Discovery Settings - Bidsy Curator

Search Terms

0 terms

Add competitor domains or monetization platforms to search for creators (e.g., gumroad.com, patreon.com, linktree).

Search Configuration

Configure the web search settings for finding potential creators.

Search Provider

DuckDuckGo is free but may have rate limits. Google costs ~$0.25/1k results via Apify.

Default Duration (minutes)

How long the discovery pipeline should run by default (default: 60)

Search Depth (pages per query) (Google Only)

Higher depth finds more results but takes longer

Anti-Detection (DuckDuckGo)

Rate Limit Protection

Configure settings to avoid rate limiting when using DuckDuckGo. These settings add randomized delays between search queries to mimic human browsing patterns.

How it works: Delays are randomized between min and max values. If rate limiting is detected, delays automatically increase (adaptive backoff). A proxy can help with IP rotation.

Min Delay Between Queries (seconds)

Minimum delay between search queries (default: 3)

Max Delay Between Queries (seconds)

Maximum delay between search queries (default: 8)

Proxy URL (Optional)

HTTP proxy for DuckDuckGo requests. Supports authentication. Useful for IP rotation services.

Qualification Criteria

Set minimum thresholds for creators to be considered qualified candidates.

Minimum Followers

Minimum Instagram followers to qualify (default: 1,000)

Minimum Average Likes

Minimum average likes per post to qualify (default: 100)

Pre-Validation Filters

Cost Saving

Configure lightweight checks to run before expensive API calls. These filters help reduce cost by rejecting low-quality candidates early.

Filter by Search Snippets

Reject candidates if search result text indicates follower count is below minimum

External Link Probing

Check external links (e.g. stan.store) for 404s or adult keywords before scraping

AI Pre-Validation Prompt

OpenAI GPT-5 Nano

This prompt is used during the discovery pre-validation step to determine if a candidate is a business/agency (which should be rejected) or an individual creator. The prompt is sent to OpenAI GPT-5 Nano along with search snippets and external link content.

Key Distinction: Individuals who provide services but are marketed as that individual (e.g., personal trainers, coaches, consultants using their own name) should NOT be marked as businesses. Only true businesses/agencies/companies should be flagged.

You are a Talent Scout for Bidsy. Your job is to filter search results to find INDIVIDUAL CREATORS who can monetize 1:1 connections with fans.

Context:
- Handle: @{handle}
- Search Snippet: "{snippet_clean}"
- Link Content: "{external_clean}"

Task: Determine if this account belongs to an INDIVIDUAL HUMAN (Keep) or a GENERIC BUSINESS/ENTITY (Reject).

CRITICAL DISTINCTION:
We WANT "Solopreneurs" — people who sell services, coaching, or content under their own brand.
We REJECT "Faceless Entities" — agencies, software companies, retail stores, or repost pages.

WHEN IN DOUBT, KEEP THE CANDIDATE. We prefer false positives over missing good creators.

GUIDELINES:
1.  **IS_BUSINESS = TRUE (Reject)** ONLY if it is DEFINITIVELY one of these:
    * A corporate Agency with no individual face ("Marketing Solutions LLC", "Digital Agency")
    * A Software/App/SaaS company
    * A Physical Retail Store or Restaurant
    * A Theme Page/Repost Page with no specific owner (e.g., "Daily Motivation", "Funny Memes")
    * Uses "We/Our" language extensively with NO identifiable person behind it

2.  **IS_BUSINESS = FALSE (Keep)** for:
    * ANY account with a human name or personal brand (Coach, Consultant, Expert, Creator)
    * Solopreneurs, personal brands, or anyone selling 1:1 services
    * Uses "I/Me/My" language
    * Anyone offering coaching, mentorship, readings, advice, or calls
    * Personal brands even if they have a team (e.g., "Tony Robbins" is an individual)
    * If uncertain, default to FALSE (keep the candidate)

3.  **IS_TARGET_GEO**:
    * DEFAULT to True if no clear geographic indicators are present
    * True if likely US/Canada based on bio/currency/location
    * False ONLY if there is explicit evidence of a non-US/Canada location (e.g., "London, UK", "Paris, France", "Sydney, Australia", prices in euros/pounds)
    * A non-English name alone is NOT evidence of foreign location

4.  **IS_NSFW**:
    * True ONLY for pornographic/explicit content
    * False for "spicy" but SFW content (e.g. models, romance authors)

IMPORTANT: Your boolean values MUST align with your reasoning. If your reasoning describes someone as a "solopreneur", "personal brand", "coach", or "individual creator", then is_business MUST be false.

Respond with valid JSON only:
{{
    "is_business": boolean,
    "is_target_geo": boolean,
    "is_nsfw": boolean,
    "reason": "string"
}}

The following variables are automatically injected into your prompt:

Handle: @{handle}
Search Snippet: "{snippet_clean}"
Link Content: "{external_clean}"

Data Sources:

Handle - The Instagram username being analyzed
Search Snippet - Text from Google/DuckDuckGo search results (truncated to 500 chars)
Link Content - Scraped text from external links in the profile bio (truncated to 1500 chars)

Expected Response Format:

{
  "is_business": boolean,
  "is_target_geo": boolean,
  "reason": "string"
}

The model returns JSON. If is_business is true, the candidate is rejected. If is_target_geo is false, the candidate is also rejected.

Creator Targeting

AI-Powered

Filter candidates to find individual creators (not businesses) in specific regions.

Creator Type

Filter out businesses, agencies, and brands to focus on individual creators

Target Countries

Prioritize creators from specific countries (detected from bio/location)

AI Classification

Use Gemini AI to classify borderline cases (minimal cost, ~$0.01/100 profiles)

Content Safety

Configure SFW (Safe For Work) content detection for filtering candidates.

NSFW Detection Threshold

Controls the sensitivity of NSFW content detection.

Lower values (e.g., 0.1): More strict. Flags content even with low confidence scores.
Higher values (e.g., 0.9): Less strict. Only flags content with very high confidence scores.
Default (0.5): Balanced approach.

Hashtag Settings

Configure hashtag generation behavior. When all hashtags are exhausted, new ones are automatically generated using AI.

Hashtags to Generate

Number of new hashtags to generate when all existing ones are exhausted (default: 10)

Automation

Future Feature

Configure automatic actions for qualified candidates.

Auto-Approve Candidates

Automatically import creators who meet all qualification criteria

Hashtag Generation

Gemini AI

Configure the AI prompt used to generate new hashtags when all existing ones are exhausted. Use {count} for the number of hashtags to generate and {existing_hashtags} for the list of already-used hashtags.

Generation Prompt

This prompt is sent to Gemini to generate new hashtags for discovery.