August’s biggest story is Agent Studio maturity: a single Genie can now serve multiple chat interfaces at once, run for up to 30 minutes, capture end-user feedback natively, and be driven headlessly through an API. On the platform side, Data Pipelines added a wave of enterprise controls and a much bigger source/destination list, and IDP’s new model now handles documents up to 50MB and 50 pages.
Agent Studio
Multiple Clients per Genie
Genies now connect to multiple chat interfaces at the same time – Slack, Microsoft Teams, and custom interfaces, all off one Genie. Chat interfaces are now managed from a new Trigger module, so every connection point lives in one centralized view.
Why this matters: Employees work across Slack, Teams, and Workato GO depending on the moment. A single Genie can now meet them everywhere at once, so the answer someone gets in Slack matches the answer they’d get in Teams, without a builder having to maintain parallel copies of the same agent.
What you can do:
- Connect a single Genie to multiple Slack workspaces, multiple Teams tenants, and custom interfaces all at the same time
- Manage all connections and triggers from one central Triggers module, with chat history kept isolated per interface
End-User Feedback on Genie Responses
Every Genie response in Slack and Microsoft Teams now carries a native thumbs-up/thumbs-down control, with API support for custom interfaces. End users can add a short comment explaining their reaction, too. Feedback is on by default across the platform – no setup or configuration required.
Why this matters: Agent quality determines adoption. Without a native way to capture how end users experience each response, quality issues stay invisible and improvements come from anecdotes instead of data. Feedback at every response turns quality into a measurable signal.
What you can do:
- Capture feedback natively in Slack and Teams, with the same experience available in custom interfaces via API
- See a positive/negative split and trend on the home page feedback widget
- Drill into any flagged conversation from conversation history to see the exact message and trace behind it
- Include feedback data in conversation exports, the Dev API, and log streaming
Increased Genie Timeout Limits
We’ve increased the timeout limits on how long Genies can run.
Why this matters: LLMs are being asked to do longer, more ambitious work including deep research, large document generation, multi-step tool chains, code execution. That work needs infrastructure that scales with it. The extended runtime is on by default, and end users will see agent messages, tool calls, and search calls streamed the whole way through.
What you can do:
- Run Genies for deep research, long-form document generation, and multi-step tool chains without hitting timeouts
- Watch tool call progress live in Slack, Teams, Workato GO, and the Headless API – every message and tool call streams in the order it happens
Test a Skill Without the Genie
Builders can now run a skill directly from the skill editor. Hit Test, run it in one window. No second tab, switching to Genie test mode, or prompt-engineering a Genie into calling the skill.
Why this matters: Fast iteration is the difference between shipping a working skill and abandoning something halfway. Testing skills directly – without needing a Genie in the loop – cuts the tightest debug loop from minutes to seconds and lets builders exercise a skill’s actual logic before wiring it up to an agent.
What you can do:
- Run a skill directly from the skill editor
- Exercise a skill’s logic, mapping, and error handling before wiring it to a Genie
- Debug a failed skill mid-run without restarting the whole conversation
MCP Gateway
New Pre-Built MCP Servers – Finance, Productivity, IT, and Sales Intelligence
We released seven new pre-built MCP servers this month, extending AI agent reach into billing, personal task management, IT service operations, and sales intelligence. These servers are accessible from the AI Hub in your Workato workspace under MCP Servers.
Why this matters: Pre-built MCP servers reduce the time to connect AI agents to enterprise systems from weeks to minutes. Each server is governed through Workato’s Verified User Access, role-based access control, and audit trails, so every action an agent takes is secure, traceable, and attributed to the person who authorized it.
Finance
- Xero Billing and AR: Run the full accounts-receivable workflow. Create and approve invoices, record payments, issue credit notes, set up recurring billing templates, and manage customer contacts. A live invoice always requires explicit confirmation.
Productivity
- Google Tasks: Capture tasks and subtasks, set due dates, organize across task lists, and mark work complete or reopen it – all through conversation.
- Microsoft To Do: Create tasks with reminders, recurrence, and importance levels. Or, break work into checklist steps, search across lists, and pull in tasks flagged from Outlook email.
IT Service Management
- ServiceNow ITSM: Log incidents from alerts or messages, search and triage the queue by state or assignment group, update work notes and priority, and resolve with proper resolution codes. Includes knowledge base lookup and user/group discovery for routing.
Sales Intelligence
- Clay Contact & Company Explorer: Research accounts and map stakeholders: pull firmographics and funding data, find decision-makers by title or seniority, and discover prospects filtered by industry, tech stack, or size.
- Waterfall Lead Enrichment: Enrich a lead against ZoomInfo, Apollo.io, and LeadIQ in sequence, merge results with field-level precedence rules, and return a confidence-scored profile. Supports single-provider lookups or the full waterfall.
Data Orchestration
Event Streams – Cross-Workspace Sharing
Event topics can now be shared across workspaces within the same Automation HQ. A topic created in one workspace can be granted to others for publishing, subscribing, or both – so recipes across multiple workspaces work off the same live topic in real time.
Why this matters: Cross-workspace sharing lets a topic created in one workspace power recipes in others, so hub-and-spoke architectures and multi-BU rollouts work end to end, without duplicating infrastructure or giving up workspace isolation.
What you can do:
- Turn on cross-workspace sharing at the Automation HQ level with a single admin switch
- Grant topics from the workspace admin console with per-topic Publish and Subscribe permissions
- Have multiple workspaces publish to and consume from one shared topic at the same time
- See which topics were granted vs. owned, with the granting workspace visible in the recipe topic dropdown
- Access the shared topic’s schema and full message history from any granted workspace
Data Pipelines – Expanded Connectivity and Enterprise Controls
Data Pipelines added a wave of new source and destination connectors, plus a set of enterprise controls: selective per-object re-sync, field-level data masking, 5-minute sync frequency, cron scheduling, and pipeline triggers that fire downstream recipes when a sync completes.
Why this matters: Data Pipelines went GA earlier this year with multi-object ingestion into Snowflake, Databricks, BigQuery, and Microsoft SQL Server. August closes the gaps enterprises hit at scale while broadening the source and destination list to more than 50 systems.
What you can do:
- Ingest from enterprise data sources including Salesforce, NetSuite, Amazon S3, Azure Blob, Google Cloud Storage, HubSpot, Oracle EBS, Oracle Cloud Fusion, Oracle Database, ServiceNow, Workday, SAP, Jira, Stripe, and Shopify, into Snowflake, Databricks, BigQuery, or Microsoft SQL Server (On-prem)
- Selectively re-sync individual objects without rebuilding the whole pipeline
- Apply field-level data masking for regulated data
- Sync as frequently as every 5 minutes, on cron schedules, or trigger downstream recipes from pipeline events
- Monitor pipelines at the object level with parallel execution for throughput
Platform
data_table_query Formula Function
The Workato Data Tables formula language now includes a data_table_query function that extends data_table_lookup with advanced filters, multi-record matches, and exposed metafields.
Why this matters: Complex data lookups shouldn’t require complex recipes. The data table query function handles filtering, matching, and metadata reads in a single formula call – so lookup logic stays where a builder can read it in one line, and recipes stay focused on the work they’re doing.
What you can do:
- Query Data Tables with advanced filter expressions across multiple fields
- Return multiple matching records that power downstream recipe actions in a single call
- Read record_id, created_at, and updated_at directly in a formula
Intelligent Document Processing
IDP by Workato – More Powerful Document Processing
IDP’s document extraction now runs on a new model that handles documents up to 50MB and 50 pages with higher accuracy, and up to 100 line items in a table.
Why this matters: Enterprise documents rarely fit neatly. Contracts, financial statements, and long-form forms run into hundreds of pages. Workato IDP now processes documents up to 50MB and 50 pages in a single pass with higher accuracy, turning full documents into structured data you can use in your workflows. And, we’ve also made improvements to complex tables, now supporting up to 100 items in a table.
What you can do:
- Improve extraction accuracy on long-form documents, tables, and dense forms without model tuning
- Extract from documents up to 50MB and 50 pages in a single pass
- Extract up to 100 line items from a table
See the latest IDP versions here.
Security & Operations
On-Prem Agent 32.2
On-Prem Agent 32.2 is a targeted release with SAP connectivity fixes, CSV formatting improvements, and library updates.
Why this matters: SAP and on-prem databases are the systems of record for many enterprise workflows. This release keeps SAP IDoc processing more stable across IDoc structures and makes CSV round-trips lossless across Postgres, Redshift, and MySQL – clean data end to end.
What you can do:
- Use SAP connection profiles with quality_of_protection set as either an integer or a string
- Rely on the SAP IDoc parser for IDocs with multiple top-level segments (32.1 regression fixed)
- Distinguish SQL NULL from empty string in CSV export and import – NULL as an unquoted empty value, empty string as a quoted empty value
- Export TIMESTAMP WITHOUT TIME ZONE (Postgres, Redshift) and DATETIME (MySQL) columns in ISO 8601 format
- Pick up the latest OWASP-recommended library updates
See the OPA 32.2 release notes.
Full AI Stack for Israel and Korea Data Centers
Enterprise MCP, Agent Studio, IDP, AI by Workato, and Go services are now deployed to Workato’s Israeli and Korean data centers, bringing the full Workato One experience to these regions.
Why this matters: Deploying the full Workato One stack in-country means agents, document processing, and MCP traffic all stay within the region – meeting residency requirements without excluding AI services.
What you can do:
- Run Enterprise MCP, Agent Studio, IDP, and AI by Workato entirely within the data center
- Meet data residency and sovereignty requirements for the region
Connectivity
For the full list of new connectors and enhancements this month, check out the community connector and platform connector blogs in the Product Hub.