Releases

Production release history. Each entry aggregates all customer-facing changes since the previous production release.

---

v1.9.4

Released: February 4, 2026 Previous production release: v1.4.1 (December 21, 2025)

Migration

The production endpoint has moved from api.wattdata.xyz to api.wattdata.ai. Authentication and tools are unchanged. Use versioned paths (/v1/mcp, /v1/mcp/m2m) for stability.

New

• build_cluster_expression — Describe your target audience in natural language and get a structured targeting expression for use with find_persons. No need to manually discover and combine cluster IDs.
• cluster_hash — Stable cluster identifiers that persist across data rebuilds. Supported in list_clusters, get_cluster, find_persons, and search_clusters. Use these instead of cluster_id for durable saved queries.
• API versioning — /v1/mcp and /v1/mcp/m2m endpoints pin your integration to the v1 API surface. Unversioned routes remain available but may shift to newer versions over time.
• Multi-identifier exports — analyze_customers accepts multiple email_columns, phone_columns, and address_columns. find_persons can export multiple identifier types (email, phone, address, name, maid) per row in a single request.
• Contact compliance metadata — get_person contact domains (email, phone, maid) now return objects with TCPA (do_not_call), CAN-SPAM (opted_in), and privacy (opted_out) fields alongside the contact value.

Breaking

• analyze_customers requires workflow_id — Obtain it from the generate_url_for_upload response before calling analyze_customers.
• generate_url_for_upload parameter renamed — Accepts filename instead of csv_key.
• analyze_customers output uses cluster_hash — Returns cluster_hash instead of cluster_id. Use cluster_hash in targeting expressions.
• Contact domains return objects — email, phone, and maid domains in get_person return objects with compliance metadata instead of string arrays. Update parsing code accordingly.
• Cluster fields renamed — cluster_name → name, cluster_domain → domain, cluster_value → value in search_clusters and analyze_customers responses.
• Address matching returns best match only — resolve_identities returns only the best-scoring person(s) per input address with a minimum quality threshold of 0.75. Maximum batch size is 1,000 addresses per request.
• Removed domains — engagement and geographic domains are no longer valid. Remove them from domain lists to avoid validation errors.

See Upgrading to v1 for detailed migration guidance on all breaking changes.

Improved

• Address resolution performance via H3-indexed lookups
• ICP synthesis reliability in analyze_customers
• Error messages now include specific validation details
• CSV handling with better error messages and more reliable schema detection
• General tool invocation reliability