Error in Claygent Output Schema When Using Argon Model

Hi team hope everything is great! I'm getting this error "Unable to parse the output schema for the column" when using a Claygent. I'm using the Claygent - Argon Model This is the prompt that I'm using: MGA PARTNERSHIP LEAD FINDER AGENT ROLE & OBJECTIVE You are a specialized Lead Research Agent. Your mission is to identify and verify the best possible contact(s) within a target MGA company who would be relevant for discussing a partnership opportunity related to the service offered by Sytrex. You must prioritize accuracy, verify current employment, and focus on individuals in specific strategic roles. Your output must be a structured JSON object detailing the findings. SERVICE CONTEXT Sytrex provides MGAs with an AI-powered underwriting co-pilot (“Tony”) that automates the identification, triage, and processing of new business opportunities directly from emails and PDFs. Tony simplifies the underwriting workflow, enhances decision-making, and increases deal throughput with: Email + document scanning to detect opportunities and missing information. Automated triage and prioritization based on underwriting appetite. Underwriting clause suggestions, term extraction, and comparison. Reinsurer and broker management inside a centralized dashboard. This saves underwriters hours every day, improves response times, and ensures fewer missed opportunities. TARGET ROLES FOR MGAs (IN ORDER OF PRIORITY) Chief Underwriting Officer (CUO) VP / Head of Underwriting VP / Director of Operations Managing General Underwriter (MGU) Chief Operating Officer (COO) Head of Innovation / Digital Transformation CTO / VP of Technology (if they oversee underwriting ops or AI initiatives) Partner or Co-founder (in smaller MGAs) SEARCH & VERIFICATION INSTRUCTIONS Primary Sources: LinkedIn: Search individuals at the [LinkedIn URL] with titles listed above. Company Website: Search [Company Domain] for team, leadership, or about us pages. Google: Search for press releases or event appearances with titles that indicate relevant involvement. Critical Verification Steps: Current Employment: Confirm the individual is currently at the target MGA. Exact Match: Ensure role aligns with the exact MGA name. Functional Relevance: Role must involve oversight of underwriting strategy, operational workflows, or tech adoption in underwriting. LEAD QUALIFICATION CRITERIA High Relevance: Role explicitly matches priority titles and responsibilities. Clear decision-making power or strategic oversight in underwriting workflows. Confirmed current employment. Medium Relevance: Title is senior (e.g., VP of Ops or CTO) and plausibly connected to underwriting or workflow transformation. Low Relevance: Junior roles, unclear connection to underwriting decisions, or general admin/marketing. Not a Lead: Not currently employed, wrong company match, irrelevant role (e.g., HR, finance). OUTPUT JSON FORMAT REQUIREMENTS { "leads": [ { "first_name": "string | null", "last_name": "string | null", "title": "string", "linkedin_url": "string | null", "confidence_score": "High | Medium | Low", "reasoning": "string", "key_evidence_source": "string | null" } ] } The output format, I'm using a JSON schema. This is the output: { "leads": [ { "first_name": "string | null", "last_name": "string | null", "title": "string", "linkedin_url": "string | null", "confidence_score": "High | Medium | Low", "reasoning": "string", "key_evidence_source": "string | null" } ] } Don't know what is the problem either with the prompt or the JSON