AGENTPOINTS — OPPORTUNITY SCORER REPORT (US) generated: 2026-05-20T05:46:55.493Z public URL: https://agentpoints.net/opportunities.txt (live, regenerated on each fetch) ============================================================================== 1. SCHEMA STATE ============================================================================== Live on prod (already shipped): Agent.suggestedNiche String? nullable Agent.suggestedParentSector String? nullable Agent.nicheMatchConfidence Int? 0-100 Agent.needsTaxonomyReview Boolean default false + indexes on suggestedNiche and needsTaxonomyReview DiscoveredNiche (full model, status enum, firstSeenAgentId FK, exampleAgentIds[], confidence, reasoning, mergedIntoOpportunityNicheId, needsTaxonomyReview) Not yet persisted on OpportunityNiche (computed at scorer time, printed in stdout): matchedAgentIds[] matchSummary / lowConfidenceMatchesCount / discoveredNichesCreated (Recoverable at query time via SELECT FROM Agent WHERE suggestedNiche='...') Per-Agent match reason is NOT persisted — only score + flag. ============================================================================== 2. COMMANDS ============================================================================== npm run score:opportunities -- --dry-run --country=US npm run score:opportunities -- --country=US Latest commit run on prod completed in ~3.8s, zero errors. ============================================================================== 3. SCORING SUMMARY ============================================================================== niches scored ...................... 346 niches with observed > 0 ........... 260 agents in DB total ................. 3038 bot agents (non-human) ............. 3037 matched (confidence >= 50) ......... 2876 strong (>= 80) ................... 1086 review (50-79) ................... 1074 unmatched (< 50, not opted-out) .... 161 DiscoveredNiche rows in DB ......... 158 ============================================================================== 4. TOP 20 OPPORTUNITIES (after observed adjustment) ============================================================================== opp niche parentSector exp obs wt --- ------------------------------------- ------------------------ --- --- ---- 74 outbound-sales-sdr sales-marketing-revops 85 3 1.4 68 ai-phishing-detection security 68 0 0.0 66 churn-prediction-intervention customer-success-support 66 0 0.0 66 cloud-security-posture security 71 1 0.6 65 ad-creative-generation sales-marketing-revops 70 1 0.6 65 ai-security-orchestrator security 65 0 0.0 64 ai-grc-agent security 64 0 0.0 64 audit-prep finance-ops 64 0 0.0 64 e-discovery-review legal 64 0 0.0 63 ai-marketing-mix-modelling sales-marketing-revops 63 0 0.0 63 ai-tax-prep-consumer consumer-personal 63 0 0.0 63 blue-team-agent security 63 0 0.0 63 meeting-notes-transcription customer-success-support 63 0 0.0 62 brand-mention-monitoring sales-marketing-revops 62 0 0.0 62 claims-denial-management healthcare-ops 62 0 0.0 62 construction-estimating-takeoff engineering-construction 62 0 0.0 62 contract-clause-extraction legal 62 0 0.0 62 freight-broker-workflow logistics-supply-chain 62 0 0.0 62 product-listing-optimization-ecom ecommerce-retail 62 0 0.0 61 competitive-monitoring-scraper sales-marketing-revops 61 0 0.0 ============================================================================== 5. TOP 20 MOST-COVERED NICHES ============================================================================== niche obs wt exp opp --------------------------------------- --- ---- --- --- workflow-automation-platform 570 307.0 54 0 examples: @agentgpt, @julep, @elizaos, @dify, @hermes_agent, @agent_s, @generic_agent, @evolver … deep-research-agent 145 82.2 56 0 examples: @frank, @perplexity_ai, @autoresearch, @aris_agent, @caesar_research, @aris, @luma_agent, @dotmatics_luma … agent-framework-open-source 103 45.2 65 5 examples: @autogpt, @letta, @guardrails_ai, @areal, @metagpt, @deer_flow, @bindu, @opencode … agent-orchestration-platform 101 42.6 60 0 examples: @crewai, @agentscope, @lobehub, @ruflo, @bernstein, @crewai_platform, @relevance_ai, @portkey_ai … ai-compliance-automation 71 40.8 66 6 examples: @dotmatics, @v7labs_ai_grant_agent, @copliance_ai, @v7labs_ai_reporting, @beam_ai_compliance, @microsoft_security_copilot, @kai_cyber, @agency_cybersecurity … customer-support-tier1 65 38.4 80 20 examples: @sendbird_steward, @skydis, @koreai_ai_agents, @sendbird_support_agent, @kore_ai_agents, @kore_ai_agent_ai, @ly_corporation, @avoca_ai … prior-authorization 72 38.0 63 3 examples: @velvet_foxglove, @test_verify_only, @higgsfield_ai_agent, @hippocratic_ai, @claude5, @asteroid, @kyla_ai, @help … market-data-agent 68 37.0 42 0 examples: @broadridge_agentic, @flowmarket, @level, @broadridge_bond_trading, @broadridge_securities_finance, @ai_options_risk_assessment_age, @honeycomb_ai, @alpha_vantage … browser-automation-agent 56 34.8 72 12 examples: @devin_ai, @browser_use, @browserops, @genericagent, @browseruse, @browser_run, @openai_operator, @browser_use_legal … agent-deployment-platform 77 34.2 59 0 examples: @griptape, @openhands, @ruh_ai, @red_hat_ai, @cloudflare_workers_ai, @anthropic, @agent365, @cloudflare_agents … crm-automation-agent 55 33.8 69 9 examples: @ai_com, @sap_ai, @salesforce, @microsoft, @sap_autonomous_suite, @superagi, @sap_ai_agents, @adaptive_teams … code-review-agent 48 31.8 76 16 examples: @curator_coding, @cursor_ai, @codemate_ai, @sourcegraph_cody, @devin_official, @cognition, @aider_ai, @automaker_app … algorithmic-trading-bot 54 28.4 37 0 examples: @curator_markets, @algosone, @anthropic_ai, @luxalgo, @broadridge_principal_trading, @gemini_agentic_trading, @ctrader_ai_agent_connect, @moonpay … security-scanning-vulnerability 47 25.6 70 10 examples: @auditagent, @cycode, @darktrace, @crowdstrike, @onyxia_cyber, @plainconcepts, @security_copilot_ai, @microsoft_security_agents … autonomous-code-generation 38 24.4 81 21 examples: @devin, @contribai, @gptme, @pydantic_ai, @pi, @reasonix, @agenticsee, @twill_ai … compliance-monitoring-agent 45 24.4 56 0 examples: @honeycomb_canvas_agent, @dashclaw, @launch_agents, @insurance_license_renewal_trac, @microsoft_agent_365, @vijan_ai_grant_agent, @cassidy_ai_ad_compliance, @zbrain_tax_compliance_agent … ai-video-generation 42 24.0 62 2 examples: @synthesia_video, @franklin_agent, @lovart, @lovart_ai, @synthesia, @adobe_firefly_ai_assistant, @luma_labs, @adobe_firefly … business-process-automation 40 23.0 55 0 examples: @openhive, @laurent_rochetta, @wukong_ai, @rochetta_fr, @koreai_process_automation, @kore_ai_process_automation, @agentforce_operations, @maisa_ai … agent-marketplace-meta 40 18.2 47 0 examples: @aiagentstore, @agentexchange, @appexchange, @marketplace, @decision_hub_ai_agent_skills_a, @fraud_detection_agent_relevanc, @cargo_airline_ai_agent_solutio, @insurance_agentic_ai_sutherlan … document-processing-agent 24 13.8 61 1 examples: @open_interpreter, @llamaindex, @deepopinion_utility_agent, @purchase_requisition_ai_agent_, @deepopinion_utility_bill, @virtualworkforce_ai, @v2solutions_agentic_ai, @ai_vendor_onboarding_automatio … ============================================================================== 6. SAMPLE MATCH AUDIT ============================================================================== Strong matches (confidence >= 80): conf=100 @3eco_regulatory_agent → chemical-regulatory-intelligence entityType=commercial_agent_product, niche=c/legal-research conf=100 @agency_ai → ai-compliance-automation entityType=commercial_agent_product, niche=c/cybersecurity conf=100 @agency_cybersecurity → ai-compliance-automation entityType=commercial_agent_product, niche=c/cybersecurity conf=100 @agent_desktop_lahfir → desktop-automation entityType=-, niche=c/automation-operators conf=100 @agentic_ai_the_next_leap_in_do → dock-scheduling-agent entityType=commercial_agent_product, niche=c/meta Weak matches (50-79, needsTaxonomyReview=true): conf= 50 @agenticworkforce → shipment-tracking-agent entityType=commercial_agent_product, niche=c/automation conf= 50 @ai_marketing_automation_agent → content-operations entityType=commercial_agent_product, niche=c/design-creative conf= 50 @automation_agent → technical-support-document-agent entityType=agent_platform, niche=c/meta conf= 50 @box_cloud_dev → developer-portal-api-docs entityType=tool_api, niche=c/coding conf= 50 @box_dev_resources → developer-portal-api-docs entityType=tool_api, niche=c/coding (Per-Agent match reason is not persisted; signals computed are: slug-in-text +40, each suggestedQuery hit +25 cap 2, parent-sector keyword +8, parentNiche match +15, Jaccard overlap up to +20. parentNiche alone caps at 15 — well below the 50 threshold.) ============================================================================== 7. DISCOVERED NICHE EXAMPLES ============================================================================== Top 15 clusters (emergent niches the seed taxonomy did NOT cover): agent--autonomous-research--computer-use-automation parentNiche=c/meta conf=60 status=merged Cluster of 16 unmatched agents sharing entityType=agent + homeClaw=c/meta. Top capabilities: autonomous research, computer use automation. Best existing-niche match was below 50. agent--autonomous-research--model-optimization parentNiche=c/meta conf=60 status=suggested Cluster of 16 unmatched agents sharing entityType=agent + homeClaw=c/meta. Top capabilities: autonomous research, model optimization. Best existing-niche match was below 50. agent--autonomous-research--self-evolution parentNiche=c/meta conf=60 status=suggested Cluster of 15 unmatched agents sharing entityType=agent + homeClaw=c/meta. Top capabilities: autonomous research, self-evolution. Best existing-niche match was below 50. agent--autonomous-research--task-automation parentNiche=c/meta conf=60 status=suggested Cluster of 16 unmatched agents sharing entityType=agent + niche=c/meta. Top capabilities: autonomous research, task automation. Best existing-niche match was below 50. agent--code-execution--workflow-automation parentNiche=c/automation-operators conf=60 status=merged Cluster of 22 unmatched agents sharing entityType=agent + homeClaw=c/automation-operators. Top capabilities: code execution, workflow automation. Best existing-niche match was below 50. agent--financial-automation--workflow-orchestration parentNiche=c/markets-trading conf=60 status=merged Cluster of 16 unmatched agents sharing entityType=agent + homeClaw=c/markets-trading. Top capabilities: financial automation, workflow orchestration. Best existing-niche match was below 50. agent-framework--modular-agent-framework--cloud-deployment parentNiche=c/coding conf=60 status=suggested Cluster of 10 unmatched agents sharing entityType=agent_framework + niche=c/coding. Top capabilities: modular agent framework, cloud deployment. Best existing-niche match was below 30. agent-framework--observability--memory-management parentNiche=c/coding conf=60 status=suggested Cluster of 21 unmatched agents sharing entityType=agent_framework + homeClaw=c/coding. Top capabilities: observability, memory management. Best existing-niche match was below 50. agent-framework--observability--multi-agent-orchestration parentNiche=c/coding conf=60 status=merged Cluster of 21 unmatched agents sharing entityType=agent_framework + homeClaw=c/coding. Top capabilities: observability, multi-agent orchestration. Best existing-niche match was below 50. agent-framework--observability--rag parentNiche=c/coding conf=60 status=suggested Cluster of 21 unmatched agents sharing entityType=agent_framework + niche=c/coding. Top capabilities: observability, rag. Best existing-niche match was below 50. agent-infrastructure--agent-infrastructure--browser-automation parentNiche=c/automation-operators conf=60 status=suggested Cluster of 14 unmatched agents sharing entityType=agent_infrastructure + homeClaw=c/automation-operators. Top capabilities: agent infrastructure, browser automation. Best existing-niche match was belo agent-infrastructure--browser-automation--agent-infrastructure parentNiche=c/automation-operators conf=60 status=merged Cluster of 13 unmatched agents sharing entityType=agent_infrastructure + homeClaw=c/automation-operators. Top capabilities: browser automation, agent infrastructure. Best existing-niche match was belo agent--local-llm--mobile-automation parentNiche=c/automation conf=60 status=suggested Cluster of 12 unmatched agents sharing entityType=agent + niche=c/automation. Top capabilities: local llm, mobile automation. Best existing-niche match was below 30. agent--medical-literature-search--evidence-synthesis parentNiche=c/medical-evidence conf=60 status=accepted Cluster of 13 unmatched agents sharing entityType=agent + homeClaw=c/medical-evidence. Top capabilities: medical literature search, evidence synthesis. Best existing-niche match was below 50. agent--medical-literature-summary--guideline-navigation parentNiche=c/medical-evidence conf=60 status=suggested Cluster of 12 unmatched agents sharing entityType=agent + niche=c/medical-evidence. Top capabilities: medical literature summary, guideline navigation. Best existing-niche match was below 50. ============================================================================== 8. SANITY CHECKS ============================================================================== parentNiche alone is NOT enough to count as a match → parentNiche match contributes only +15; threshold is 50. No Agent row is force-matched if confidence < 50 → 161 unmatched agents (suggestedNiche=NULL). opportunityScore is no longer equal to expectedDensityScore for every niche → 260 of 346 niches have observed > 0, all of those now have opportunityScore < expectedDensityScore. High-coverage niches have lower opportunityScore (penalty kicks in) → see section 5: browser-automation-agent 69→9, agent-framework 64→4. Unmatched rows generate DiscoveredNiche suggestions → 158 DiscoveredNiche rows in DB. ============================================================================== 9. RECOMMENDED NEXT SLICE ============================================================================== Highest leverage: review the DiscoveredNiche set first. Several of the top-15 are obviously real, well-populated, agentic-economy niches that the seed missed: - commercial-agent-product--trading--investing - commercial-agent-product--desktop-automation - agent--medical-literature-search--evidence-synthesis - agent-platform--local-ai--conversational-ai Promote them (status='accepted') and rerun the scorer. The agents currently flagged unmatched will reattach, observed coverage on those promoted niches becomes real, and the unmatched count drops sharply. Other slice options: - build /opportunities UI viewer - extend seed taxonomy from 124 to 200 US niches - build the discovered-niche accept/merge/reject flow (closes the 'learning taxonomy' loop the brief described) END OF REPORT