August 2026

Three and a half minutes on fifteen of this month's changes — invites and free viewers, the guided first run, native ERP steps, Essbase, external lakehouses, workflow containers and resume, SCD-2, allocation dimensions, geospatial, dashboards and Dash apps, HTML report emails, and project copy — and what each one saves you.

Releases Shipped

Version	Released
v5.11.12	2026-08-05
v5.11.13	2026-08-05
v5.11.14	2026-08-07
v5.11.15	2026-08-10
v5.11.16	2026-08-12
v5.11.17	2026-08-14
v5.11.18	2026-08-14
v5.11.19	2026-08-28

Added

• A guided first run for new workspaces. Empty screens now explain their purpose and offer the appropriate next action. You can now choose from several guided sample projects — retail sales analysis, marketing campaign performance, supply-chain inventory status, and finance budget-versus-actual reporting — and load and run one in a single step to watch its progress as it imports, cleans, and summarizes the data. Once the run finishes, PlaidCloud automatically publishes the result to reporting and builds a dashboard from it — a KPI summary, a trend over time, a category breakdown, and (for the retail and inventory samples) a breakdown by US state — so you land on a finished dashboard, not just a table. The screen sets expectations up front: the first run warms your warehouse and takes about a minute, while later runs are near-instant. A Need help? link is always within reach if you get stuck, and you can dismiss the guidance when you no longer need it and restart it later from the user menu. See Start With the Sample Project.

• One-click password setup, and cleaner account emails. A new member’s welcome email now includes a single Set your password link — there’s no separate step to first request a one-time code. The link is secure, works once, and expires after 24 hours. Welcome emails are delivered more reliably, too: an undelivered one is automatically retried, so a new member is never left unable to sign in. Account and notification emails have also dropped their marketing opt-out footer, matching their transactional purpose. See New Member Welcome Email.

• Invite teammates as free, unlimited viewers. Adding a member is now an invite: pick a role — Viewer or Builder — and enter one or more email addresses to invite everyone at once. Viewers get free, unlimited read-only access on every plan, so there’s no reason not to bring your whole team in to look around; Builders can create and edit content and count toward your plan’s build-seat limit. A seat-usage note updates as you switch roles so you always know what an invite will cost. See Inviting New Members.

• External lakehouses for Snowflake and Databricks. Projects can now use your Snowflake or Databricks lakehouse as their data backend, keeping data under your own governance and billing while PlaidCloud runs workflows and dashboards. File imports use each warehouse’s native bulk-load path, and dashboards support both engines. See Use an External Lakehouse.

• A native connection and step catalog for major ERPs. PlaidCloud now connects natively to NetSuite, Acumatica, Business Central, Dynamics 365 Finance & Operations, Oracle Fusion, Workday Financials, Sage Intacct, and SAP S/4HANA Cloud. Ready-made import steps pull financial and operational data — balances, GL detail, master data, and transactional records — with no SuiteQL, OData, or custom REST calls to write, and rich server-side filters, including date, account, and dimension ranges, so a pull can be scoped without downloading everything. Matching post steps write journal entries, invoices, bills, and payments back, with a preview/validate mode that checks every document before anything posts, an idempotent write path that guards against double-posting a retried or re-run batch, and — for corrections — a posted reversing entry rather than an in-place edit or delete. On a NetSuite account running Multi-Book Accounting, pulls now total the primary book by default rather than silently summing every book together, and a wide date-bounded pull is automatically chunked to stay under NetSuite’s per-query row ceiling, with a clear error asking for a date range if an unbounded pull would otherwise be cut short. Sage Intacct imports can include statistical accounts and now cover GL detail, balances, entries, and dimension masters with automatic pagination; its Post Journal Entry step supports previewing and posting journal entries from workflow tables; and a companion step attaches supporting documents from a table onto Intacct records. SAP S/4HANA Cloud Public Edition connects directly over OAuth2 and OData/SOAP, with no PlaidLink agent required, distinct from the existing agent-based SAP ECC / S/4HANA connector for on-premises systems. ERP Post History provides a read-only record of posting attempts and their current status across all of these ERPs plus SAP. See Import NetSuite Financials, Connect to Acumatica, Connect to Business Central, Connect to Dynamics 365 Finance & Operations, Connect to Oracle Fusion, Connect to Workday Financials, Connect to Sage Intacct, Connect to SAP S/4HANA Cloud, and Review ERP Post History.

• Safer connection setup. Connection editors can mark a connection as sandbox or production, and database connections can be tested before use. See Create and Manage a Connection.

• Hosted Plotly Dash apps. Deploy Dash apps from PlaidCloud Git with the same build, autoscaling, single sign-on, per-user data access, logs, and rebuild workflow as server-side Panel apps. See Deploy a Dash App From PlaidCloud Git.

• Advanced workflow controls. Execution containers run a related group of steps as a unit, with container-level conditions, iteration, status, and targeted reruns. The canvas now makes conditional steps visible and editable, supports copying and pasting steps with their links and groups, and lets a failed workflow resume from the failing step rather than rerunning completed work. See Execution Containers, Seeing Conditions on the Visual Canvas, Copy & Paste steps, and Run a Workflow.

• More flexible projects, agents, and access controls. Copy selected project content into an existing project and re-sync it later; assign PlaidLink agents to environments so a project environment selects the matching agent; and apply security-group row access to dashboards. See Copying a Project, Agent Environments, and Managing Row Access.

• Published Table Performance recommendations. Table Optimizer can now show passive, evidence-based recommendations for published-table layout improvements. This first phase never changes a table automatically; your team remains in control of every change. See Review Published Table Performance.

• Clearer workspace provisioning. Workspace lists and details now report provisioning state and show service links only when the corresponding service is ready. See Provisioning Status.

• Geospatial services and workflow tools. Use PlaidCloud’s geocoding API for address lookup, reverse geocoding, autocomplete, and place lookup; optionally use drive-time routing; and enrich workflow results with geohashes or Plus Codes. Make Grid and Poly-Split now have designer forms. See Geocoding API, Drive-Time Routing API, and Geospatial Analytics.

• A guided allocation-dimension builder. Create rule-based allocation dimensions from the Dimensions view without manually assembling the underlying hierarchy. See Creating an Allocation Dimension.

• Point-in-time master data. Two new workflow steps let you keep and use the history of your master data instead of only its current state. The SCD-2 Master Builder turns a source table into a history table that records every value each key has held, stamped with the date range it was in effect; the SCD-2 As-Of Join then attaches to each fact row the version that was in effect on that row’s own date — so a transaction is attributed to the region, cost center, or price that applied when it happened, not the one in effect today. See SCD-2 Master Builder and SCD-2 As-Of Join.

• Live queries against Oracle Essbase cubes. Connect to Oracle Essbase 21c or Oracle Analytics Cloud with a service account, then run an MDX query against a cube from a workflow and land the result directly in a table. It’s a true live query — each run reflects the cube’s current numbers, with no extract job to wait on. Build the query by browsing the cube’s dimensions and picking members, or write MDX directly, and the result’s columns are derived from the returned grid so you don’t declare them up front. Row access applies to the landed table like any other. See Essbase Query and the Oracle Essbase Connector.

• Rich HTML data-report emails. The Notify Via Email step now composes its message as a rich HTML report — titles, summaries, and rendered tables — from a template over one or more of your workflow’s tables. Add each table as a named frame (with its own column selection, filter, and aggregation) and lay them out in a single template editor with a live preview, a Send Test to Me button, a Prettify button, and an insert palette for tables, the workspace name, loops, and number and date formatting. There’s no separate plain-text body to keep in sync, and existing steps carry over automatically with their old body converted for you. See Send Templated HTML Email Notifications.

Changed

• Cleaner REST and workspace APIs. The unused all workspaces option has been removed from project, agent, and warehouse listings. REST callers may continue sending it while they update; JSON-RPC callers should remove it. The workspace numeric-ID request now returns the workspace ID rather than the caller’s user ID.

• Faster, clearer reports and exports. PDF project reports render with less layout overhead, gather data concurrently, and may use fewer pages. Excel exports stream large files with lower memory use, preserve date formatting, and identify Excel size limits before the export starts. Report text now resolves column references, and report/table reads surface their actual error instead of silently producing empty content. See Export to Excel and Report Text.

• A more usable workflow designer. The Visual Workflow Designer adds a foldered, accessible step palette; collapsible and resizable palette and Inspector panes; and a dark canvas that follows the application theme. Read-only users now see steps as read-only rather than discovering a permission problem at save time. See Advanced Workflows and The Step Palette.

• More current and better-governed dashboards. Republishing a table immediately invalidates its dashboard cache, while the default cache lifetime is now one hour. Dashboard visibility respects restrictions set in the dashboard service, and dashboard accounts follow PlaidCloud membership changes more reliably. See Dashboard Caching and Which Dashboards You See.

• More efficient connected AI agents. MCP results now support smaller pages, selected fields, column-oriented results, explicit truncation, bounded query and batch sizes, and a leaner corrected tool catalog. These changes reduce agent context use and prevent a single request from monopolizing shared capacity. See Keeping Results Small and Batching Requests.

• Clearer allocation and cost-trace answers. Allocation analysis now states the amounts, periods, scope, and measurement behind its findings; distinguishes complete results from top movers; explains confidence and limits in plain language; produces stable ordering; and suggests next steps that match what it could and could not determine. What-if answers also better identify stale data, rounding limits, and the scope of their estimate. The calculations have not changed; the explanations are more transparent and consistent. See Analyzing Allocations with AI, Honest Answers, and Tracing Allocations.

• Better performance under load. Workflow runs request less repeated information, large table and dimension operations resolve dependencies in batches, idle Lakehouse warehouses scale query capacity down and back up as demand returns, and long-running external authentication or import work no longer needlessly occupies shared capacity. See Lakehouse Management.

• Refined integrations and organization. The Microsoft 365 Copilot agent is now named PlaidCloud and uses PlaidCloud branding at sign-in. Sage Intacct workflow steps are grouped under External Systems. See Microsoft 365 Copilot and Sage Intacct Steps.

Fixed

• Member lifecycle and workspace administration. Deleting a member now removes that member, rather than the administrator, from distribution lists and revokes the member’s connection and document-account access. Disabled members can be re-enabled, and the dimension-member picker reliably opens and applies changes to the intended rows. Review distribution lists, connections, and document accounts for members deleted before this release. See Member Management and Dimension-Type Properties and the Member Picker.

• Workflow, connection, and project reliability. REST requests and Slack or Teams notification steps can again retrieve their connection credentials. Deleting a workflow also removes its scheduled events; connection Where Used includes REST v2 steps; Macro Concurrent detects a stopped child instead of waiting indefinitely; large ML models can be saved; and step configuration and interruption states report accurately. An allocation or rule-based tagging step whose assignment dimension has no active rules — for example, when every rule is turned off by the include property — now stops with a clear message pointing you to the assignment dimension instead of failing with an internal error. See REST Request, Notify via Slack, and When a Child Invocation Stops Responding.

• The Postman request step no longer truncates results. A request that fails partway through a paginated response now leaves the target table unchanged instead of partially loaded, and very large paginated responses import in full rather than only their final portion. This legacy step is superseded by the REST Request step. See REST Request.

• More reliable data import, export, and project transfer. Excel and Lakehouse timestamp imports preserve values; Access imports fail clearly when a source table cannot be read; table anti-joins preserve text columns; and archive imports and exports accurately report outcomes, preserve distinguishable duplicate copies, and keep the interface responsive. Alteryx imports now close on success and render ordinary comments without warning styling. See Import Access and Import a Workflow.

• Connected and built-in AI agents complete more intended work. Agents can create documented text, document, spatial, and report steps; save run conditions and related workflow settings; render requested diagrams as images without inventing values; send email and SMS; upload documents; and accurately detect changed records and warehouse search results. Text Sentiment sentence-level scoring is now documented. See Using the AI Assistant and Text Sentiment.

• Small but important interface corrections. Downloads now name their failure, document-folder Launcher pins work on S3 and Wasabi accounts, Import and Export Table Bundle dialogs omit placeholder text, and stalled request bodies no longer degrade a workspace for other users.

Security

• Security-group permissions are preserved correctly. Editing a security group no longer removes permissions the editor does not hold, and a non-administrator still cannot grant permissions they do not have. Permission and membership changes are now recorded in the security log. Review security groups edited by non-administrators before this release to confirm their permission sets. See Managing Security Groups and Assignments.

• Archive and project access checks are stronger. Importing an archive into an existing project preserves that project’s access restrictions, archives exclude table data from users who cannot read it, and project-content, project-lock, and workflow-step-copy requests now verify access to the named source project. Review restricted projects that received an archive before this release.

• Stronger boundaries for queries, assistants, and automation credentials. Submitted query components cannot carry arbitrary SQL; assistant workflow edits reject malformed saved structure; and a project’s automation credential is confined to that project instead of being able to manage workspace-level folders or projects. Existing automation that manages projects from a workflow may need support to re-enable that behavior.