The CData Arc Q4 2026 release puts AI to work inside your integrations. Ask Arc from your AI client to rerun a connector or requeue a failed message, get a plain-language explanation of a failed transaction in one click, and let Arc pinpoint what’s slowing down your worker pool.
This release expands Arc’s built-in Model Context Protocol (MCP) server with tools that take action, adds AI-assisted diagnosis for failed transactions and worker pool performance, and introduces new connectors, including Partner Mailbox for browser-based file exchange, Microsoft 365 Email built on Microsoft Graph, and Finalize, which confirms delivery before Arc cleans up source files. It also adds batch group processing, external secret providers for Azure Key Vault and environment variables, per-workspace Git repositories, and a broad set of improvements to automation visibility, the Admin API, and connectors.
More operational tools in the MCP server
Arc’s built-in MCP server launched in Q3 release with read-only diagnostic tools. Now it can act on the running system too, without touching your configuration. From your MCP-compatible client, you can ask Arc in plain language to run a connector, requeue a failed message, or clean up old files. You stay in control: every action requires explicit confirmation, works within your existing Arc permissions, and is recorded in the audit log.
Diagnostics get sharper too. Your client can now pull service-level logs, such as SFTP Server connection failures that happen before a transaction even exists. Arc also ties the application log, request log, and transaction records together for each message, so when you ask why a message failed, you get the full story in one answer.
AI diagnosis for failed transactions and worker pools
When a transaction fails, Arc shows the error and leaves the investigation to you. A new Diagnose error link appears on failed transactions in Activity, Message Detail, and a connector’s Transactions tab. Click it and Arc asks your configured AI provider to analyze the failure, then returns an explanation with suggested corrective actions. It runs only when you click, it doesn’t change anything, and a copy button lets you paste the result into a support case.
Tuning performance shouldn’t mean digging through diagnostic logs by hand. Analyze Worker Pool appears in Performance settings on the Advanced tab once you’ve configured an AI provider. An admin starts a logging session, and Arc collects worker pool logs from that point forward for the window you set. Collection can’t capture the past, so you start it before the slowdown you want to study. When collection finishes, Arc analyzes the session with your AI provider and gives you findings and recommendations. The recommendations are only suggestions, so Arc never changes a setting on its own, and past analyses are saved so you can come back to them. Learn more about Arc AI.
Partner Mailbox for browser-based file exchange
Many hosts support trading partners who can’t run an SFTP client or an automated endpoint, and who still need to send and receive files.
The new Partner Mailbox connector gives a partner a browser-based mailbox with its own credentials, so the partner doesn’t need an Arc login. Each partner gets its own connector, login, and web address.
Partners see two views. Files for you lists what your flow has staged, with a download button on each file. Your submissions is where partners upload files by drag and drop or a file picker and see what they’ve sent.
You define the folder structure as two separate trees of subfolders, one for downloads and one for uploads, and you edit them by drag and drop. Arc tells your flow which folder each upload landed in, so you can route by folder. That matters for flat-file and other non-EDI data, where Arc can’t infer the document type from the content. Partner uploads enter your flow like any other message, so there’s nothing special to build downstream.
Files are removed after a retention period you set, seven days by default. Setting retention to zero, or choosing delete on download, removes a file as soon as the partner downloads it.
Microsoft 365 Email through Microsoft Graph
Shared and departmental mailboxes are awkward over SMTP and IMAP, which expect per-user OAuth consent. Outlook categories are invisible to those protocols altogether.
The new Microsoft 365 Email connector sends and receives mail through Microsoft Graph. It supports Azure AD sign-in and app-only authentication with a client secret or a certificate. With app-only authentication, one connector can reach any mailbox in the tenant, such as support@ or billing@, without per-user consent. On receive, you can filter on the server by folder, read status, importance, attachments, and category. After processing, Arc can mark a message as read, tag it with an Outlook category, or move it to a folder, which gives you a way to mark mail as handled without deleting it. You can also send on behalf of or as another user.
The existing Email Send and Email Receive connectors are unchanged. They remain the right choice for Gmail, on-premises Exchange, and other SMTP and IMAP servers.
Batch group processing and deferred finalization
Processing a folder of files as separate messages makes one question hard to answer: did every file arrive? It matters when source files should be deleted, a trigger file written, and one notification sent only after all of them are delivered.
The File and Amazon S3 connectors now accept batch groups as input on send. Arc sends each file in the batch through its normal path, and if one fails, the whole batch is marked failed. Arc remembers which files already went out, so a retry picks up where it left off.
Scripts get batch group support too. Your scripts and Email Send templates can work through every file in a batch, so you can send one notification per batch instead of counting messages. The Script connector can also build new batches on output.
For source cleanup, the File connector has a new Defer received-file finalization setting. With it on, Arc holds each received file as pending until the flow finishes. A new Finalize connector at the end of the flow tells the File connector that delivery succeeded, and the File connector then runs its configured cleanup on the source file. If the flow times out or fails, the source file stays put and is picked up again on the next receive, so nothing gets lost.
External secret providers
Many security teams require that credentials live in one governed system. Until now, Arc’s vault stored secrets locally, so teams that manage credentials in Azure Key Vault or inject them as environment variables had to copy each secret into Arc and keep the copies in sync.
Arc can now read secrets from external providers. Arc resolves a secret from the provider when a job needs it, so connectors use the current value, and anything added to the external vault is available in Arc without an import step. External providers are read-only from Arc, so you add and edit items only in the local vault. Connector settings still reference secrets through the vault picker, which means connectors don’t need to know where a secret lives.
Two providers are available. The Azure Key Vault provider supports more than one vault, and Save and Test confirms the connection right away. The Environment Variable provider uses wildcard patterns to choose which variables Arc picks up and whether each one is treated as encrypted.
Every item from every provider appears in the vault table with a Provider column. Provider settings export and import with the rest of your configuration, and credentials stay protected. Vault references now also work in workspace settings and on the global Settings page.
The Azure Key Vault provider requires an Enterprise license. The Environment Variable provider is available in every edition.
Per-workspace Git repositories
Git-based configuration management arrived in v26, and Q3 added local repositories. In this release, a workspace can have its own repository, with its own remote URL, branch, and credentials, set in the workspace’s Version Control settings. The instance repository stays the default for everything else.
A repository list in the version control footer shows which repository your next action targets. Commit, push, pull, and discard act on one repository at a time. On the Flows page the list follows the workspace you’re in, and you can click a repository to switch. Commit History can be filtered by repository.
The release also tightens up daily Git work. You can switch an existing repository from local to remote in the UI without editing Git config by hand. A failed push in Commit and Push no longer deletes or reverts your changes on disk. The diff view can show a connector with all of its settings, and Arc now saves its configuration files in an indented format, so diffs are much easier to read.
Visibility into automation and the application database
It’s frustrating not to know whether a scheduled connector ran and found nothing, or never ran at all. Each connector’s Automation tab now shows a live Automation Status card with the result of the last run and when the next one is due, and a new Automation Logs view under Activity records every scheduling decision Arc makes. The Dashboard adds a card for connectors in backoff and an auto-refresh option, so problems surface without anyone going looking for them.
Arc also keeps a closer eye on its own health. A banner now warns you when the application database grows past the tested limit for your database type, and SLA At Risk and Violated events are written to the application log, so your existing monitoring tools can pick them up without relying on email.
Admin API, security, and administration
For teams that manage Arc through scripts and pipelines, the Admin API now handles user management, including regenerating API tokens and client secrets, and can move a group of connectors to another workspace without breaking the connections between them.
Audit logs now capture exactly which setting changed, with its previous and new values, across connectors, workspaces, users, roles, and more. If your team answers to change management or internal audit, that’s a much clearer record of who changed what, and when. A new Message Directory setting also lets you store message data separately from the rest of Arc’s application data.
Connectors, mapping, and file transfer
Mapping gets smarter in this release. Database connectors can use customized prompts for AI-assisted mapping, the Lookup action can pick out a value by qualifier in EDI data, and XML Map conditions can compare one XPath, header, or variable against another.
For trading partner work, the AS2 connector now signs the MDN with your rollover certificate when that’s the certificate that decrypted the message, so a certificate rollover no longer means extra setup on every connector. The Trading Partner Management Console also turns on functional acknowledgement routing for you when you add a 997 or 999.
Application and database connectors can now use certificates managed in Arc, shared connections show their status right in the connection picker, and the Webhook connector can promote incoming HTTP headers to message headers. Beyond that, you’ll find dozens of smaller improvements and fixes across SFTP, Box, Azure Blob, SOAP, Excel, and the Arc UI. The full list is in the release notes.
Upgrade notes
The classic Activity page, the classic Reports UI, and the classic flow designer are removed. Version Control is now limited to administrators, and non-admin users no longer see its options. and Arc 26.4 uses the v26 drivers, which change several defaults. The release notes have the full list.
Get started with the Q4 2026 release
On-premises customers can download the latest version from our builds page. Arc Cloud customers should contact support to schedule their upgrade.
For full details on everything in this release, visit our documentation or reach out to our support team.
