Review uncertain AI candidates and preserve unresolved business scope

apps/docs/tests/research.test.ts

@@ -23,9 +23,9 @@ test("research revisions share the authentication database and preserve replay l
     for (let i = 0; i < 20; i++) consumeCode(config, "invalid");
     assert.equal(consumeCode(config, "invalid"), "limited");
     const initial = readResearchSnapshot();
-    assert.equal(initial.companies.length, 1051);
-    assert.equal(initial.evidence.length, 2697);
-    assert.equal(initial.observations.length, 366);
+    assert.equal(initial.companies.length, 1065);
+    assert.equal(initial.evidence.length, 2764);
+    assert.equal(initial.observations.length, 372);
     assert.deepEqual(initial.candidateBatches, seed.candidateBatches, "candidate gaps persist in the same database metadata");
     assert.deepEqual(initial.directoryRechecks, seed.directoryRechecks, "follow-up classifications persist without rewriting the discovery audit");
     assert.deepEqual(readResearchSnapshot(), initial, "reopening does not duplicate records");
@@ -64,8 +64,8 @@ test("research validation rejects broken evidence and unsafe source links", () =
 
 test("reviews link explicit companies without turning global or excluded cases into companies", () => {
   const snapshot = structuredClone(seed) as ResearchSnapshot;
-  assert.equal(snapshot.reviews.filter((r) => r.scope === "companies").length, 170);
-  assert.equal(snapshot.reviews.filter((r) => r.scope === "dataset").length, 12);
+  assert.equal(snapshot.reviews.filter((r) => r.scope === "companies").length, 173);
+  assert.equal(snapshot.reviews.filter((r) => r.scope === "dataset").length, 13);
   assert.equal(snapshot.reviews.filter((r) => r.scope === "excluded-candidate").length, 44);
   const review = snapshot.reviews.find((r) => r.id === "R4")!;
   assert.deepEqual(reviewCompanyIds(snapshot, review), ["AI007", "AI013", "AI021", "AI034"]);
@@ -85,11 +85,11 @@ test("reviews link explicit companies without turning global or excluded cases i
 test("disclosure counts exclude client income, biographies, targets and duplicate milestones", () => {
   const snapshot = seed as ResearchSnapshot;
   const summary = categorySummary(snapshot);
-  assert.equal(summary.reduce((n, c) => n + c.count, 0), 1051);
-  assert.equal(summary.reduce((n, c) => n + c.self, 0), 52);
+  assert.equal(summary.reduce((n, c) => n + c.count, 0), 1065);
+  assert.equal(summary.reduce((n, c) => n + c.self, 0), 53);
   assert.equal(summary.reduce((n, c) => n + c.estimate, 0), 2);
   assert.equal(summary.reduce((n, c) => n + c.platform, 0), 19);
-  assert.equal(summary.reduce((n, c) => n + c.unknown, 0), 978);
+  assert.equal(summary.reduce((n, c) => n + c.unknown, 0), 991);
   assert.equal(snapshot.companies.find((c) => c.name === "Rork")?.revenueStatus, "unknown");
   assert.equal(snapshot.companies.find((c) => c.name === "Result")?.revenueStatus, "unknown");
   assert.ok(snapshot.observations.filter((o) => o.selection === "signal").every((o) => o.scope === "公司整体"));
@@ -124,8 +124,8 @@ test("service run rates, legacy freight revenue and paid adoption do not become
 
 test("technical reviews keep accessible materials separate from tested capabilities", () => {
   const snapshot = structuredClone(seed) as ResearchSnapshot;
-  assert.equal(snapshot.companies.filter((c) => c.technicalReview).length, 83);
-  assert.equal(snapshot.companies.filter((c) => c.technicalReview?.level === "public-artifact").length, 18);
+  assert.equal(snapshot.companies.filter((c) => c.technicalReview).length, 85);
+  assert.equal(snapshot.companies.filter((c) => c.technicalReview?.level === "public-artifact").length, 19);
   assert.equal(snapshot.companies.filter((c) => c.technicalReview?.level === "independent-test").length, 1);
   assert.equal(snapshot.companies.find((c) => c.id === "AI051")?.revenueStatus, "unknown");
   assert.match(snapshot.companies.find((c) => c.id === "AI051")!.technicalReview!.testScope!, /Sep 2026/);
@@ -443,6 +443,55 @@ test("follow-up evidence can admit an uncertain directory candidate without rewr
   assert.throws(() => validateResearch(snapshot), /directory recheck/);
 });
 
+test("AI relevance follow-ups retain unresolved businesses and reuse existing company identities", () => {
+  const snapshot = structuredClone(seed) as ResearchSnapshot;
+  const batch = snapshot.candidateBatches!.find((b) => b.id === "YC-AI-UNCLEAR-1")!;
+  assert.equal(batch.complete, false);
+  assert.equal(batch.records.length, 24);
+  const original = coverageAudit.records.filter((r) => r.status === "ai-unclear").slice(0, 24);
+  assert.deepEqual(batch.records.map((r) => r.url), original.map((r) => r.source));
+  const linked = batch.records.filter((r) => r.status === "linked");
+  assert.equal(linked.length, 17);
+  assert.equal(linked.filter((r) => r.companyId!.startsWith("YC")).length, 14);
+  for (const [name, id] of [["Bud", "AI070"], ["Prism", "AI030"], ["Parametric", "AI014"]]) {
+    assert.equal(linked.find((r) => r.name === name)?.companyId, id);
+  }
+  assert.equal(snapshot.companies.filter((c) => c.profileOnly || c.directoryOrigin === "yc-2025-2026").length, 928);
+  for (const row of linked) {
+    const ycId = Number(row.id.replace("YC-AI1-", ""));
+    assert.equal(snapshot.directoryRechecks!.find((r) => r.ycId === ycId)?.companyId, row.companyId);
+  }
+  const pending = batch.records.filter((r) => r.status === "pending-ai");
+  assert.equal(pending.length, 7);
+  assert.ok(pending.every((r) => !r.companyId && r.evidenceIds!.length && /2025/.test(r.note)));
+  assert.match(pending.find((r) => r.name === "Better Auth")!.note, /AI.*模块/);
+  validateResearch(snapshot);
+  pending[0].companyId = "AI014";
+  assert.throws(() => validateResearch(snapshot), /research candidate/);
+  delete pending[0].companyId;
+  pending[0].evidenceIds = [];
+  assert.throws(() => validateResearch(snapshot), /research candidate/);
+});
+
+test("historical monthly revenue and customer examples do not enter recurring platform comparisons", () => {
+  const snapshot = seed as ResearchSnapshot;
+  const monthly = snapshot.observations.find((o) => o.id === "UNCLEAR1-O-SCALAR-MONTHLY")!;
+  assert.equal(monthly.value, 74000);
+  assert.equal(monthly.selection, "historical");
+  assert.equal(monthly.metric, undefined, "monthly revenue alone does not establish MRR or ARR");
+  assert.equal(monthly.comparisonGroup, null);
+  assert.equal(snapshot.companies.find((c) => c.id === "YC30514")!.revenueStatus, "self");
+  for (const id of ["AMBY-CLIENT", "NIXO-EXAMPLE", "AUTOSTEP-WASTE"]) {
+    assert.equal(snapshot.observations.find((o) => o.id === `UNCLEAR1-O-${id}`)!.selection, "excluded");
+  }
+  assert.ok(!platformIncomeSample(snapshot, { windowDays: null }).points.some((p) => p.observation.id.startsWith("UNCLEAR1-")));
+  const cua = snapshot.companies.find((c) => c.id === "YC30420")!;
+  assert.equal(cua.cat, "C03");
+  assert.equal(cua.revenueStatus, "unknown");
+  assert.match(cua.businessModel!, /自备计算机/);
+  assert.equal(cua.technicalReview!.level, "public-artifact");
+});
+
 test("Chinese discovery entries preserve unread articles and use separate primary evidence", () => {
   const snapshot = structuredClone(seed) as ResearchSnapshot;
   const batch = snapshot.candidateBatches!.find((b) => b.id === "JQ-2026-targeted-1")!;