Productivity

Audience Reaction Engine (by Zinnawa)

by David Blum — Zinnawa

Overview

Audience Reaction Engine lets Claude run your product launch, price change, campaign message, announcement, crisis statement or personal post past a panel of AI personas before it goes public. You get a GO / CAUTION / NO-GO verdict, every persona's reaction and objection, the main risks and opportunities, concrete rewrite suggestions, and a link to a full visual report. Compare two versions head-to-head, or interview a single persona to understand what would change their mind. Consumer pack (16 personas) and B2B pack (12 enterprise stakeholders). Works in any language. Honest limits: this is a directional simulation, not a prediction. In two pre-registered external backtests on 440 real campaigns, the panel did not beat chance at predicting real-world outcomes. Use it to catch blind spots and sharpen a message, not to forecast sales. Regulated categories (gambling, crypto, health claims, weight loss, trading, cosmetic surgery, tobacco) are refused at intake.

Tools

compare_versions

Claude

A/B test two variants of the same content head-to-head on the panel. Both variants run through the full engine in parallel against the SAME personas; returns the overall winner, how many personas prefer each version, and the per-persona duel. Use it to pick between two headlines, two price framings, two apology drafts… Args: stimulus_type: same values as test_audience_reaction. fields_a: version A content fields (same shape as test_audience_reaction). fields_b: version B content fields. pack: "consumer" (default) or "b2b".

interview_persona

Claude

Ask ONE persona a follow-up question, in character. Run test_audience_reaction first, then dig into a persona's reaction: why they object, what would change their mind, how they'd phrase it. Pass the same stimulus fields (and optionally the persona's initial reaction/objection from the test) so the interview stays consistent. Args: persona_id: the persona id from list_personas (e.g. "skeptical", "cfo"). question: your question, any language. stimulus_type: same values as test_audience_reaction. fields: the same content fields the persona reacted to. initial_reaction: (optional) that persona's reaction text from the test. initial_objection: (optional) that persona's objection from the test.

list_personas

Claude

List the AI personas of an audience pack. Args: pack: "consumer" (16 personas) or "b2b" (12 enterprise stakeholders).

test_audience_reaction

Claude

Simulate how a whole audience reacts to content BEFORE publishing it. Runs the stimulus through a panel of AI personas (each with its own values, web grounding and perception), aggregates sentiment across platforms, and returns an executive verdict: GO / CAUTION / NO-GO, plus each persona's reaction and objection, top risks/opportunities, recommended actions and a link to the full visual report. Args: stimulus_type: one of "product_launch", "price_change", "campaign_message", "crisis_statement", "company_announcement", "personal_post", "personal_photo", "personal_opinion". fields: content fields, depending on the type — product_launch: {"name", "category", "price", "description"}; price_change: {"name", "current_price", "new_price", "reason"}; campaign_message: {"brand_name", "target_audience", "channel", "message"}; crisis_statement: {"company_name", "crisis_type", "statement"}; company_announcement: {"company_name", "announcement_type", "announcement_text"}; personal_post: {"content", "platform"}; personal_photo: {"caption", "platform"}; personal_opinion: {"opinion", "platform"}. Write the text fields in the language you want the panel to answer in. pack: "consumer" (default) or "b2b".

Links

Documentation ↗Privacy Policy ↗

App Stats

Tools

Claude

Platforms