diff --git a/internal/ai/orchestrator.go b/internal/ai/orchestrator.go
new file mode 100644
index 0000000..3eae24a
--- /dev/null
+++ b/internal/ai/orchestrator.go
@@ -0,0 +1,72 @@
+package ai
+
+import (
+	"context"
+	"log"
+
+	"git.georgsen.dk/hwlab/internal/inventory"
+)
+
+// Orchestrator manages the three-tier AI pipeline.
+// Tier1 is local oMLX (fast, low cost). Tier2 is OpenRouter (slower, better).
+// Tier3 (Lab Advisor) is out of scope for Phase 2.
+type Orchestrator struct {
+	tier1     AIClient
+	tier2     AIClient
+	threshold float64 // confidence threshold for escalation; default 0.75
+}
+
+// NewOrchestrator creates an Orchestrator. Both tier1 and tier2 must be non-nil.
+func NewOrchestrator(tier1, tier2 AIClient, threshold float64) *Orchestrator {
+	if threshold <= 0 {
+		threshold = 0.75
+	}
+	return &Orchestrator{tier1: tier1, tier2: tier2, threshold: threshold}
+}
+
+// Analyze runs tier1 and escalates to tier2 if confidence is below threshold.
+// Never returns an error from individual tier failures — tier errors cause escalation.
+// Returns a non-nil IntakeResult in all cases (may be zero-value on total failure).
+// The returned CatalogStatus is either StatusIndexed or StatusNeedsResearch.
+func (o *Orchestrator) Analyze(ctx context.Context, req IntakeRequest) (*IntakeResult, inventory.CatalogStatus, error) {
+	result, err := o.tier1.AnalyzePhotos(ctx, req)
+	if err != nil {
+		log.Printf("orchestrator: tier1 error (escalating to tier2): %v", err)
+		result = nil
+	}
+
+	// Escalate if tier1 result is missing, nil, or low confidence
+	if result == nil || result.Confidence < o.threshold {
+		log.Printf("orchestrator: tier1 confidence=%.2f < threshold=%.2f — escalating to tier2",
+			confidenceOf(result), o.threshold)
+		result2, err2 := o.tier2.AnalyzePhotos(ctx, req)
+		if err2 != nil {
+			log.Printf("orchestrator: tier2 error: %v", err2)
+		} else if result2 != nil {
+			result = result2
+		}
+	}
+
+	// Map confidence to CatalogStatus
+	if result == nil {
+		return &IntakeResult{
+			AINotes:        "all AI tiers failed",
+			Confidence:     0.0,
+			ConfidenceNote: "no result from any tier",
+		}, inventory.StatusNeedsResearch, nil
+	}
+
+	status := inventory.StatusIndexed
+	if result.Confidence < o.threshold {
+		status = inventory.StatusNeedsResearch
+	}
+	return result, status, nil
+}
+
+// confidenceOf returns 0.0 for nil results, otherwise result.Confidence.
+func confidenceOf(r *IntakeResult) float64 {
+	if r == nil {
+		return 0.0
+	}
+	return r.Confidence
+}
diff --git a/internal/ai/orchestrator_test.go b/internal/ai/orchestrator_test.go
new file mode 100644
index 0000000..690a93c
--- /dev/null
+++ b/internal/ai/orchestrator_test.go
@@ -0,0 +1,126 @@
+package ai
+
+import (
+	"context"
+	"errors"
+	"testing"
+
+	"git.georgsen.dk/hwlab/internal/inventory"
+)
+
+func testReq() IntakeRequest {
+	return IntakeRequest{PhotosBase64: []string{"data:image/jpeg;base64,/9j/"}, JobID: "test-job"}
+}
+
+// TestOrchestratorHighConfidence: tier1 returns confidence 0.95; tier2 never called; status == StatusIndexed
+func TestOrchestratorHighConfidence(t *testing.T) {
+	tier1 := &MockAIClient{FixedResult: HighConfidenceResult()}
+	tier2 := &MockAIClient{}
+
+	o := NewOrchestrator(tier1, tier2, 0.75)
+	result, status, err := o.Analyze(context.Background(), testReq())
+
+	if err != nil {
+		t.Fatalf("expected no error, got: %v", err)
+	}
+	if result == nil {
+		t.Fatal("expected non-nil result")
+	}
+	if status != inventory.StatusIndexed {
+		t.Errorf("expected StatusIndexed, got %q", status)
+	}
+	if len(tier2.Calls) != 0 {
+		t.Errorf("expected tier2 never called, got %d calls", len(tier2.Calls))
+	}
+	if len(tier1.Calls) != 1 {
+		t.Errorf("expected tier1 called once, got %d calls", len(tier1.Calls))
+	}
+}
+
+// TestOrchestratorLowConfidenceEscalates: tier1 returns confidence 0.40; tier2 called once; tier2 returns 0.85; status == StatusIndexed
+func TestOrchestratorLowConfidenceEscalates(t *testing.T) {
+	tier1 := &MockAIClient{FixedResult: LowConfidenceResult()}
+	tier2 := &MockAIClient{FixedResult: HighConfidenceResult()}
+
+	o := NewOrchestrator(tier1, tier2, 0.75)
+	result, status, err := o.Analyze(context.Background(), testReq())
+
+	if err != nil {
+		t.Fatalf("expected no error, got: %v", err)
+	}
+	if result == nil {
+		t.Fatal("expected non-nil result")
+	}
+	if status != inventory.StatusIndexed {
+		t.Errorf("expected StatusIndexed after tier2 escalation, got %q", status)
+	}
+	if len(tier2.Calls) != 1 {
+		t.Errorf("expected tier2 called once, got %d calls", len(tier2.Calls))
+	}
+	if result.Confidence < 0.75 {
+		t.Errorf("expected tier2 result confidence >= 0.75, got %.2f", result.Confidence)
+	}
+}
+
+// TestOrchestratorBothTiersFail: both tiers error; result is non-nil, status == StatusNeedsResearch, err == nil
+func TestOrchestratorBothTiersFail(t *testing.T) {
+	tier1 := &MockAIClient{FixedError: errors.New("tier1 connection refused")}
+	tier2 := &MockAIClient{FixedError: errors.New("tier2 connection refused")}
+
+	o := NewOrchestrator(tier1, tier2, 0.75)
+	result, status, err := o.Analyze(context.Background(), testReq())
+
+	if err != nil {
+		t.Fatalf("orchestrator must not propagate tier errors, got: %v", err)
+	}
+	if result == nil {
+		t.Fatal("orchestrator must return non-nil result even on total failure")
+	}
+	if status != inventory.StatusNeedsResearch {
+		t.Errorf("expected StatusNeedsResearch on total failure, got %q", status)
+	}
+}
+
+// TestOrchestratorTier1NilResult: tier1 returns nil result with nil error; orchestrator escalates to tier2
+func TestOrchestratorTier1NilResult(t *testing.T) {
+	tier1 := &MockAIClient{FixedResult: nil, FixedError: nil}
+	tier2 := &MockAIClient{FixedResult: HighConfidenceResult()}
+
+	o := NewOrchestrator(tier1, tier2, 0.75)
+	result, status, err := o.Analyze(context.Background(), testReq())
+
+	if err != nil {
+		t.Fatalf("expected no error, got: %v", err)
+	}
+	if result == nil {
+		t.Fatal("expected non-nil result from tier2 escalation")
+	}
+	if len(tier2.Calls) != 1 {
+		t.Errorf("expected tier2 called once after tier1 nil result, got %d calls", len(tier2.Calls))
+	}
+	if status != inventory.StatusIndexed {
+		t.Errorf("expected StatusIndexed from tier2, got %q", status)
+	}
+}
+
+// TestOrchestratorNeedsResearch: tier1 returns 0.40; tier2 also returns 0.40; final status == StatusNeedsResearch
+func TestOrchestratorNeedsResearch(t *testing.T) {
+	tier1 := &MockAIClient{FixedResult: LowConfidenceResult()}
+	tier2 := &MockAIClient{FixedResult: LowConfidenceResult()}
+
+	o := NewOrchestrator(tier1, tier2, 0.75)
+	result, status, err := o.Analyze(context.Background(), testReq())
+
+	if err != nil {
+		t.Fatalf("expected no error, got: %v", err)
+	}
+	if result == nil {
+		t.Fatal("expected non-nil result")
+	}
+	if status != inventory.StatusNeedsResearch {
+		t.Errorf("expected StatusNeedsResearch when both tiers return low confidence, got %q", status)
+	}
+	if len(tier2.Calls) != 1 {
+		t.Errorf("expected tier2 called once, got %d calls", len(tier2.Calls))
+	}
+}
diff --git a/internal/ai/research.go b/internal/ai/research.go
new file mode 100644
index 0000000..13d1b66
--- /dev/null
+++ b/internal/ai/research.go
@@ -0,0 +1,24 @@
+package ai
+
+import "context"
+
+// SearchResult is a single result from a SearXNG research query.
+type SearchResult struct {
+	Title   string
+	URL     string
+	Snippet string
+}
+
+// ResearchClient abstracts the SearXNG search backend.
+// Phase 7 will provide a real implementation.
+type ResearchClient interface {
+	Search(ctx context.Context, query string) ([]SearchResult, error)
+}
+
+// NoOpResearchClient is a Phase 2 stub that returns empty results.
+// Replace with SearXNG HTTP client in Phase 7.
+type NoOpResearchClient struct{}
+
+func (n *NoOpResearchClient) Search(_ context.Context, _ string) ([]SearchResult, error) {
+	return nil, nil
+}