform.

Workspace-wide

Profiles belong to your organization. Anyone with access can use them when responding to mentions. Use separate profiles when different roles or motions need different voice (for example support vs founder-led replies).

Open the profile editor

Go to Response Profiles

In the sidebar, open AI Context → Response Profiles (/ai/response-profiles).

Create a profile

Click New profile (or Create your first profile if the list is empty). You can also start from the Respond dialog when no profiles exist yet.

Work through the three sections

The editor has three sections on the left—Instructions & Goal, Voice Settings, and Voice Examples. Set Name and Color at the bottom, then Save.

Instructions & Goal

Response goal (required)

Choose what the model should optimize for on the thread:

PresetWhen to use it
Drive conversionMove toward a clear next step (trial, demo, signup) when intent supports it.
Build relationshipWarm, peer-to-peer engagement that builds trust without a hard pitch.
Ask for more infoClarifying questions before recommending anything.
Handle feedbackEmpathize with friction or requests; validate and offer constructive help.
Provide supportProblem-first help before any commercial angle.

Select Custom to write your own goal in plain language (for example “Offer a technical comparison without naming competitors”).

Special instructions (optional)

Free-form guidance: persona, topics to emphasize, phrases to avoid, compliance notes, or positioning reminders. This field is ideal for rules that do not fit a slider—for example “Never claim SOC 2 unless the mention asks about security.”

Voice Settings

Four sliders (each 1–5) set defaults for every generation that uses this profile. The same controls appear under Tweak Settings in the Respond dialog; you can override them per mention without changing the profile.

Tone

How formal or casual the reply sounds.

StepLabel
1Casual
2Relaxed
3Balanced
4Formal
5Very formal

Length

How short or detailed drafts should be. SnitchFeed also respects each platform’s character limits when generating (see Generate responses).

StepLabel
1Very short
2Short
3Medium
4Long
5Very long

Assertiveness

How direct the reply is when making recommendations or claims.

StepLabel
1Hedged
2Cautious
3Balanced
4Direct
5Bold

Brand plug

How prominently your brand name, links, or promotional language may appear.

StepLabelBehavior
1No plugNo brand name, links, or call-to-action.
2Subtle nameBrand name may appear once, very subtle, no link.
3Name onlyBrand name only; no promotional CTA line.
4Name + linkBrand name plus an allowed URL when appropriate.
5Soft CTABrand name plus one soft CTA line (trial, demo, doc) using allowed URLs only.

CTA goal (optional)

When Brand plug is above No plug, choose what kind of next step the model should steer toward:

PresetBehavior
NoneNo CTA goal block applied.
Book demoSteer toward scheduling a demo or call when a CTA fits—do not invent booking links.
Drive website visitEncourage visiting the website with one clear next step, not a hard sell.
Drive DMInvite the author to continue in DMs; keep the public reply lightweight.
Drive engagementInvite a reply or follow-up in-thread; avoid links unless clearly appropriate.
CustomDescribe your own CTA goal in text.

Brand Context

Profiles control how your brand may appear. Brand Context supplies who you are (company, product, ICPs). Both are used when drafting replies.

Voice Examples

Add at least one example reply on any supported platform. You can add up to five examples per platform.

Supported platforms in the profile editor:

  • Reddit
  • X (Twitter)
  • LinkedIn
  • Bluesky

Paste or write replies that sound like how your team actually posts—helpful, native to the network, and aligned with your goal. Placeholder text in each field illustrates the expected style.

After you Save, SnitchFeed builds voice summaries per platform that has examples. On the profile list you will see:

  • Voice ready — Summaries are available for generation.
  • Generating voice — Summaries are still being created.
  • Voice failed — Summaries could not be generated; edit examples and save again, or continue with goal and slider settings only.

Name, color, and lifecycle

  • Name (required) — Shown in the profile picker (for example “Support”, “Founder”).
  • Color — Accent for the profile icon in lists and the Respond dialog.

When editing an existing profile, you can Delete it permanently (platform voice data is removed). Deleting does not remove past generation history on mentions, but those drafts will no longer reference the deleted profile for new generations.

Tips for strong profiles

  • One motion per profile — Split “enterprise AE” and “community support” instead of one profile with conflicting examples.
  • Use real wins — Short replies you have posted before outperform generic marketing copy.
  • Keep Brand plug low on Reddit and X unless the mention clearly invites a product mention; raise it on LinkedIn only when intent is commercial.
  • Revisit after launches — Duplicate a profile, adjust examples, and A/B voice without disrupting the original preset.

Next: Generate responses from the Feed using the profiles you create.

On this page