Account Enrichment Review

AI-classified industry taxonomy — review before pushing to HubSpot

\n \n CRM Dashboard\n \n \n View Source\n \n

\n\n \n

\n \n

\n \n \n

\n\n

\n\n \n \n \n \n\n

\n\n \n \n \n

\n\n \n

\n \n Running...\n

\n\n \n

\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n

	Company	Current Industry	Tier 1	Tier 2	Tier 3 (Niche)	ICP	Segment	Conf	Status	Actions
Loading...

\n \n

\n \n \n

\n\n \n \n

\n\n \n \n\n\n","label",[18],"pulls","branches",1,"forks","envVars",{"_162":156,"_164":26},"connections",{"_162":26,"_163":26},"count","unfilledCount","emptyCount",{"_32":245,"_13":212,"_62":206,"_42":-5,"_34":35,"_147":246,"_148":84,"_36":247,"_38":26,"_64":248,"_150":-5,"_65":24,"_152":-5,"_182":249,"_17":250},{"_32":237,"_13":238,"_62":239,"_42":206,"_34":35,"_147":240,"_148":84,"_36":241,"_38":26,"_64":189,"_150":242,"_65":24,"_152":-5,"_182":243,"_17":244},{"_32":229,"_13":230,"_62":231,"_42":206,"_34":35,"_147":232,"_148":84,"_36":233,"_38":26,"_64":189,"_150":234,"_65":24,"_152":-5,"_182":235,"_17":236},{"_32":221,"_13":222,"_62":223,"_42":206,"_34":35,"_147":224,"_148":84,"_36":225,"_38":26,"_64":189,"_150":226,"_65":24,"_152":-5,"_182":227,"_17":228},{"_32":213,"_13":214,"_62":215,"_42":206,"_34":35,"_147":216,"_148":84,"_36":217,"_38":26,"_64":189,"_150":218,"_65":24,"_152":-5,"_182":219,"_17":220},{"_32":203,"_13":204,"_62":205,"_42":206,"_34":35,"_147":207,"_148":84,"_36":208,"_38":26,"_64":189,"_150":209,"_65":24,"_152":-5,"_182":210,"_17":211},{"_32":196,"_13":197,"_62":198,"_42":-5,"_34":35,"_147":111,"_148":84,"_36":199,"_38":156,"_64":21,"_150":200,"_65":24,"_152":-5,"_182":201,"_17":202},{"_32":146,"_13":18,"_62":63,"_42":-5,"_34":35,"_147":39,"_148":84,"_36":193,"_38":26,"_64":21,"_150":151,"_65":24,"_152":-5,"_182":194,"_17":195},{"_32":185,"_13":186,"_62":187,"_42":-5,"_34":35,"_147":179,"_148":84,"_36":188,"_38":26,"_64":189,"_150":190,"_65":24,"_152":-5,"_182":191,"_17":192},{"_32":175,"_13":176,"_62":82,"_42":-5,"_34":35,"_147":177,"_148":84,"_36":178,"_38":179,"_64":180,"_150":181,"_65":24,"_152":-5,"_182":183,"_17":184},"26c16bcc-5a2e-11f1-9429-ee650bb23af1","main.ts",13,["D",1779928969029],4,"http","import { parseVal, serveFile } from \"https://esm.town/v/std/utils/index.ts\";\nimport { fetchContacts, fetchDeals, fetchCompanies, hasToken } from \"./hubspot.ts\";\nimport {\n initDB,\n getClassifications,\n getStats,\n updateReviewStatus,\n bulkUpdateReviewStatus,\n getLatestRun,\n} from \"./enrichment/db.ts\";\nimport { runPipeline } from \"./enrichment/pipeline.ts\";\n\nasync function handleApiRequest(fetcher: () => Promise) {\n try {\n const data = await fetcher();\n if (!data) return Response.json({ error: \"no_token\" }, { status: 401 });\n return Response.json(data);\n } catch (e: any) {\n return Response.json({ error: e.message }, { status: 500 });\n }\n}\n\nexport default async function(req: Request): Promise {\n const url = new URL(req.url);\n\n if (url.pathname === \"/source\") {\n return new Response(null, { status: 302, headers: { Location: parseVal().links.self.val } });\n }\n\n // ── Original CRM dashboard routes ──────────────────────────────\n if (url.pathname === \"/api/contacts\") return handleApiRequest(fetchContacts);\n if (url.pathname === \"/api/deals\") return handleApiRequest(fetchDeals);\n if (url.pathname === \"/api/companies\") return handleApiRequest(fetchCompanies);\n\n // ── Enrichment pipeline routes ─────────────────────────────────\n\n // Initialize DB (idempotent)\n if (url.pathname.startsWith(\"/api/enrichment\")) {\n await initDB();\n }\n\n // Start a pipeline run\n if (url.pathname === \"/api/enrichment/run\" && req.method === \"POST\") {\n const body = await req.json().catch(() => ({}));\n const maxCompanies = body.maxCompanies || 5; // Default to 5 for safety\n const dryRun = body.dryRun || false;\n\n // Fire-and-forget — respond immediately, process in background\n const responsePromise = runPipeline({\n maxCompanies,\n dryRun,\n delayBetweenMs: 2000,\n });\n\n // Don't await — let it run in background\n responsePromise.catch((err) => console.error(\"Pipeline error:\", err));\n\n return Response.json({\n status: \"started\",\n message: `Processing up to ${maxCompanies} companies${dryRun ? \" (DRY RUN)\" : \"\"}`,\n });\n }\n\n // Get pipeline status / latest run\n if (url.pathname === \"/api/enrichment/status\") {\n const run = await getLatestRun();\n const stats = await getStats();\n return Response.json({ run, stats });\n }\n\n // Get classifications (with filters)\n if (url.pathname === \"/api/enrichment/classifications\") {\n const status = url.searchParams.get(\"status\") || undefined;\n const icpFitTier = url.searchParams.get(\"icp\") || undefined;\n const minConfidence = url.searchParams.get(\"minConfidence\")\n ? parseInt(url.searchParams.get(\"minConfidence\")!)\n : undefined;\n const limit = parseInt(url.searchParams.get(\"limit\") || \"50\");\n const offset = parseInt(url.searchParams.get(\"offset\") || \"0\");\n\n const data = await getClassifications({ status, icpFitTier, minConfidence, limit, offset });\n return Response.json(data);\n }\n\n // Review a single classification\n if (url.pathname === \"/api/enrichment/review\" && req.method === \"POST\") {\n const body = await req.json();\n await updateReviewStatus(body.hubspotId, body.status, body.notes);\n return Response.json({ ok: true });\n }\n\n // Bulk review\n if (url.pathname === \"/api/enrichment/bulk-review\" && req.method === \"POST\") {\n const body = await req.json();\n await bulkUpdateReviewStatus(body.hubspotIds, body.status);\n return Response.json({ ok: true, count: body.hubspotIds.length });\n }\n\n // ── Page routing ───────────────────────────────────────────────\n if (url.pathname === \"/enrichment\") {\n return serveFile(\"./enrichment-dashboard.html\", import.meta.url);\n }\n\n return serveFile(\"./dashboard.html\", import.meta.url);\n}\n","children",[],[176],"cc95e79c-59f4-11f1-b759-ee650bb23af1","hubspot.ts","cc95e850-59f4-11f1-b759-ee650bb23af1",["D",1779904336432],"script","const TOKEN = Deno.env.get(\"TOKEN\") || \"\";\n\nconst HEADERS = {\n Authorization: `Bearer ${TOKEN}`,\n \"Content-Type\": \"application/json\",\n};\n\nexport function hasToken() {\n return !!TOKEN;\n}\n\nasync function fetchObjects(objectType: string, properties: string[], limit = 100) {\n if (!TOKEN) return null;\n const props = properties.join(\",\");\n const res = await fetch(\n `https://api.hubapi.com/crm/v3/objects/${objectType}?limit=${limit}&properties=${props}`,\n { headers: HEADERS },\n );\n if (!res.ok) {\n const err = await res.text();\n throw new Error(`HubSpot API error (${res.status}): ${err}`);\n }\n return res.json();\n}\n\nexport function fetchContacts() {\n return fetchObjects(\"contacts\", [\n \"firstname\", \"lastname\", \"email\", \"phone\", \"company\", \"lifecyclestage\", \"createdate\",\n ]);\n}\n\nexport function fetchDeals() {\n return fetchObjects(\"deals\", [\n \"dealname\", \"amount\", \"dealstage\", \"pipeline\", \"closedate\", \"createdate\", \"hubspot_owner_id\",\n ]);\n}\n\nexport function fetchCompanies() {\n return fetchObjects(\"companies\", [\n \"name\", \"domain\", \"industry\", \"numberofemployees\", \"city\", \"state\", \"country\", \"createdate\",\n ]);\n}\n",[],[186],["D",1779929047493],[],[18],"5f524420-5a2e-11f1-b7fd-ee650bb23af1","dashboard.html","dfdbe284-59f4-11f1-ba23-ee650bb23af1",["D",1779929063931],"\n\n\n \n \n HubSpot CRM Dashboard\n \n \n\n\n

\n\n \n

\n \n

HubSpot CRM Dashboard

\n \n \n Enrichment Review\n \n \n View Source\n \n

\n\n \n

\n \n \n \n

\n\n \n

\n \n

Loading data...

\n\n \n

\n \n \n \n

\n\n \n \n\n\n",[],[197],"f12c16ba-5a2d-11f1-83ca-ee650bb23af1","web-search.ts","f12c1778-5a2d-11f1-83ca-ee650bb23af1","9c18a224-5a2d-11f1-822b-ee650bb23af1",10,["D",1779928879136],"/**\n * Web Search Enrichment — fetches publicly available info about a company\n * to provide context for the AI classifier.\n *\n * Strategy:\n * 1. If domain exists, fetch the homepage and extract a description\n * 2. Search for \"{company name} + lab/testing/manufacturing\" for context\n */\n\nexport interface WebEnrichmentResult {\n context: string;\n sources: string[];\n}\n\n/**\n * Enrich a company by fetching its website and searching for context.\n * Returns a text summary suitable for the LLM classifier.\n */\nexport async function enrichCompany(\n name: string,\n domain: string | null,\n): Promise {\n const parts: string[] = [];\n const sources: string[] = [];\n\n // 1. Try fetching the company homepage\n if (domain) {\n try {\n const url = domain.startsWith(\"http\") ? domain : `https://${domain}`;\n const res = await fetch(url, {\n headers: { \"User-Agent\": \"QBench-Enrichment/1.0\" },\n signal: AbortSignal.timeout(8000),\n redirect: \"follow\",\n });\n if (res.ok) {\n const html = await res.text();\n const extracted = extractPageInfo(html);\n if (extracted) {\n parts.push(`[Website (${domain})]: ${extracted}`);\n sources.push(url);\n }\n }\n } catch {\n // Domain fetch failed — that's fine, we'll rely on search\n }\n }\n\n // 2. Web search for company context\n try {\n const searchQuery = buildSearchQuery(name, domain);\n const searchResult = await webSearch(searchQuery);\n if (searchResult) {\n parts.push(`[Search]: ${searchResult}`);\n sources.push(`search: \"${searchQuery}\"`);\n }\n } catch {\n // Search failed — classifier will work with whatever we have\n }\n\n return {\n context: parts.join(\"\\n\\n\") || \"No web enrichment data available.\",\n sources,\n };\n}\n\nfunction buildSearchQuery(name: string, domain: string | null): string {\n // Clean company name for search\n const cleanName = name.replace(/[,.\\-()]/g, \" \").replace(/\\s+/g, \" \").trim();\n // Limit to first 4 words to keep search focused\n const shortName = cleanName.split(\" \").slice(0, 4).join(\" \");\n return `${shortName} laboratory testing manufacturing`;\n}\n\n/**\n * Simple web search using a public search API.\n * Falls back gracefully if unavailable.\n */\nasync function webSearch(query: string): Promise {\n try {\n // Use DuckDuckGo instant answer API (no key required)\n const encoded = encodeURIComponent(query);\n const res = await fetch(\n `https://api.duckduckgo.com/?q=${encoded}&format=json&no_html=1&skip_disambig=1`,\n { signal: AbortSignal.timeout(5000) },\n );\n if (!res.ok) return null;\n const data = await res.json();\n\n const parts: string[] = [];\n if (data.Abstract) parts.push(data.Abstract);\n if (data.RelatedTopics?.length) {\n const topics = data.RelatedTopics\n .slice(0, 3)\n .map((t: any) => t.Text)\n .filter(Boolean);\n if (topics.length) parts.push(topics.join(\". \"));\n }\n return parts.join(\" \") || null;\n } catch {\n return null;\n }\n}\n\n/**\n * Extract useful text from an HTML page — title, meta description,\n * and first meaningful text content.\n */\nfunction extractPageInfo(html: string): string | null {\n const parts: string[] = [];\n\n // Title\n const titleMatch = html.match(/]*>(.*?)<\\/title>/si);\n if (titleMatch) parts.push(`Title: ${cleanText(titleMatch[1])}`);\n\n // Meta description\n const metaMatch = html.match(\n /]+name=[\"']description[\"'][^>]+content=[\"']([^\"']+)[\"']/si,\n );\n if (metaMatch) parts.push(`Description: ${cleanText(metaMatch[1])}`);\n\n // OG description as fallback\n if (!metaMatch) {\n const ogMatch = html.match(\n /]+property=[\"']og:description[\"'][^>]+content=[\"']([^\"']+)[\"']/si,\n );\n if (ogMatch) parts.push(`Description: ${cleanText(ogMatch[1])}`);\n }\n\n // Extract first ~500 chars of visible body text\n const bodyMatch = html.match(/]*>([\\s\\S]*?)<\\/body>/si);\n if (bodyMatch) {\n const stripped = bodyMatch[1]\n .replace(//gi, \"\")\n .replace(//gi, \"\")\n .replace(//gi, \"\")\n .replace(//gi, \"\")\n .replace(//gi, \"\")\n .replace(/<[^>]+>/g, \" \")\n .replace(/\\s+/g, \" \")\n .trim();\n if (stripped.length > 50) {\n parts.push(`Content: ${stripped.slice(0, 500)}`);\n }\n }\n\n return parts.length > 0 ? parts.join(\". \") : null;\n}\n\nfunction cleanText(text: string): string {\n return text\n .replace(/&/g, \"&\")\n .replace(/</g, \"<\")\n .replace(/>/g, \">\")\n .replace(/&#?\\w+;/g, \" \")\n .replace(/\\s+/g, \" \")\n .trim();\n}\n",[],[212,204],"enrichment","cc04dcbe-5a2d-11f1-af29-ee650bb23af1","taxonomy.ts","cc04ddae-5a2d-11f1-af29-ee650bb23af1",8,["D",1779928816802],"/**\n * QBench Industry Taxonomy v4.4\n * Encodes the full 3-tier industry hierarchy, ICP fit, segment rules,\n * lab operation types, and regulatory frameworks.\n *\n * DETERMINISTIC RULES are enforced here — the LLM only picks Tier 2 + Tier 3.\n * Everything else (Tier 1, ICP, Segment) is derived from lookups.\n */\n\n// ── Tier 2 → Tier 1 lookup (deterministic, never set by LLM) ──────────\nexport const TIER2_TO_TIER1: Record = {\n \"Therapeutic Development\": \"Life Sciences\",\n \"Contract Research & Manufacturing\": \"Life Sciences\",\n \"Life Science Tools & Platforms\": \"Life Sciences\",\n \"Precision Medicine & Diagnostics\": \"Life Sciences\",\n \"Food & Beverage Manufacturing\": \"Food & Beverage\",\n \"Contract Food & Beverage Testing\": \"Food & Beverage\",\n \"Health Supplements\": \"Food & Beverage\",\n \"Agriculture\": \"Agriculture\",\n \"Chemicals & Materials\": \"Manufacturing\",\n \"Metals, Minerals & Construction\": \"Manufacturing\",\n \"Engineered Products\": \"Manufacturing\",\n \"Packaging & Paper\": \"Manufacturing\",\n \"Contract Engineering & Analytical\": \"Manufacturing\",\n \"Consumer Goods\": \"Manufacturing\",\n \"Personal Care Products\": \"Manufacturing\",\n \"Oil, Gas & Petrochemical\": \"Energy & Natural Resources\",\n \"Power & Utilities\": \"Energy & Natural Resources\",\n \"Clean Energy & Storage\": \"Energy & Natural Resources\",\n \"Environmental Testing\": \"Environmental & Compliance Testing\",\n \"Product Safety & Compliance\": \"Environmental & Compliance Testing\",\n \"Cannabis & Hemp Testing\": \"Cannabis & Hemp Testing\",\n \"Electronics & Semiconductor\": \"Electronics & Semiconductor\",\n \"Independent Clinical\": \"Healthcare\",\n \"Veterinary Diagnostics\": \"Healthcare\",\n \"Hospital & Health System\": \"Healthcare\",\n \"University Research\": \"Academic & Government Research\",\n \"Government & National Research\": \"Academic & Government Research\",\n \"Forensic & Public Safety\": \"Forensic & Public Safety\",\n};\n\n// ── Tier 3 → ICP Fit lookup (deterministic, overrides mid-tier ICP) ────\nexport type ICPFitTier = \"Bullseye\" | \"Ring 1\" | \"Ring 2\" | \"Not Target\";\n\nexport const TIER3_TO_ICP: Record = {\n // Life Sciences\n \"Therapeutic Biotech (Preclinical–Phase II)\": \"Bullseye\",\n \"Pharma R&D & QC\": \"Ring 1\",\n \"Preclinical / Discovery CROs\": \"Bullseye\",\n \"Bioanalytical / GLP CROs (DMPK, PK/PD)\": \"Bullseye\",\n \"CDMOs\": \"Bullseye\",\n \"Platform/Tools Biotech\": \"Bullseye\",\n \"Molecular and Genetic Specialty\": \"Ring 1\",\n \"Infectious Disease\": \"Ring 1\",\n \"Cancer Diagnostic\": \"Ring 1\",\n // Food & Beverage\n \"Meat, Poultry & Seafood Processing\": \"Ring 1\",\n \"Dairy Processing\": \"Ring 1\",\n \"Beverage Production (Breweries, Wineries, Distilleries)\": \"Ring 1\",\n \"Bakery, Snacks & Confectionery\": \"Ring 1\",\n \"Fruit & Vegetable Processing\": \"Ring 1\",\n \"Ingredients, Flavors & Additives\": \"Ring 1\",\n \"Pet Food & Animal Food Manufacturing\": \"Ring 2\",\n \"Tobacco Manufacturing\": \"Not Target\",\n \"Contract Food Testing\": \"Ring 1\",\n \"Nutraceuticals\": \"Bullseye\",\n \"Health Supplements\": \"Bullseye\",\n \"Vitamins\": \"Bullseye\",\n // Agriculture\n \"Seed & Trait R&D\": \"Ring 2\",\n \"Soil Testing Services\": \"Ring 2\",\n \"Plant Pathology & Diagnostics\": \"Ring 2\",\n \"Public Agricultural Research\": \"Not Target\",\n \"Ag Biotech Startups (Biofertilizers, Biocontrol)\": \"Ring 2\",\n // Manufacturing — Chemicals & Materials\n \"Specialty Chemicals\": \"Bullseye\",\n \"Basic & Commodity Chemicals\": \"Ring 1\",\n \"Polymers, Resins & Plastics\": \"Bullseye\",\n \"Coatings, Paints & Adhesives\": \"Bullseye\",\n \"Fertilizers & Agrochemical Manufacturing\": \"Ring 2\",\n // Manufacturing — Metals, Minerals & Construction\n \"Metals & Metallurgy (Primary + Fabricated)\": \"Bullseye\",\n \"Mining & Mineral Processing\": \"Ring 2\",\n \"Battery & Rare-Earth Minerals Processing\": \"Ring 2\",\n \"Building Materials (Cement, Glass, Ceramics)\": \"Bullseye\",\n \"Engineered Wood & Lumber\": \"Ring 2\",\n \"Construction & Geotechnical Testing\": \"Ring 2\",\n // Manufacturing — Engineered Products\n \"Automotive & EV (OEMs + Suppliers)\": \"Bullseye\",\n \"Aerospace & Defense Manufacturing\": \"Bullseye\",\n \"Medical Devices & Diagnostics Manufacturing\": \"Bullseye\",\n \"Industrial Machinery & Instrumentation\": \"Ring 2\",\n \"Plastics & Rubber Products\": \"Bullseye\",\n \"Appliances & Home Equipment\": \"Bullseye\",\n // Manufacturing — Packaging & Paper\n \"Mills & Converters\": \"Ring 2\",\n // Manufacturing — Contract Engineering & Analytical\n \"Contract Physical & Engineering R&D\": \"Ring 1\",\n // Manufacturing — Consumer Goods\n \"Household Cleaning & Chemical Products\": \"Bullseye\",\n \"Contract Manufacturers (CPG)\": \"Bullseye\",\n \"Apparel, Footwear & Textiles\": \"Ring 2\",\n \"Sports & Recreation Equipment\": \"Ring 2\",\n \"Furniture\": \"Not Target\",\n // Manufacturing — Personal Care Products\n \"Cosmetics & Personal Care\": \"Bullseye\",\n \"Fragrance & Flavor Houses\": \"Bullseye\",\n // Energy & Natural Resources\n \"Refinery QC\": \"Bullseye\",\n \"Upstream Core Analysis & Reservoir Fluids\": \"Bullseye\",\n \"Petrochemical Process\": \"Bullseye\",\n \"Fuel & Lubricant Analysis Services\": \"Bullseye\",\n \"Power Plants\": \"Bullseye\",\n \"Water & Wastewater Utility\": \"Bullseye\",\n \"Energy Storage & Battery R&D\": \"Bullseye\",\n \"Renewable Energy\": \"Bullseye\",\n // Environmental & Compliance Testing\n \"Water & Wastewater Testing\": \"Ring 2\",\n \"Soil & Groundwater Testing\": \"Ring 2\",\n \"Air & Emissions Testing\": \"Ring 2\",\n \"Industrial Hygiene & Occupational Health\": \"Ring 2\",\n \"Ecology & Radiochemistry\": \"Not Target\",\n \"EMC / Compliance (NRTL & 3rd-Party)\": \"Ring 2\",\n \"Consumer Product Safety & Claims\": \"Ring 2\",\n // Cannabis & Hemp Testing\n \"Cannabis Compliance Testing\": \"Not Target\",\n \"Hemp / CBD Testing\": \"Not Target\",\n // Electronics & Semiconductor\n \"Front-End Fabs\": \"Ring 2\",\n \"OSAT / Test & Packaging\": \"Ring 2\",\n \"Device OEM R&D\": \"Ring 2\",\n \"PCB Assembly QA\": \"Ring 2\",\n // Healthcare (all Not Target)\n \"Regional Reference Labs\": \"Not Target\",\n \"Pathology / Anatomic Pathology Groups\": \"Not Target\",\n \"Toxicology & Drug Testing\": \"Not Target\",\n \"Fertility & Reproductive Medicine\": \"Not Target\",\n \"National Reference\": \"Not Target\",\n \"Private Vet Diagnostic\": \"Not Target\",\n \"Public Vet Diagnostic (AAVLD)\": \"Not Target\",\n \"Aquaculture Diagnostic\": \"Not Target\",\n \"Community Hospital\": \"Not Target\",\n \"Academic Medical Center\": \"Not Target\",\n \"Physician Office (POL)\": \"Not Target\",\n \"Outpatient Clinics\": \"Not Target\",\n \"Retail Clinic / Pharmacy\": \"Not Target\",\n \"Dialysis & Specialty Clinic\": \"Not Target\",\n // Academic & Government Research (all Not Target)\n \"R1 Biomedical Research\": \"Not Target\",\n \"R1 Engineering & Materials\": \"Not Target\",\n \"R2 / Other University Research\": \"Not Target\",\n \"University Core Facilities\": \"Not Target\",\n \"Federal National Research\": \"Not Target\",\n // Forensic & Public Safety (all Not Target)\n \"Public Crime\": \"Not Target\",\n \"Private Forensic\": \"Not Target\",\n \"Medicolegal Death Investigation\": \"Not Target\",\n};\n\n// ── Tier 3 → Parent Tier 2 lookup ──────────────────────────────────────\nexport const TIER3_TO_TIER2: Record = {\n \"Therapeutic Biotech (Preclinical–Phase II)\": \"Therapeutic Development\",\n \"Pharma R&D & QC\": \"Therapeutic Development\",\n \"Preclinical / Discovery CROs\": \"Contract Research & Manufacturing\",\n \"Bioanalytical / GLP CROs (DMPK, PK/PD)\": \"Contract Research & Manufacturing\",\n \"CDMOs\": \"Contract Research & Manufacturing\",\n \"Platform/Tools Biotech\": \"Life Science Tools & Platforms\",\n \"Molecular and Genetic Specialty\": \"Precision Medicine & Diagnostics\",\n \"Infectious Disease\": \"Precision Medicine & Diagnostics\",\n \"Cancer Diagnostic\": \"Precision Medicine & Diagnostics\",\n \"Meat, Poultry & Seafood Processing\": \"Food & Beverage Manufacturing\",\n \"Dairy Processing\": \"Food & Beverage Manufacturing\",\n \"Beverage Production (Breweries, Wineries, Distilleries)\": \"Food & Beverage Manufacturing\",\n \"Bakery, Snacks & Confectionery\": \"Food & Beverage Manufacturing\",\n \"Fruit & Vegetable Processing\": \"Food & Beverage Manufacturing\",\n \"Ingredients, Flavors & Additives\": \"Food & Beverage Manufacturing\",\n \"Pet Food & Animal Food Manufacturing\": \"Food & Beverage Manufacturing\",\n \"Tobacco Manufacturing\": \"Food & Beverage Manufacturing\",\n \"Contract Food Testing\": \"Contract Food & Beverage Testing\",\n \"Nutraceuticals\": \"Health Supplements\",\n \"Health Supplements\": \"Health Supplements\",\n \"Vitamins\": \"Health Supplements\",\n \"Seed & Trait R&D\": \"Agriculture\",\n \"Soil Testing Services\": \"Agriculture\",\n \"Plant Pathology & Diagnostics\": \"Agriculture\",\n \"Public Agricultural Research\": \"Agriculture\",\n \"Ag Biotech Startups (Biofertilizers, Biocontrol)\": \"Agriculture\",\n \"Specialty Chemicals\": \"Chemicals & Materials\",\n \"Basic & Commodity Chemicals\": \"Chemicals & Materials\",\n \"Polymers, Resins & Plastics\": \"Chemicals & Materials\",\n \"Coatings, Paints & Adhesives\": \"Chemicals & Materials\",\n \"Fertilizers & Agrochemical Manufacturing\": \"Chemicals & Materials\",\n \"Metals & Metallurgy (Primary + Fabricated)\": \"Metals, Minerals & Construction\",\n \"Mining & Mineral Processing\": \"Metals, Minerals & Construction\",\n \"Battery & Rare-Earth Minerals Processing\": \"Metals, Minerals & Construction\",\n \"Building Materials (Cement, Glass, Ceramics)\": \"Metals, Minerals & Construction\",\n \"Engineered Wood & Lumber\": \"Metals, Minerals & Construction\",\n \"Construction & Geotechnical Testing\": \"Metals, Minerals & Construction\",\n \"Automotive & EV (OEMs + Suppliers)\": \"Engineered Products\",\n \"Aerospace & Defense Manufacturing\": \"Engineered Products\",\n \"Medical Devices & Diagnostics Manufacturing\": \"Engineered Products\",\n \"Industrial Machinery & Instrumentation\": \"Engineered Products\",\n \"Plastics & Rubber Products\": \"Engineered Products\",\n \"Appliances & Home Equipment\": \"Engineered Products\",\n \"Mills & Converters\": \"Packaging & Paper\",\n \"Contract Physical & Engineering R&D\": \"Contract Engineering & Analytical\",\n \"Household Cleaning & Chemical Products\": \"Consumer Goods\",\n \"Contract Manufacturers (CPG)\": \"Consumer Goods\",\n \"Apparel, Footwear & Textiles\": \"Consumer Goods\",\n \"Sports & Recreation Equipment\": \"Consumer Goods\",\n \"Furniture\": \"Consumer Goods\",\n \"Cosmetics & Personal Care\": \"Personal Care Products\",\n \"Fragrance & Flavor Houses\": \"Personal Care Products\",\n \"Refinery QC\": \"Oil, Gas & Petrochemical\",\n \"Upstream Core Analysis & Reservoir Fluids\": \"Oil, Gas & Petrochemical\",\n \"Petrochemical Process\": \"Oil, Gas & Petrochemical\",\n \"Fuel & Lubricant Analysis Services\": \"Oil, Gas & Petrochemical\",\n \"Power Plants\": \"Power & Utilities\",\n \"Water & Wastewater Utility\": \"Power & Utilities\",\n \"Energy Storage & Battery R&D\": \"Clean Energy & Storage\",\n \"Renewable Energy\": \"Clean Energy & Storage\",\n \"Water & Wastewater Testing\": \"Environmental Testing\",\n \"Soil & Groundwater Testing\": \"Environmental Testing\",\n \"Air & Emissions Testing\": \"Environmental Testing\",\n \"Industrial Hygiene & Occupational Health\": \"Environmental Testing\",\n \"Ecology & Radiochemistry\": \"Environmental Testing\",\n \"EMC / Compliance (NRTL & 3rd-Party)\": \"Product Safety & Compliance\",\n \"Consumer Product Safety & Claims\": \"Product Safety & Compliance\",\n \"Cannabis Compliance Testing\": \"Cannabis & Hemp Testing\",\n \"Hemp / CBD Testing\": \"Cannabis & Hemp Testing\",\n \"Front-End Fabs\": \"Electronics & Semiconductor\",\n \"OSAT / Test & Packaging\": \"Electronics & Semiconductor\",\n \"Device OEM R&D\": \"Electronics & Semiconductor\",\n \"PCB Assembly QA\": \"Electronics & Semiconductor\",\n \"Regional Reference Labs\": \"Independent Clinical\",\n \"Pathology / Anatomic Pathology Groups\": \"Independent Clinical\",\n \"Toxicology & Drug Testing\": \"Independent Clinical\",\n \"Fertility & Reproductive Medicine\": \"Independent Clinical\",\n \"National Reference\": \"Independent Clinical\",\n \"Private Vet Diagnostic\": \"Veterinary Diagnostics\",\n \"Public Vet Diagnostic (AAVLD)\": \"Veterinary Diagnostics\",\n \"Aquaculture Diagnostic\": \"Veterinary Diagnostics\",\n \"Community Hospital\": \"Hospital & Health System\",\n \"Academic Medical Center\": \"Hospital & Health System\",\n \"Physician Office (POL)\": \"Hospital & Health System\",\n \"Outpatient Clinics\": \"Hospital & Health System\",\n \"Retail Clinic / Pharmacy\": \"Hospital & Health System\",\n \"Dialysis & Specialty Clinic\": \"Hospital & Health System\",\n \"R1 Biomedical Research\": \"University Research\",\n \"R1 Engineering & Materials\": \"University Research\",\n \"R2 / Other University Research\": \"University Research\",\n \"University Core Facilities\": \"University Research\",\n \"Federal National Research\": \"Government & National Research\",\n \"Public Crime\": \"Forensic & Public Safety\",\n \"Private Forensic\": \"Forensic & Public Safety\",\n \"Medicolegal Death Investigation\": \"Forensic & Public Safety\",\n};\n\n// ── Segment from employee count (deterministic) ────────────────────────\nexport type Segment = \"SMB\" | \"Scaled SMB\" | \"Mid-Market\" | \"Enterprise\";\n\nexport function getSegmentFromEmployeeCount(count: number | null): Segment | null {\n if (count === null || count === undefined) return null;\n if (count < 50) return \"SMB\";\n if (count < 250) return \"Scaled SMB\";\n if (count <= 2000) return \"Mid-Market\";\n return \"Enterprise\";\n}\n\n// ── Lab Operation Types ────────────────────────────────────────────────\nexport const LAB_OPERATION_TYPES = [\n \"Process / Continuous\",\n \"Discrete / Lot-Based\",\n \"Research / Non-Production\",\n] as const;\nexport type LabOperationType = typeof LAB_OPERATION_TYPES[number];\n\n// ── Regulatory Frameworks (multi-select) ───────────────────────────────\nexport const REGULATORY_FRAMEWORKS = [\n \"GxP (FDA/EMA)\",\n \"FDA FSMA / USDA\",\n \"NELAP / ISO 17025\",\n \"CLIA / CAP\",\n \"ASTM / ISO 9001\",\n \"EPA Compliance\",\n \"USDA / APHIS\",\n] as const;\nexport type RegulatoryFramework = typeof REGULATORY_FRAMEWORKS[number];\n\n// ── Tier 1 values (for validation) ─────────────────────────────────────\nexport const TIER1_VALUES = [\n \"Life Sciences\",\n \"Food & Beverage\",\n \"Agriculture\",\n \"Manufacturing\",\n \"Energy & Natural Resources\",\n \"Environmental & Compliance Testing\",\n \"Cannabis & Hemp Testing\",\n \"Electronics & Semiconductor\",\n \"Healthcare\",\n \"Academic & Government Research\",\n \"Forensic & Public Safety\",\n] as const;\n\n// ── All valid Tier 2 values ────────────────────────────────────────────\nexport const TIER2_VALUES = Object.keys(TIER2_TO_TIER1);\n\n// ── All valid Tier 3 values ────────────────────────────────────────────\nexport const TIER3_VALUES = Object.keys(TIER3_TO_ICP);\n\n// ── Derive all fields from Tier 3 (the single source of truth) ─────────\nexport function deriveFromTier3(tier3: string): {\n tier2: string;\n tier1: string;\n icpFitTier: ICPFitTier;\n} | null {\n const tier2 = TIER3_TO_TIER2[tier3];\n if (!tier2) return null;\n const tier1 = TIER2_TO_TIER1[tier2];\n if (!tier1) return null;\n const icpFitTier = TIER3_TO_ICP[tier3];\n if (!icpFitTier) return null;\n return { tier2, tier1, icpFitTier };\n}\n\n// ── Contract Model per Tier 3 ──────────────────────────────────────────\nexport type ContractModel = \"In-House\" | \"Contract\" | \"Mixed\";\n\nexport const TIER3_CONTRACT_MODEL: Record = {\n \"Therapeutic Biotech (Preclinical–Phase II)\": \"In-House\",\n \"Pharma R&D & QC\": \"In-House\",\n \"Preclinical / Discovery CROs\": \"Contract\",\n \"Bioanalytical / GLP CROs (DMPK, PK/PD)\": \"Contract\",\n \"CDMOs\": \"Contract\",\n \"Platform/Tools Biotech\": \"In-House\",\n \"Molecular and Genetic Specialty\": \"Mixed\",\n \"Infectious Disease\": \"Mixed\",\n \"Cancer Diagnostic\": \"Mixed\",\n \"Meat, Poultry & Seafood Processing\": \"In-House\",\n \"Dairy Processing\": \"In-House\",\n \"Beverage Production (Breweries, Wineries, Distilleries)\": \"In-House\",\n \"Bakery, Snacks & Confectionery\": \"In-House\",\n \"Fruit & Vegetable Processing\": \"In-House\",\n \"Ingredients, Flavors & Additives\": \"In-House\",\n \"Pet Food & Animal Food Manufacturing\": \"In-House\",\n \"Tobacco Manufacturing\": \"In-House\",\n \"Contract Food Testing\": \"Contract\",\n \"Nutraceuticals\": \"Mixed\",\n \"Health Supplements\": \"Mixed\",\n \"Vitamins\": \"Mixed\",\n \"Seed & Trait R&D\": \"In-House\",\n \"Soil Testing Services\": \"Contract\",\n \"Plant Pathology & Diagnostics\": \"Mixed\",\n \"Public Agricultural Research\": \"In-House\",\n \"Ag Biotech Startups (Biofertilizers, Biocontrol)\": \"In-House\",\n \"Specialty Chemicals\": \"In-House\",\n \"Basic & Commodity Chemicals\": \"In-House\",\n \"Polymers, Resins & Plastics\": \"In-House\",\n \"Coatings, Paints & Adhesives\": \"In-House\",\n \"Fertilizers & Agrochemical Manufacturing\": \"In-House\",\n \"Metals & Metallurgy (Primary + Fabricated)\": \"Mixed\",\n \"Mining & Mineral Processing\": \"Mixed\",\n \"Battery & Rare-Earth Minerals Processing\": \"Mixed\",\n \"Building Materials (Cement, Glass, Ceramics)\": \"In-House\",\n \"Engineered Wood & Lumber\": \"In-House\",\n \"Construction & Geotechnical Testing\": \"Contract\",\n \"Automotive & EV (OEMs + Suppliers)\": \"In-House\",\n \"Aerospace & Defense Manufacturing\": \"In-House\",\n \"Medical Devices & Diagnostics Manufacturing\": \"In-House\",\n \"Industrial Machinery & Instrumentation\": \"In-House\",\n \"Plastics & Rubber Products\": \"In-House\",\n \"Appliances & Home Equipment\": \"In-House\",\n \"Mills & Converters\": \"In-House\",\n \"Contract Physical & Engineering R&D\": \"Contract\",\n \"Household Cleaning & Chemical Products\": \"In-House\",\n \"Contract Manufacturers (CPG)\": \"Contract\",\n \"Apparel, Footwear & Textiles\": \"Mixed\",\n \"Sports & Recreation Equipment\": \"Mixed\",\n \"Furniture\": \"In-House\",\n \"Cosmetics & Personal Care\": \"In-House\",\n \"Fragrance & Flavor Houses\": \"In-House\",\n \"Refinery QC\": \"In-House\",\n \"Upstream Core Analysis & Reservoir Fluids\": \"In-House\",\n \"Petrochemical Process\": \"In-House\",\n \"Fuel & Lubricant Analysis Services\": \"Contract\",\n \"Power Plants\": \"In-House\",\n \"Water & Wastewater Utility\": \"In-House\",\n \"Energy Storage & Battery R&D\": \"In-House\",\n \"Renewable Energy\": \"In-House\",\n \"Water & Wastewater Testing\": \"Contract\",\n \"Soil & Groundwater Testing\": \"Contract\",\n \"Air & Emissions Testing\": \"Contract\",\n \"Industrial Hygiene & Occupational Health\": \"Contract\",\n \"Ecology & Radiochemistry\": \"Contract\",\n \"EMC / Compliance (NRTL & 3rd-Party)\": \"Contract\",\n \"Consumer Product Safety & Claims\": \"Contract\",\n \"Cannabis Compliance Testing\": \"Contract\",\n \"Hemp / CBD Testing\": \"Contract\",\n \"Front-End Fabs\": \"In-House\",\n \"OSAT / Test & Packaging\": \"In-House\",\n \"Device OEM R&D\": \"In-House\",\n \"PCB Assembly QA\": \"In-House\",\n \"Regional Reference Labs\": \"Contract\",\n \"Pathology / Anatomic Pathology Groups\": \"Contract\",\n \"Toxicology & Drug Testing\": \"Contract\",\n \"Fertility & Reproductive Medicine\": \"Contract\",\n \"National Reference\": \"Contract\",\n \"Private Vet Diagnostic\": \"Contract\",\n \"Public Vet Diagnostic (AAVLD)\": \"Contract\",\n \"Aquaculture Diagnostic\": \"Contract\",\n \"Community Hospital\": \"In-House\",\n \"Academic Medical Center\": \"In-House\",\n \"Physician Office (POL)\": \"In-House\",\n \"Outpatient Clinics\": \"In-House\",\n \"Retail Clinic / Pharmacy\": \"In-House\",\n \"Dialysis & Specialty Clinic\": \"In-House\",\n \"R1 Biomedical Research\": \"In-House\",\n \"R1 Engineering & Materials\": \"In-House\",\n \"R2 / Other University Research\": \"In-House\",\n \"University Core Facilities\": \"Contract\",\n \"Federal National Research\": \"In-House\",\n \"Public Crime\": \"In-House\",\n \"Private Forensic\": \"Contract\",\n \"Medicolegal Death Investigation\": \"In-House\",\n};\n",[],[212,214],"1a823c7e-5a2e-11f1-bf62-ee650bb23af1","pipeline.ts","1a823d28-5a2e-11f1-bf62-ee650bb23af1",12,["D",1779928948486],"/**\n * Enrichment Pipeline — orchestrates the full flow:\n * 1. Fetch companies from HubSpot (paginated)\n * 2. Web-enrich each company\n * 3. Classify with AI\n * 4. Store results in SQLite\n *\n * Handles rate limiting, retries, and progress tracking.\n */\nimport { type CompanyProfile, classifyCompany } from \"./classifier.ts\";\nimport { enrichCompany } from \"./web-search.ts\";\nimport {\n initDB,\n createRun,\n updateRunProgress,\n completeRun,\n upsertClassification,\n} from \"./db.ts\";\n\nconst TOKEN = Deno.env.get(\"TOKEN\") || \"\";\nconst HEADERS = {\n Authorization: `Bearer ${TOKEN}`,\n \"Content-Type\": \"application/json\",\n};\n\n// ── HubSpot paginated company fetch ────────────────────────────────────\n\ninterface HubSpotCompany {\n id: string;\n properties: Record;\n}\n\nasync function fetchAllCompanies(\n limit: number = 100,\n maxPages: number = 100,\n): Promise {\n const allCompanies: HubSpotCompany[] = [];\n let after: string | undefined;\n const properties = [\n \"name\", \"domain\", \"industry\", \"numberofemployees\",\n \"city\", \"state\", \"country\", \"description\", \"createdate\",\n ].join(\",\");\n\n for (let page = 0; page < maxPages; page++) {\n const url = new URL(\"https://api.hubapi.com/crm/v3/objects/companies\");\n url.searchParams.set(\"limit\", String(limit));\n url.searchParams.set(\"properties\", properties);\n if (after) url.searchParams.set(\"after\", after);\n\n const res = await fetch(url.toString(), { headers: HEADERS });\n if (!res.ok) {\n const err = await res.text();\n throw new Error(`HubSpot API error (${res.status}): ${err}`);\n }\n\n const data = await res.json();\n allCompanies.push(...(data.results || []));\n\n if (data.paging?.next?.after) {\n after = data.paging.next.after;\n } else {\n break; // No more pages\n }\n\n // Respect rate limits — brief pause between pages\n await sleep(100);\n }\n\n return allCompanies;\n}\n\n// ── Process a single company ───────────────────────────────────────────\n\nasync function processCompany(\n company: HubSpotCompany,\n runId: number,\n): Promise<{ success: boolean; error?: string }> {\n const p = company.properties;\n const name = p.name || \"Unknown\";\n const domain = p.domain || null;\n\n try {\n // Step 1: Web enrichment\n const enrichment = await enrichCompany(name, domain);\n\n // Step 2: Build profile\n const profile: CompanyProfile = {\n hubspotId: company.id,\n name,\n domain,\n industry: p.industry || null,\n employeeCount: p.numberofemployees ? parseInt(p.numberofemployees) : null,\n city: p.city || null,\n state: p.state || null,\n country: p.country || null,\n description: p.description || null,\n webSearchContext: enrichment.context,\n };\n\n // Step 3: AI classification\n const result = await classifyCompany(profile);\n\n // Step 4: Store result\n await upsertClassification({\n runId,\n hubspotId: company.id,\n companyName: name,\n domain,\n employeeCount: profile.employeeCount,\n existingIndustry: p.industry || null,\n tier1: result.tier1,\n tier2: result.tier2,\n tier3: result.tier3,\n icpFitTier: result.icpFitTier,\n segment: result.segment,\n labOperationType: result.labOperationType,\n regulatoryFrameworks: result.regulatoryFrameworks,\n confidence: result.confidence,\n reasoning: result.reasoning,\n webEnrichmentContext: enrichment.context,\n });\n\n return { success: true };\n } catch (error: any) {\n console.error(`Error processing ${name} (${company.id}):`, error.message);\n return { success: false, error: error.message };\n }\n}\n\n// ── Main pipeline runner ───────────────────────────────────────────────\n\nexport interface PipelineOptions {\n batchSize?: number; // companies to process per batch before pausing\n delayBetweenMs?: number; // ms delay between companies (rate limiting)\n maxCompanies?: number; // cap for testing (0 = all)\n dryRun?: boolean; // if true, fetch but don't classify\n}\n\nexport interface PipelineProgress {\n runId: number;\n total: number;\n processed: number;\n succeeded: number;\n failed: number;\n status: \"running\" | \"completed\" | \"error\";\n errors: Array<{ company: string; error: string }>;\n}\n\nexport async function runPipeline(\n options: PipelineOptions = {},\n): Promise {\n const {\n batchSize = 10,\n delayBetweenMs = 2000, // 2s between companies (LLM + web search cost)\n maxCompanies = 0,\n dryRun = false,\n } = options;\n\n // Initialize database\n await initDB();\n\n // Fetch all companies from HubSpot\n console.log(\"Fetching companies from HubSpot...\");\n let companies = await fetchAllCompanies();\n console.log(`Fetched ${companies.length} companies`);\n\n if (maxCompanies > 0) {\n companies = companies.slice(0, maxCompanies);\n console.log(`Limited to ${companies.length} for this run`);\n }\n\n const runId = await createRun(companies.length);\n\n const progress: PipelineProgress = {\n runId,\n total: companies.length,\n processed: 0,\n succeeded: 0,\n failed: 0,\n status: \"running\",\n errors: [],\n };\n\n if (dryRun) {\n console.log(\"DRY RUN — fetched companies but skipping classification\");\n progress.status = \"completed\";\n return progress;\n }\n\n // Process in batches\n for (let i = 0; i < companies.length; i++) {\n const company = companies[i];\n const result = await processCompany(company, runId);\n\n progress.processed++;\n if (result.success) {\n progress.succeeded++;\n } else {\n progress.failed++;\n progress.errors.push({\n company: company.properties.name || company.id,\n error: result.error || \"Unknown error\",\n });\n }\n\n // Update run progress every 10 companies\n if (progress.processed % 10 === 0 || progress.processed === companies.length) {\n await updateRunProgress(runId, progress.processed);\n console.log(\n `Progress: ${progress.processed}/${progress.total} ` +\n `(${progress.succeeded} ok, ${progress.failed} failed)`,\n );\n }\n\n // Rate limit delay\n if (i < companies.length - 1) {\n await sleep(delayBetweenMs);\n }\n }\n\n await completeRun(runId);\n progress.status = \"completed\";\n\n console.log(\n `Pipeline complete: ${progress.succeeded}/${progress.total} succeeded, ` +\n `${progress.failed} failed`,\n );\n\n return progress;\n}\n\nfunction sleep(ms: number): Promise {\n return new Promise((resolve) => setTimeout(resolve, ms));\n}\n",[],[212,222],"081e5194-5a2e-11f1-9d85-ee650bb23af1","db.ts","081e528e-5a2e-11f1-9d85-ee650bb23af1",11,["D",1779928917630],"/**\n * SQLite schema and queries for the enrichment pipeline.\n * Stores classification results, review status, and audit log.\n */\nimport { sqlite } from \"https://esm.town/v/std/sqlite/main.ts\";\n\nexport async function initDB() {\n await sqlite.batch([\n {\n sql: `CREATE TABLE IF NOT EXISTS enrichment_runs (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n started_at TEXT NOT NULL DEFAULT (datetime('now')),\n completed_at TEXT,\n total_companies INTEGER DEFAULT 0,\n processed INTEGER DEFAULT 0,\n status TEXT NOT NULL DEFAULT 'running'\n )`,\n },\n {\n sql: `CREATE TABLE IF NOT EXISTS company_classifications (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n run_id INTEGER,\n hubspot_id TEXT NOT NULL,\n company_name TEXT NOT NULL,\n domain TEXT,\n employee_count INTEGER,\n -- Existing HubSpot values\n existing_industry TEXT,\n -- AI Classification results\n tier1 TEXT,\n tier2 TEXT,\n tier3 TEXT,\n icp_fit_tier TEXT,\n segment TEXT,\n lab_operation_type TEXT,\n regulatory_frameworks TEXT,\n confidence INTEGER,\n reasoning TEXT,\n web_enrichment_context TEXT,\n -- Review status\n review_status TEXT NOT NULL DEFAULT 'pending',\n reviewed_by TEXT,\n reviewed_at TEXT,\n review_notes TEXT,\n -- Push status\n pushed_to_hubspot INTEGER DEFAULT 0,\n pushed_at TEXT,\n -- Timestamps\n created_at TEXT NOT NULL DEFAULT (datetime('now')),\n updated_at TEXT NOT NULL DEFAULT (datetime('now')),\n UNIQUE(hubspot_id)\n )`,\n },\n {\n sql: `CREATE TABLE IF NOT EXISTS audit_log (\n id INTEGER PRIMARY KEY AUTOINCREMENT,\n hubspot_id TEXT NOT NULL,\n action TEXT NOT NULL,\n old_values TEXT,\n new_values TEXT,\n performed_by TEXT DEFAULT 'system',\n created_at TEXT NOT NULL DEFAULT (datetime('now'))\n )`,\n },\n ]);\n}\n\nexport interface StoredClassification {\n id: number;\n run_id: number;\n hubspot_id: string;\n company_name: string;\n domain: string | null;\n employee_count: number | null;\n existing_industry: string | null;\n tier1: string;\n tier2: string;\n tier3: string;\n icp_fit_tier: string;\n segment: string | null;\n lab_operation_type: string | null;\n regulatory_frameworks: string | null;\n confidence: number;\n reasoning: string;\n web_enrichment_context: string | null;\n review_status: string;\n reviewed_by: string | null;\n reviewed_at: string | null;\n review_notes: string | null;\n pushed_to_hubspot: number;\n pushed_at: string | null;\n created_at: string;\n updated_at: string;\n}\n\nexport async function upsertClassification(data: {\n runId: number;\n hubspotId: string;\n companyName: string;\n domain: string | null;\n employeeCount: number | null;\n existingIndustry: string | null;\n tier1: string;\n tier2: string;\n tier3: string;\n icpFitTier: string;\n segment: string | null;\n labOperationType: string | null;\n regulatoryFrameworks: string[];\n confidence: number;\n reasoning: string;\n webEnrichmentContext: string | null;\n}) {\n await sqlite.execute({\n sql: `INSERT INTO company_classifications (\n run_id, hubspot_id, company_name, domain, employee_count,\n existing_industry, tier1, tier2, tier3, icp_fit_tier,\n segment, lab_operation_type, regulatory_frameworks,\n confidence, reasoning, web_enrichment_context,\n review_status, updated_at\n ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'pending', datetime('now'))\n ON CONFLICT(hubspot_id) DO UPDATE SET\n run_id = excluded.run_id,\n company_name = excluded.company_name,\n domain = excluded.domain,\n employee_count = excluded.employee_count,\n existing_industry = excluded.existing_industry,\n tier1 = excluded.tier1,\n tier2 = excluded.tier2,\n tier3 = excluded.tier3,\n icp_fit_tier = excluded.icp_fit_tier,\n segment = excluded.segment,\n lab_operation_type = excluded.lab_operation_type,\n regulatory_frameworks = excluded.regulatory_frameworks,\n confidence = excluded.confidence,\n reasoning = excluded.reasoning,\n web_enrichment_context = excluded.web_enrichment_context,\n review_status = CASE WHEN company_classifications.pushed_to_hubspot = 1 THEN company_classifications.review_status ELSE 'pending' END,\n updated_at = datetime('now')`,\n args: [\n data.runId,\n data.hubspotId,\n data.companyName,\n data.domain,\n data.employeeCount,\n data.existingIndustry,\n data.tier1,\n data.tier2,\n data.tier3,\n data.icpFitTier,\n data.segment,\n data.labOperationType,\n JSON.stringify(data.regulatoryFrameworks),\n data.confidence,\n data.reasoning,\n data.webEnrichmentContext,\n ],\n });\n}\n\nexport async function getClassifications(filters: {\n status?: string;\n icpFitTier?: string;\n minConfidence?: number;\n limit?: number;\n offset?: number;\n}): Promise<{ rows: StoredClassification[]; total: number }> {\n const where: string[] = [];\n const args: any[] = [];\n\n if (filters.status) {\n where.push(\"review_status = ?\");\n args.push(filters.status);\n }\n if (filters.icpFitTier) {\n where.push(\"icp_fit_tier = ?\");\n args.push(filters.icpFitTier);\n }\n if (filters.minConfidence !== undefined) {\n where.push(\"confidence >= ?\");\n args.push(filters.minConfidence);\n }\n\n const whereClause = where.length > 0 ? `WHERE ${where.join(\" AND \")}` : \"\";\n const limit = filters.limit || 50;\n const offset = filters.offset || 0;\n\n const countResult = await sqlite.execute({\n sql: `SELECT COUNT(*) as count FROM company_classifications ${whereClause}`,\n args,\n });\n const total = (countResult.rows[0] as any)?.count || 0;\n\n const result = await sqlite.execute({\n sql: `SELECT * FROM company_classifications ${whereClause}\n ORDER BY confidence DESC, company_name ASC\n LIMIT ? OFFSET ?`,\n args: [...args, limit, offset],\n });\n\n return { rows: result.rows as StoredClassification[], total };\n}\n\nexport async function updateReviewStatus(\n hubspotId: string,\n status: \"approved\" | \"rejected\",\n notes?: string,\n) {\n await sqlite.execute({\n sql: `UPDATE company_classifications\n SET review_status = ?, reviewed_at = datetime('now'), review_notes = ?, updated_at = datetime('now')\n WHERE hubspot_id = ?`,\n args: [status, notes || null, hubspotId],\n });\n}\n\nexport async function bulkUpdateReviewStatus(\n hubspotIds: string[],\n status: \"approved\" | \"rejected\",\n) {\n const placeholders = hubspotIds.map(() => \"?\").join(\",\");\n await sqlite.execute({\n sql: `UPDATE company_classifications\n SET review_status = ?, reviewed_at = datetime('now'), updated_at = datetime('now')\n WHERE hubspot_id IN (${placeholders})`,\n args: [status, ...hubspotIds],\n });\n}\n\nexport async function createRun(totalCompanies: number): Promise {\n const result = await sqlite.execute({\n sql: `INSERT INTO enrichment_runs (total_companies) VALUES (?)`,\n args: [totalCompanies],\n });\n return Number(result.lastInsertRowid);\n}\n\nexport async function updateRunProgress(runId: number, processed: number) {\n await sqlite.execute({\n sql: `UPDATE enrichment_runs SET processed = ? WHERE id = ?`,\n args: [processed, runId],\n });\n}\n\nexport async function completeRun(runId: number) {\n await sqlite.execute({\n sql: `UPDATE enrichment_runs SET status = 'completed', completed_at = datetime('now') WHERE id = ?`,\n args: [runId],\n });\n}\n\nexport async function getLatestRun() {\n const result = await sqlite.execute(\n `SELECT * FROM enrichment_runs ORDER BY id DESC LIMIT 1`,\n );\n return result.rows[0] || null;\n}\n\nexport async function getStats() {\n const result = await sqlite.execute(`\n SELECT\n COUNT(*) as total,\n SUM(CASE WHEN review_status = 'pending' THEN 1 ELSE 0 END) as pending,\n SUM(CASE WHEN review_status = 'approved' THEN 1 ELSE 0 END) as approved,\n SUM(CASE WHEN review_status = 'rejected' THEN 1 ELSE 0 END) as rejected,\n SUM(CASE WHEN pushed_to_hubspot = 1 THEN 1 ELSE 0 END) as pushed,\n AVG(confidence) as avg_confidence,\n SUM(CASE WHEN icp_fit_tier = 'Bullseye' THEN 1 ELSE 0 END) as bullseye,\n SUM(CASE WHEN icp_fit_tier = 'Ring 1' THEN 1 ELSE 0 END) as ring1,\n SUM(CASE WHEN icp_fit_tier = 'Ring 2' THEN 1 ELSE 0 END) as ring2,\n SUM(CASE WHEN icp_fit_tier = 'Not Target' THEN 1 ELSE 0 END) as not_target\n FROM company_classifications\n `);\n return result.rows[0] || {};\n}\n",[],[212,230],"e1f0c2b8-5a2d-11f1-83ca-ee650bb23af1","classifier.ts","e1f0c376-5a2d-11f1-83ca-ee650bb23af1",9,["D",1779928853581],"/**\n * AI Classifier — Uses Claude to classify companies into the QBench taxonomy.\n * The LLM picks Tier 2 (mid-tier) and Tier 3 (niche).\n * Everything else is derived deterministically from taxonomy lookups.\n */\nimport Anthropic from \"npm:@anthropic-ai/sdk@0.39.0\";\nimport {\n TIER2_VALUES,\n TIER3_VALUES,\n TIER3_TO_TIER2,\n LAB_OPERATION_TYPES,\n REGULATORY_FRAMEWORKS,\n deriveFromTier3,\n getSegmentFromEmployeeCount,\n type ICPFitTier,\n type LabOperationType,\n type RegulatoryFramework,\n type Segment,\n} from \"./taxonomy.ts\";\n\nconst client = new Anthropic();\n\nexport interface CompanyProfile {\n hubspotId: string;\n name: string;\n domain: string | null;\n industry: string | null; // existing HubSpot industry value\n employeeCount: number | null;\n city: string | null;\n state: string | null;\n country: string | null;\n description: string | null; // from HubSpot\n webSearchContext: string | null; // from web enrichment\n}\n\nexport interface ClassificationResult {\n tier3: string;\n tier2: string;\n tier1: string;\n icpFitTier: ICPFitTier;\n segment: Segment | null;\n labOperationType: LabOperationType | null;\n regulatoryFrameworks: RegulatoryFramework[];\n confidence: number; // 0-100\n reasoning: string;\n}\n\nconst SYSTEM_PROMPT = `You are an expert B2B industry classifier for QBench, a Laboratory Information Management System (LIMS) company. Your job is to classify companies into QBench's proprietary industry taxonomy based on all available evidence.\n\n## YOUR TASK\nGiven a company profile (name, domain, location, employee count, existing HubSpot industry, web search context), classify it into:\n1. **Tier 3 (Niche)** — the most specific classification. This is the PRIMARY output.\n2. **Lab Operation Type** — Process/Continuous, Discrete/Lot-Based, or Research/Non-Production\n3. **Regulatory Frameworks** — which regulatory regimes apply (can be multiple or none)\n4. **Confidence** — 0-100 score reflecting classification certainty\n5. **Reasoning** — 2-3 sentences explaining your classification logic\n\n## VALID TIER 3 VALUES (you MUST pick exactly one):\n${TIER3_VALUES.map(v => `- ${v}`).join(\"\\n\")}\n\n## VALID LAB OPERATION TYPES (pick exactly one, or null if unclear):\n${Array.from(LAB_OPERATION_TYPES).map(v => `- ${v}`).join(\"\\n\")}\n\n## VALID REGULATORY FRAMEWORKS (pick zero or more):\n${Array.from(REGULATORY_FRAMEWORKS).map(v => `- ${v}`).join(\"\\n\")}\n\n## CLASSIFICATION RULES\n- Focus on what the company DOES and what KIND OF LAB they operate (or would operate)\n- A \"lab\" in QBench's context = any facility doing testing, QC, R&D, or analytical work\n- If a company has no lab or testing component at all, classify into the CLOSEST industry niche anyway — the ICP system will handle disqualification\n- Prefer SPECIFIC niches over general ones\n- If a company could fit multiple niches, pick the PRIMARY business activity\n- Use employee count for segment sizing but NOT for niche selection\n- Web search context is your most valuable signal — use it heavily\n\n## CONFIDENCE SCORING\n- 90-100: Clear match — company description, domain, and industry all align to one niche\n- 70-89: Strong match — most signals align, minor ambiguity\n- 50-69: Moderate — some signals conflict or limited info available\n- 30-49: Low — mostly guessing from company name/domain alone\n- 0-29: Very low — almost no useful information\n\n## RESPONSE FORMAT\nRespond with ONLY a JSON object, no markdown:\n{\n \"tier3\": \"exact Tier 3 value from the list above\",\n \"labOperationType\": \"Process / Continuous\" | \"Discrete / Lot-Based\" | \"Research / Non-Production\" | null,\n \"regulatoryFrameworks\": [\"framework1\", \"framework2\"],\n \"confidence\": 75,\n \"reasoning\": \"Brief explanation of why this classification was chosen.\"\n}`;\n\nexport async function classifyCompany(profile: CompanyProfile): Promise {\n const userMessage = buildUserMessage(profile);\n\n const response = await client.messages.create({\n model: \"claude-sonnet-4-20250514\",\n max_tokens: 500,\n temperature: 0.1, // low temp for consistency\n system: SYSTEM_PROMPT,\n messages: [{ role: \"user\", content: userMessage }],\n });\n\n const text = response.content[0].type === \"text\" ? response.content[0].text : \"\";\n const parsed = parseClassificationResponse(text, profile.employeeCount);\n\n return parsed;\n}\n\nfunction buildUserMessage(profile: CompanyProfile): string {\n const parts = [`## Company Profile`];\n parts.push(`**Name:** ${profile.name}`);\n if (profile.domain) parts.push(`**Domain:** ${profile.domain}`);\n if (profile.industry) parts.push(`**Existing HubSpot Industry:** ${profile.industry}`);\n if (profile.employeeCount) parts.push(`**Employee Count:** ${profile.employeeCount}`);\n const location = [profile.city, profile.state, profile.country].filter(Boolean).join(\", \");\n if (location) parts.push(`**Location:** ${location}`);\n if (profile.description) parts.push(`**Description:** ${profile.description}`);\n if (profile.webSearchContext) {\n parts.push(`\\n## Web Search Context`);\n parts.push(profile.webSearchContext);\n }\n return parts.join(\"\\n\");\n}\n\nfunction parseClassificationResponse(text: string, employeeCount: number | null): ClassificationResult {\n // Strip markdown code fences if present\n const cleaned = text.replace(/```json\\n?/g, \"\").replace(/```\\n?/g, \"\").trim();\n const parsed = JSON.parse(cleaned);\n\n const tier3 = parsed.tier3;\n if (!TIER3_VALUES.includes(tier3)) {\n throw new Error(`Invalid Tier 3 value from LLM: \"${tier3}\"`);\n }\n\n // Derive Tier 2, Tier 1, and ICP deterministically\n const derived = deriveFromTier3(tier3);\n if (!derived) {\n throw new Error(`Could not derive taxonomy from Tier 3: \"${tier3}\"`);\n }\n\n // Validate lab operation type\n const labOp = parsed.labOperationType;\n const validLabOp = labOp && LAB_OPERATION_TYPES.includes(labOp) ? labOp : null;\n\n // Validate regulatory frameworks\n const regFrameworks = (parsed.regulatoryFrameworks || []).filter(\n (f: string) => REGULATORY_FRAMEWORKS.includes(f as any)\n );\n\n return {\n tier3,\n tier2: derived.tier2,\n tier1: derived.tier1,\n icpFitTier: derived.icpFitTier,\n segment: getSegmentFromEmployeeCount(employeeCount),\n labOperationType: validLabOp,\n regulatoryFrameworks: regFrameworks,\n confidence: Math.max(0, Math.min(100, parsed.confidence || 0)),\n reasoning: parsed.reasoning || \"No reasoning provided\",\n };\n}\n",[],[212,238],"9c18a184-5a2d-11f1-822b-ee650bb23af1",7,["D",1779928736398],"directory",[166,167,168,169,170],[212],["D",1778773249291],["D",1789509672838],{"_11":12,"_124":-5},{"_11":12,"_120":12,"_68":-5,"_30":-5,"_121":122,"_123":24},{"_38":111,"_13":16,"_32":35,"_112":71,"_113":-5,"_114":-5,"_115":-5,"_116":-5,"_117":-5,"_118":14,"_119":12},"ctx",{"_52":-5,"_143":289,"_290":291,"_292":293,"_294":24,"_295":24,"_296":24,"_297":84,"_298":-5,"_299":-7,"_300":-7,"_301":26,"_302":24,"_303":24,"_304":-5,"_305":24,"_306":24,"_307":24,"_308":-5},"ENV",{"_260":261,"_262":263,"_264":16,"_265":266,"_267":268,"_269":270,"_271":272,"_273":274,"_275":276,"_277":278,"_279":280,"_281":282,"_283":284,"_285":286,"_287":288},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

History · enrichment-dashboard.html