Changelog

✨ July's release brings six new community connectors spanning business operations, AI infrastructure, product data, and scheduling. Odoo gets double coverage this month — a JSON-2 connector with fully dynamic schemas for any Odoo object (contacts, products, sales orders) plus support for object functions and guided wizards, and a JSON-RPC connector for teams on Odoo's classic external API, both letting you trigger recipes on record creation or updates. Pinecone opens the door to vector search and RAG workflows directly from Workato, letting recipes upsert, query, and manage embeddings at scale. Akeneo brings product information management into reach, with dynamic product search, channel- and locale-specific attribute updates, and asset downloads. Cal.com powers real-time scheduling automation — checking availability, booking, rescheduling, and reacting to booking changes, ideal for AI voice agents and sales workflows. Zscaler Digital Experience (ZDX) rounds things out with monitoring for application, device, and network performance.

🔧 Enhancements this month lean heavily into authentication flexibility, data integrity, and dynamic metadata. GraphQL adds configurable auth header keys and token type overrides across all OAuth flows, making it easier to work with nonstandard APIs like Shopify's. Veeva Vault's document upload gets a rework with cascading "Type → Subtype → Classification" fields and dynamically resolved metadata, alongside a fix for corrupted uploads. Microsoft Fabric SQL now runs dedicated, isolated connection pools per connection with smarter validation and automatic cleanup of unhealthy sessions, while Microsoft Fabric and Microsoft OneLake both add OAuth2 Client Credentials support alongside their existing Authorization Code flows. Jira Assets extends connection support to service accounts for Jira Cloud, and Apollo.io adds new polling triggers for accounts and contacts plus automatic rate-limit handling. The Positional File Tool introduces record position validation and per-column default values for flat-file parsing, and Aconex, Kickflow, and Ivanti round out the month with richer output fields, multi-row slip item support, and cached metadata for faster performance.

🛡 Stability improvements landed across Google Pub/Sub (fixing duplicate ACKs and permanent message loss on errors), Monday (missing method causing list/multiselect column failures), Filevine (multiple field rendering and control-type fixes so datapills and formulas work correctly), and Adaptive Insights (automatic token refresh on Workday JWT expiration). These fixes tighten reliability across data sync and workflow automation use cases.

The data_table_query is a powerful upgrade to data_table_lookup — bringing multi-record returns, advanced filter operators, record identity, and built-in correctness guards, all directly in formula mode.

• Record Identity — Every match includes record_id, created_at, and updated_at, so query results flow straight into downstream Update, Delete, or relation steps.
• Real Filter Operators — $gt, $gte, $lt, $lte, $ne, $in, $starts_with, and $and combinations replace equality-only matching.
• Multiple Records — mode: 'all' or mode: 'some' returns every match as an array up to a configurable limit, instead of silently returning just the first row.
• Built-In Correctness Guards — Six mode options make the formula throw an explicit error when match-count assumptions are wrong, instead of quietly returning the wrong row.
• Sorting — order ranks results by any field so you can grab the most recent or highest-value match with one_first_optional.

Learn more about data_table_query formula.

MCP Registry is a governed repository of every MCP server in your environment — what exists, who owns it, and who can use it. Builders, admins, and AI agents all work from the same single source of truth.

• Governed Discovery — A searchable UI gives builders and admins full visibility into every MCP server, its owner, and who has access — eliminating word-of-mouth sprawl and duplicate server creation.
• Programmatic Access — AI agents can discover available tools and verify authorization at runtime by calling the registry directly.

Learn more about MCP Registry.

MCP servers support multiple named, individually-auditable tokens — so many users can act through one server without sharing credentials. Each token is independently revocable, rotatable, and logged.

• Per-User Attribution — Named tokens make it clear which user performed which actions, replacing the ambiguity of a single shared credential.
• Independent Rotation — Individual tokens can be rotated without disrupting any other users or applications on the same server.

Learn more in MCP Access Methods.

Builders can label and annotate MCP tools so AI clients can distinguish routine actions from sensitive ones — keeping approval prompts meaningful and reducing the "approve everything" reflex.

• Editable Tool Titles — Relabel a tool's display name without breaking clients already invoking it by its programmatic name.
• Behavior Hints — Set destructiveHint, readonlyHint, idempotentHint, and openWorldHint on each tool so AI clients know how a tool behaves, not just what it's called.
• Automated Control — Read-only tools can be set to always allow; write and delete actions still require explicit approval.
• Reduced Approval Fatigue — Routine lookups execute without interruption, so approval prompts surface only when they actually matter.

Learn more about Tool Annotations.

This month brings a wide set of enhancements and improvements across Platform Connectors, expanding capabilities and accelerating workflows across recruiting, finance, document management, and collaboration tools alike. Learn more here.

IDP handles larger, more complex documents — long contracts, financial reports, and purchase orders with extensive line items — with sharper extraction accuracy across the board.

• Larger Document Support — PDFs up to 50 pages (previously 15) and files up to 50MB (previously 20MB) are now supported, eliminating the need to split and reconcile documents in pieces.
• Longer Tables — Tables up to 100 line items (previously 15) are extracted in a single pass, with more complete line item capture and more accurate values.
• Backwards Compatible — Existing recipes stay on the version they were built with; new recipes get the latest model by default. Upgrades are manual for existing recipes.

Learn how to migrate your existing recipe to a new version in Workato Docs.

OPA 32.2 is a focused release with SAP fixes, database improvements, and security updates.

• SAP — quality_of_protection config property now accepts both integer and string values; fixed an IDoc parser issue introduced in v32.1 where IDocs with multiple top-level segments failed to parse.
• Database — NULL and empty strings are now distinct in CSV export and import; TIMESTAMP and DATETIME columns now conform to ISO 8601 formatting.
• Security — Updated libraries per OWASP scanner recommendations.

Learn more in OPA 32.2 release notes.

Two significant enhancements enhance Workato API Gateway's alignment with OIDC standards.

• Public API Endpoints — API collections can expose unauthenticated endpoints with no access profile or credentials required, enabling standards-compliant discovery documents like /.well-known/openid-configuration (OIDC) and /.well-known/smart-configuration (SMART on FHIR). A clear warning is shown when configuring public endpoints to prevent accidental exposure.
• Custom Claims and Scopes Validation — Access Profiles can be resolved through OIDC token claims with no IdP modification required, enabling enterprise customers with locked-down shared IdPs — PingFederate, Okta, Azure AD, ADFS — to integrate end-to-end without involving their IdP admin teams.
• Clients Tab Renamed to Applications — The API Keys tab under Clients is now called Applications, reflecting support for multiple credential types beyond API keys alone.

Learn more in API Platform.

AIRO is the multi-agent system for the entire solution lifecycle. Describe what you want and AIRO plans, builds, fixes, and answers your questions — all from a single conversation, inside the Workato UI or via the AIRO MCP from Claude, Cursor, and other AI clients.

• Architect & Build from Intent — Plans and builds across recipes, Genies, Skills, APIs, Recipe Functions, and MCP servers from a single description.
  
• Ask Anything — Q&A grounded in Workato docs and best practices: how-to guidance, explanations of existing recipes, and the latest releases.
• Troubleshoot — Reads job logs, diagnoses failures, explains what went wrong, and fixes it — without leaving the conversation.
• Knows Your World — Grounded in your workspace assets and company knowledge so it applies best practices and reuses what you already have.
• Enterprise Guardrails — Respects RBAC controls out of the box, with no additional configuration required.

Learn more about AIRO.