📢 Changelog

🗓️ 2026-02-20

🚀 Features

🤖🐐 @file mentions in Otto chat — type @ to search and tag Project files as context, with autocomplete dropdown showing files with folder hierarchy and interactive file pills in messages
🤖🐐 Otto now supports Microsoft Foundry as an AI model provider (Private Preview), configurable in AI settings with API base URL and API key
New X-Small and 2X-Small runtime sizes available for Deployments

🌟 Improvements

🤖🐐 Improved conversation summarization to prevent Otto from restarting completed work after context compression in long conversations
🤖🐐 Improved thread event streaming reliability with proper event ordering, deduplication for new subscribers, and cleaner thread completion signaling

🛠️ Bug fixes

🤖🐐 Fixed Workspace auto-snooze interrupting active Otto chat sessions; snooze timer now resets during Otto interactions
🤖🐐 Fixed Otto Automations failing to extract assistant responses
Fixed memory leak from orphaned event listeners on Server-Sent Events connections causing unnecessary reconnections
Fixed build state not updating correctly during runtime state transitions

🗓️ 2026-02-13

🚀 Features

🤖🐐 Comprehensive markdown rendering overhaul in Otto:
- Improved code blocks with syntax highlighting, language labels, and line citations
- Mermaid diagram rendering with copy, PNG export, and fullscreen toolbar
- KaTeX math equation support (inline and block)
- GitHub-style alert callouts (NOTE, TIP, IMPORTANT, WARNING, CAUTION)
- Smart entity linking — type #component:name, #flow:name, etc. — with rich hover previews and direct navigation
- SVG rendering with lightbox preview, interactive tables, and table of contents sidebar
🤖🐐 Otto outputs are being renamed to Applications and are now discoverable in a dedicated Applications tab in the conversation sidebar, with search filtering and quick navigation to the originating conversation
🤖🐐 Otto now has awareness of your Project file structure from the start of each conversation, improving code analysis and assistance quality
🤖🐐 Conversation context summarization now displays real-time status and compression statistics (tokens saved, messages compressed) when Otto automatically optimizes conversation context
S3 Connections now support custom endpoints and addressing styles for S3-compatible services such as DigitalOcean Spaces, Wasabi, and MinIO

🌟 Improvements

Runtime pods are now shut down when an Instance is suspended, preventing unnecessary resource consumption
Improved locked account messaging with contextual explanations for different lock reasons (credit overage, plan expiration, plan limitations)
Improved Workspace startup and resizing reliability with better health tracking, graceful shutdown handling, and clearer error reporting during pod transitions

🛠️ Bug fixes

🤖🐐 Fixed AI provider settings (OpenAI, Azure OpenAI, Vertex AI, Google AI Studio, Bedrock) not displaying feedback when using verify, save, or clear buttons
Fixed DuckLake regression when using PostgreSQL as the metadata catalog with a custom catalog name
Fixed error viewing partitions on Snowflake caused by a column count mismatch between the partition view and the underlying table after metadata columns were added
Fixed duplicate Project creation during first-time Instance setup caused by a race condition
Fixed expand/collapse buttons on Flow and Application groups in the graph editor not responding to clicks
Fixed plan downgrade flow for trial users failing to detect current plan level and collect payment information
Fixed Repository SSH keys not consistently being stored in Vault

🗓️ 2026-02-05

🚀 Features

🤖🐐 Otto now supports Google Vertex AI and Google AI Studio as AI providers, enabling Gemini models with bring-your-own-key configuration

🌟 Improvements

🤖🐐 Improved conversation summarization to better preserve artifact content and provide more accurate status reporting during long conversations
🤖🐐 Added open-in-new-tab button to artifact cards for viewing artifacts directly in a new browser tab
🤖🐐 Improved error messages when MCP server connections fail, now showing the underlying server error instead of a generic "Connection closed" message
🤖🐐 Improved Otto chat response latency by caching project details and reducing synchronization delays
🤖🐐 Improved Otto dashboard welcome screen with a more consistent layout for new and returning users
🤖🐐 Otto suggested prompts now transition from curated prompts to personalized OttoPilot suggestions after 8 conversations
🤖🐐 Otto chat input now shows connection status and disables input until connected
Improved command palette (Cmd-K) search with fuzzy matching, theme switching actions, and keyboard shortcut improvements
Improved in-app feedback modal with category selection — support requests are now directed to the Ascend Community Slack and documentation
Faster CLI and runtime startup
Faster Instance startup by running first-time setup in the background instead of blocking the login flow
DuckDB Connections now support a threads configuration option for performance tuning, and validate that the memory limit does not exceed the runner process limit
The max_files setting in load_strategy is now honored during blob and Local File ingestion
Improved Databricks reliability by automatically retrying operations that fail with Delta Lake MetadataChangedException errors

🛠️ Bug fixes

🤖🐐 Fixed Otto query failures when a build was still in progress — Otto now automatically falls back to the most recent successful build
🤖🐐 Fixed an issue where Otto could become unresponsive after the LLM generated invalid tool names
🤖🐐 Fixed # autocomplete menu sometimes appearing at the wrong position on the page
Fixed partitioned Write Component failures on Snowflake and Databricks caused by incorrect column name casing in SQL partition queries
Fixed ad hoc queries referencing cross-Flow data failing to resolve when using DuckLake
Fixed cross-Flow references in DuckLake not correctly including all input dependencies

🗓️ 2026-01-29

🚀 Features

🤖🐐 Otto now executes SQL queries directly in chat:
- Two execution modes: auto-run (read-only queries execute immediately) and ask (requires user confirmation)
- Query results display in an inline data grid
- Open queries directly in worksheets for further exploration
- "Ask Otto" and "Fix" buttons added to worksheet results for seamless integration
🤖🐐 Otto creates rich interactive artifacts:
- Supports React components, HTML, SVG, Mermaid diagrams, code snippets, and markdown
- Inline clickable cards in chat with full modal viewer
- Version history with artifact browser to navigate all artifacts in a conversation
- Shareable URLs for direct linking to specific artifacts
- Artifacts can dynamically bind to query results with window.refreshData() API
🤖🐐 Otto can switch context dynamically during conversations:
- Automatically wakes paused Workspaces when needed
- Type # in chat input to quickly add Workspaces and Deployments to context
- New prompt bubble guides users to add a runtime when none is selected

🌟 Improvements

🤖🐐 Otto conversations now sync in real-time across browser windows with automatic updates for creation, changes, and deletions
🤖🐐 Cancel running Otto requests from any browser window
🤖🐐 Rename Otto conversations inline in the sidebar
🤖🐐 Automatic conversation summarization for long conversations
Improved Otto first-time user experience with refined categories and action-oriented prompts with documentation links
Improved Otto interface: messages start at top, new icon sidebar, enhanced loading screen, click logo to start new conversation
Improved Otto sidebar visual integration with semi-transparent backgrounds
Added Deployments settings page to Command+K quick search
Improved usage chart with better legend/tooltip display, date picker presets (Last 7/28/30 days, Last 3/6/12 months), and smart calendar defaults
Improved Workspace startup animation with updated process icons
Improved loading screen with animated status messages
DuckLake: Optimized data maintenance performance by leveraging table partitioning metadata, reducing data pruning time from minutes to seconds
DuckLake: Improved metadata catalog sync performance by automatically compacting databases when data inlining is disabled
DuckLake: Preview mode now connects directly to remote DuckDB metadata catalogs for faster, always-current data
DuckDB query results in artifact viewers now display complete results (previously limited to 1000 rows)
Improved Workspace startup time by approximately 10 seconds using local storage for small Workspaces
Faster Workspace and Deployment startup through event-driven runtime management
Faster initial Instance setup by running first-time initialization tasks in the background

🛠️ Bug fixes

🤖🐐 Fixed inline diff view not rendering for Claude Sonnet model responses
🤖🐐 Fixed issue where toggling query execution mode mid-conversation would cause queries to hang
🤖🐐 Fixed conversation history persistence issues
🤖🐐 Fixed tool outputs appearing stuck or pending after page refresh
🤖🐐 Fixed Otto failing on second message when Workspace context was attached
🤖🐐 Fixed conversations created after midnight UTC not appearing in conversation list
🤖🐐 Fixed blank screen when toggling auto-run mode
🤖🐐 Fixed flashing disconnect error when sending messages
🤖🐐 Fixed artifact version selector not updating displayed artifact
DuckLake: Fixed race condition causing "catalog does not exist" errors when running multiple worksheet queries in parallel
DuckLake: Fixed "catalog not found" errors when querying newly computed Components
DuckLake: Fixed issue where newly written data was not visible in the Records tab after running a Component

🗓️ 2026-01-22

🚀 Features

🤖🐐 New Otto capabilities:
- Fullscreen Otto experience: New users see an Otto-centric home page with guided prompts, suggested actions, and quick access to Workspaces and Deployments. Existing users can opt in via Settings.
- Agent inheritance: Custom agents can now extend built-in agents using the extends field, inheriting instructions, tools, and settings while adding custom behavior.
- Test Connections: Ask Otto to validate your Connection configurations before using them in Flows.
Cancel running Worksheet queries: Stop long-running queries in Worksheets with the new cancel button. Query execution is now asynchronous with real-time status updates. Supported on BigQuery, Databricks, MySQL, Oracle, PostgreSQL, Microsoft SQL Server, and Snowflake.
dbt projects with DuckDB Data Plane: dbt projects now support DuckDB as the Data Plane, with automatic Connection configuration — Ascend Connections are available to dbt without additional setup.
Billing usage analytics: View detailed usage history in Settings → Billing with charts showing credit consumption by day, week, or month. Filter by Deployment, Workspace, or AI usage, drill down by environment and runtime, and export to CSV.

🌟 Improvements

Deployments settings page is now searchable
DuckDB/DuckLake performance:
- Full refresh Flow runs skip unnecessary metadata catalog downloads
- Metadata catalogs are automatically compacted before syncing to blob storage
Improved reliability of real-time updates across the UI

🛠️ Bug fixes

Fixed dbt sources failing validation when no seed files exist. Sources can now be defined without corresponding seeds.
Fixed AWS credentials file corruption when multiple processes update credentials concurrently. Config writes are now atomic.

🗓️ 2026-01-14

🚀 Features

🤖🐐 New Otto capabilities:
- Send emails: Ask Otto to email you, specific Instance members, or all Instance members at once.
- Interactive forms: Otto can generate structured forms in chat to collect data from users. Forms support text, number, and secret fields with validation. Secret fields integrate with Vaults — users can select existing secrets or create new ones without exposing sensitive values. LLMs/AI service providers do not see secrets submitted through forms.
In-app product feedback: Submit feedback and feature requests directly from the app via the new feedback button in the header. Feedback routes to our product team for review.
Animated Workspace startup: A new animated visualization plays while your Workspace is starting up.

🌟 Improvements

🤖🐐 Increased default Otto chat sidebar width for better usability

🛠️ Bug fixes

Fixed Worksheets triggering unnecessary builds when saved. Worksheets now save without interrupting your development workflow.
Fixed DuckLake catalog exceptions during concurrent data operations. When using DuckDB as the DuckLake metadata catalog, concurrent INSERT operations could fail with catalog errors. All INSERT statements now properly acquire DDL locks when data inlining is enabled.
Fixed errors in blob file conversion being silently swallowed. Exceptions during file-to-parquet conversion now properly propagate to the main thread and surface to users instead of causing indefinite hangs.
Fixed the Run Flow button repeatedly showing loading state.
Fixed intermittent PostgreSQL connection errors (SSL connection has been closed unexpectedly) on long-idle Instances. Database connections now recycle after 30 minutes to prevent server-side timeouts.

🗓️ 2026-01-07

🚀 Features

Worksheets (Public Preview)

Worksheets

Run SQL queries directly in the Ascend UI with the new Worksheets editor. Position your cursor on any query and click Run to execute it, or select multiple queries to run them in sequence. A visual highlight bar shows exactly which queries will execute.

Worksheets are stored per-user in worksheets/<your-email>/ as regular .sql files you can rename and reorganize. Saving doesn't trigger builds. You can also run Worksheets in Deployments to query production data, though saving is disabled there.

Learn more about Worksheets →

Regex-based retry clauses

Define pattern-specific retry rules for Components with independent counters per error type:

retry_strategy:
  retry_clauses:
    - pattern: "503.*Service Unavailable"
      max_attempts: 10
    - pattern: "timeout"
      max_attempts: 3

Each pattern maintains its own retry count, so transient 503 errors won't exhaust your timeout retry budget.

Learn more about retry clauses →

Notification center

A new notification center in the navbar collects toast notifications with persistent history. Notifications are scoped to your user — click the bell icon to review your past notifications, dismiss individual items, or clear all. Toasts now dismiss 3x faster since you can always find them in the notification center.

Custom and high-memory runner sizes (Private Preview)

Configure CPU and memory for Workspaces, Flows, and Flow runs with new high-memory presets and custom size options. Contact your Ascend representative to enable this feature.

🌟 Improvements

Improved responsiveness of real-time updates across the UI
Preview operations are up to 3x faster due to caching
Flow run history displays timestamps above each run column with detailed start/finish times in tooltip
Flow run history adds logarithmic scale toggle for duration bars — useful when outlier runs distort the chart
Run filter dropdowns include All/Clear buttons for bulk selection

🛠️ Bug fixes

Fixed Flow runs timing out during graceful shutdown when tasks continue running
Fixed DuckLake data table compaction crashing with memory overflow on large tables
Fixed S3 credentials expiring after 1 hour in Kubernetes environments

🗓️ 2026-01-01

🚀 Features

DuckLake table partitioning: Configure partition_by in your DuckLake Data Plane options to partition tables by columns or expressions, enabling partition pruning for improved query performance.
data_plane: duckdb: ducklake: partition_by: - region
SQL expressions are also supported:
data_plane: duckdb: ducklake: partition_by: - year(created_at) - month(created_at)

🌟 Improvements

Flow names in the Deployment timeline view are now clickable, navigating directly to the Flow's run history.

🛠️ Bug fixes

Fixed S3 authentication failures in long-running Flows. Credentials using IAM roles now automatically refresh, preventing failures after the 1-hour token expiration.
Fixed Flows with large partition counts (100K+) failing on Snowflake due to LOB size limits. Partition metadata is now stored in individual rows instead of aggregated arrays.

🗓️ 2025-12-25

Minor bug fixes and performance improvements.

🗓️ 2025-12-18

🚀 Features

Ascend-managed DuckLake without PostgreSQL (Public Preview): Ascend-managed DuckLake now supports a fully cloud storage-based architecture — no PostgreSQL required. Both metadata and data are stored in object storage, simplifying infrastructure and reducing overhead in a way that can improve performance for certain workloads. Configure with $ascend_managed.duckdb_ducklake. See self-hosted DuckLake to bring your own storage.
dbt project integration (Private Preview): Run dbt projects as Ascend Applications with full support for dbt models, tests, and builds via the new dbt Component type.
🤖🐐 Otto improvements:
- Extended thinking support for Claude models — enable advanced reasoning with configurable token budgets
- Improved response conciseness and parallel tool execution

🌟 Improvements

DuckDB upgraded to 1.4.3 with MotherDuck compatibility.

🛠️ Bug fixes

Fixed identifier quoting across databases: database, schema, and table names starting with digits or containing special characters are now automatically quoted for Databricks, MS SQL Server, MySQL, Oracle, PostgreSQL, and Snowflake.
Fixed incremental merge writes to PostgreSQL failing when the target table lacked a unique index. Ascend now automatically creates the required unique index on unique_key columns when creating new tables.
Fixed Flow Graph freezing in an infinite loop when zooming to a node via the build panel.
Fixed Flow Graph not updating to reflect row counts and Component statuses after data changes.
Fixed branch selector crashing when typing an existing branch name instead of selecting it from the dropdown.

🗓️ 2025-12-11

🌟 Improvements

Super Graph rendering is 85% faster. Lazy tooltip initialization, memoized build graph calculations, and pre-computed derived values significantly reduce DOM overhead and recomputation on large Projects.

DuckLake table compaction now supports data tables in addition to metadata tables. Configure compaction thresholds per Component or globally via Profile defaults:

data_plane:
  duckdb:
    ducklake_data_table_compaction:
      small_file_record_count_limit: 100000
      small_file_count_threshold: 50
      small_file_ratio_threshold: 0.25

Component defaults now support wildcards in the flow_name field, enabling global configuration across all Flows using regex patterns.
🤖🐐 MCP tool calls now include explanations and descriptions for better visibility into Otto's actions.
Deployment pages now load Flow runs beyond the initial 30-day window. Historical runs are fetched in the background and stored locally for faster subsequent access.
DuckLake table compaction now includes helper tables created by Component store adapters, reducing fragmentation across all metadata tables.

🗓️ 2025-12-04

🚀 Features

🤖🐐 Otto can now navigate you through the Ascend UI. Ask Otto to open a specific file, go to a Deployment page, or view a Flow run, and it will take you there directly.
🤖🐐 OttoPilot is now available on the Dashboard. Get quick help, instance summaries, and suggested prompts without leaving the main view.

🛠️ Bug fixes

Fixed Microsoft SQL Server datetime predicates failing due to incorrect format. Datetime values are now formatted without timezone info and with millisecond precision.
Fixed DuckLake compaction using wrong catalog name when catalog override is configured.
Fixed Automation runs table only showing one run when multiple Automations triggered from the same event.
Fixed Flow run links in email notifications using incorrect project identifier.
Fixed Deployment Flow pages incorrectly redirecting to Super Graph view.
Fixed Snowflake Connection URLs generated with incorrect case for account locator.
Fixed SSH key display flickering when loading Repository settings.
Fixed clicking Component in Deployment Flow run history opening the wrong file.

🗓️ 2025-11-27

🚀 Features

🤖🐐 Otto can fetch and analyze web content. Enable web_access to let Otto read documentation, API references, and other online resources to better assist with your work.
Read Components now support strategy: full for native database connectors. Use this for small datasets or one-time loads that don't need partitioning:
read: connection: my_database strategy: full

🌟 Improvements

DuckLake now uses block-level row count statistics for dramatically faster output row calculations. Row counts that previously took minutes on large datasets are now near-instantaneous. Statistics are automatically computed during partition writes and backfilled for existing Components.
🤖🐐 Otto max turns handling improved. When approaching the turn limit (default 50), Otto receives a warning and gracefully summarizes progress instead of abruptly stopping.
Workspace status on Project page now correctly reflects actual state (Online, Paused, etc.).

🛠️ Bug fixes

Fixed Snowflake Read Components dropping _ascend_source and _ascend_ingested_at columns due to case sensitivity. Snowflake returns column names in uppercase, which now matches correctly.
Fixed Cmd+K quick search opening the wrong item when using keyboard navigation. The navigation index now correctly maps to visible items only.
Fixed Component form overwriting extra YAML fields when switching between code and form views. Custom configuration options are now preserved.

🗓️ 2025-11-20

🚀 Features

🤖🐐 Otto now defaults to agent mode, providing more proactive, capable assistance. Otto can now autonomously plan, execute, and iterate on tasks without requiring explicit step-by-step prompts.

🌟 Improvements

DuckLake metadata tables now automatically compact to prevent fragmentation. Tables with too many small files are detected and rewritten as single files during data maintenance.
MotherDuck no longer uses DDL locking, allowing concurrent schema operations.
Databricks Instance Store connections now handle invalid session handles gracefully with automatic reconnection. Added hourly connection pool recycling to prevent stale sessions.
Blob storage list operations (S3, GCS, ABFS) now use native client prefix filtering instead of directory traversal, significantly improving performance for prefix-based listings.
Table names with special characters (hyphens, spaces) now work correctly across all Data Planes. Quoting is applied automatically based on identifier validity.
Default Workspace auto-snooze timeout increased from 20 to 30 minutes.
Deployment Flows page loads faster with lazy timeline calculation, optimized date formatting, and a single shared popover instead of per-run popovers.

🛠️ Bug fixes

Fixed missing _ascend_source column in database-based Read Components. This regression caused schema mismatch errors when processing new data.
Fixed Incremental Read Components not respecting the normalize setting when no columns config is specified. Also fixed Smart Table meta columns incorrectly appearing in output when using columns config.
Fixed Flow run listing not filtering by Profile and Project path for runtime queries.

🗓️ 2025-11-13

🚀 Features

🤖🐐 New OttoPilot experience with onboarding, welcome chat, and dashboard integration. Otto now provides a guided introduction for new users and quick access from the dashboard.
DuckLake now supports use_nightly and force_install options for extension management. Use use_nightly: true for latest development builds or force_install: true to reinstall the extension.

🌟 Improvements

🤖🐐 Otto improvements:
- Jinja templating is now disabled by default for custom rules and agents. Rules with {{ }} syntax that aren't valid Jinja will no longer silently fail. Add jinja: true to frontmatter to enable Jinja rendering.
Smart Schema processing on DuckDB now uses temporary tables for intermediate operations, reducing write times by 33-47%.
Flow run data loads faster with improved browser caching.

🛠️ Bug fixes

🤖🐐 Otto fixes:
- Fixed Otto Automation failures when no explicit model was configured. Model selection now properly falls back through the priority chain (agent → defaults → Instance → provider).
Fixed unable to disable Workspace auto-snooze in settings.
Fixed MotherDuck logo not displaying in Explore tab.

🗓️ 2025-11-06

🛠️ Bug fixes

Fixed suspended Instance lockout controls to prevent content flashing before redirect.

🗓️ 2025-11-03

🚀 Features

🤖🐐 New Otto capabilities:
- Otto agent configuration now supports agent overrides, custom instructions, and configurable max_turns settings via otto.yaml, including commit message generation with context integration.
Write Components now support full refresh operations for object storage (GCS, S3, ABFS), BigQuery, MySQL, Oracle, and PostgreSQL Connection types. Full refresh resets and rebuilds destination tables from scratch.

🌟 Improvements

Snowflake authentication workflows updated to use key-pair authentication as password-based auth approaches deprecation.

🛠️ Bug fixes

Flow run errors now include the actual error message instead of generic timeout or segfault messages.
Fixed DuckDB Connections to PostgreSQL metadata failing with no password supplied errors due to race conditions. Atomic file replacement ensures complete password files are written before reading.
Fixed Flow Components in subdirectories not displaying run history due to path resolution issues.
Fixed Automation run error display formatting. Error messages now render with resize handles and copy buttons.
Fixed editor focus behavior when saving files or accepting AI changes.

🗓️ 2025-10-27

🚀 Features

🤖🐐 New Otto capabilities:
- Otto now generates commit messages using Conventional Commits with type prefixes, scopes, and breaking change indicators.
- Otto agent system updated with improved context gathering that traces symbols and enhanced tool descriptions for error-prone operations like RegEx escaping.
- Otto displays inline file diffs consistently across all AI model providers.
SQLFrame DataFrame support added across BigQuery, Snowflake, Databricks, and DuckDB Data Planes with Spark-like API and lazy evaluation.
Project defaults now support Flow runner size configurations using RegEx patterns to match Flow names.

🌟 Improvements

Added syntax highlighting for .ascendignore and other ignore files.
Snowflake authentication updated to use key pair authentication instead of deprecated password method.
Increased MySQL connection pool size to handle higher concurrent loads, reducing timeout errors for Instance-to-cloud GRPC communications.
Workspace pane automatically closes docked views when no tabs are present.
Deployment names in merge menus now match labels used elsewhere in the UI.
Tab loading logic enhanced with AbortController support for better resource management and error handling.

🛠️ Bug fixes

Otto fixes:
- Fixed patch parser to detect and correct malformed patch endings with extra + characters.
- Fixed Otto Chat title generation failing silently.
- Fixed commit message generation wrapping text in triple backticks or including commentary. Now outputs clean text without formatting artifacts.
Fixed empty data tables being dropped during maintenance, breaking downstream Components. Tables referenced by committed partitions are now protected from deletion.
Fixed table creation race conditions causing Clustering Key has changed errors. Unified SQL generation with DDL locking ensures consistent behavior across Data Planes.
Fixed Run Flow button text not updating when builds became out-of-date.
Fixed delete dialogs remaining open after successful deletions.
Improved local storage error handling for corrupted data, quota exceeded errors, and other storage failures.

🗓️ 2025-10-20

🚀 Features

🤖🐐 New Otto capabilities:
- Otto communication style updated to remove filler phrases like "Certainly!" and "Of course!", with improved context gathering and tool descriptions for error-prone operations.
Projects now support .ascendignore files to exclude parts of your Project during development.
- Profiles also support an ignore field for additional ignore patterns.

🌟 Improvements

Smart Schema support expanded:
- MySQL, Oracle, PostgreSQL, Microsoft SQL Server, BigQuery, Snowflake, and SFTP Read Components now support Smart Schema.
- Smart SQL Transforms, Smart Python Transforms, and Custom Python Transforms now support Smart Schema.

🛠️ Bug fixes

Fixed Component tests with severity: warn causing hard failures. Test severity parsing now handles all configuration patterns correctly.
Fixed authorization checks hitting rate limits during rapid UI interactions. Authorization checks are now cached.
Fixed polling being lost when switching browser tabs. Polling now pauses and resumes based on tab visibility.
Fixed Project page crashing with Cannot read properties of undefined errors when refreshing. Page now handles empty states and removed unnecessary polling.
Fixed Workspace tours breaking when encountering auto-snoozed Workspaces. Tours now detect Workspace state and prompt users to start paused Workspaces.

🗓️ 2025-10-13

🚀 Features

Added product overview and welcome video.
Flow configurations now support Flow runner size overrides for custom resource allocation per Flow.
Added nested subdirectories support for Flow and Component organization with collapsible directory views and summary statistics.

🌟 Improvements

PostgreSQL authentication updated to use pgpass files with automatic IAM token refresh.
Smart schema now supported on the Databricks Data Plane.
DuckDB upgraded to version 1.4.1.

🛠️ Bug fixes

Fixed concurrent DuckDB operations causing segmentation faults due to missing thread locks. Added locking for DDL operations, query execution, and batch insertions.
Fixed Smart Schema components with NULL partitioning versions failing during concurrent backfills with Invalid partitioning version: None errors.
Fixed DuckLake Connections on S3 failing after 1 hour due to expired AWS credentials. DuckLake now automatically refreshes credentials every hour.
Fixed multiple UI memory leaks from Monaco editors, polling handlers, and window event listeners.
Fixed paused Workspaces timing out when attempting to load Git status, causing 503 errors. Paused Workspaces now display a message instead of attempting to load Git status.

🗓️ 2025-10-06

🚀 Features

Otto can now send automated email alerts with optional AI-powered explanations.
Otto conversation limit increased from 25 to 50 maximum turns per request.

🌟 Improvements

DuckDB now defaults to max_combined_sql_statements=1 when using DuckLake for better performance and resource utilization. Prevents memory issues and CPU usage inefficiencies.
Build performance improved by up to 67% for large projects through global Jinja2 template caching, file I/O optimization, and threading improvements.
Otto Bedrock integration now supports prompt caching for Anthropic models, reducing costs and latency.
Flow runner resource allocation now supports configurable size overrides for CPU, memory, and disk allocation per runner.

🛠️ Bug fixes

Fixed MCP tool call responses failing due to serialization issues with complex data types. JSON serialization now handles all data types correctly.
Fixed segmentation fault in DuckLake by implementing safer check for partitioning version column existence.

🗓️ 2025-09-29

🚀 Features

🤖🐐 New Otto capabilities:
- Claude Sonnet 4.5 model is now the default LLM for Otto.
Added guided tour for first-time users.

🌟 Improvements

File upload memory management improved with single-threaded parquet conversion to prevent OOM crashes.
DuckDB DDL operations now properly locked to prevent race conditions when multiple operations run with task-threads > 1.
Documentation site redesigned with updated visual styling.

🛠️ Bug fixes

Fixed BigQuery Read Components with column casting throwing syntax errors due to missing fully qualified table names.
Fixed BigQuery Flows with concurrent components writing to the same table failing with Transaction is aborted due to concurrent update errors. Added automatic retry logic with exponential backoff.
Fixed COALESCE expressions in partitioned Data Plane operations causing SQL type inference issues. All COALESCE expressions now use explicit CAST(NULL AS {type}) syntax.
Fixed adding columns to existing Databricks tables failing with [PARSE_SYNTAX_ERROR] errors. Column addition now works correctly across all Data Planes.

🗓️ 2025-09-22

🚀 Features

🤖🐐 New Otto capabilities:
- Otto now supports LLMs across providers with Instance-level model management settings. Configure under AI & Models.
- Otto now works with Jinja macros, parameters, Automations, Tasks, and SSH Tunnels with updated rules and reduced token usage.
Smart schema evolution is now the default schema change strategy for Read Components with object storage and Local File Connections on all Data Planes. Data partitions store their own schemas and reconcile type differences without copying data on schema changes.
- Existing Components continue to use the full strategy. New Components default to smart strategy.
Workspace auto-snooze automatically pauses inactive Workspaces after a configurable timeout (5-120 minutes), with detection of ongoing Flow runs to avoid interruption.
- Auto-snooze applies to all new Workspaces with a default timeout of 10 minutes.

🌟 Improvements

Focus on a Component in the build panel to jump to its location in the Flow Graph, or on a Flow to jump to its location in the Super Graph.
Data Plane Connections now appear in Flow Connections list. Build info panel displays the number of Connections used in a Flow.
Lists throughout Ascend Instance now sort alphabetically including Deployment, Project, Git branch, Profile, and Environment selectors.

🛠️ Bug fixes

PostgreSQL Read Components now preserve array and JSON column types instead of converting to VARCHAR. Added arrays_as_json parameter for non-mappable array data.
Fixed incremental merger resets failing with malformed SQL syntax errors. Qualified table names now generate proper DROP statements across Data Planes.

🗓️ 2025-09-15

⚠️ Breaking changes

GCS and ABFS Connections now use more stable metadata fields (md5Hash, crc32c for GCS; content_md5, etag for ABFS) for fingerprinting to prevent unnecessary re-ingests when storage tiers change. Requires re-ingesting existing data.

🌟 Improvements

Deployment run pages now link directly to individual Flows instead of the Super Graph.
Flex Code Flow creation forms streamlined to include only essential fields (name, description, Data Plane Connection, parameters, and defaults).
BigQuery Data Plane replaced custom UDF logic with native IF functions for better performance.
DuckLake on S3 now uses httpfs as the default Connection method.
Added local file caching support for DuckLake with S3, GCS, and Azure Blob Storage.

🛠️ Bug fixes

Fixed DuckDB SAFE_STRING macro creation failing in DuckLake environments with expression depth limit errors. Added Python UDF fallback.
Fixed long-running DuckLake jobs losing connection to metadata database.
Fixed DuckDB-based ingestion not working with multiple PostgreSQL Components simultaneously.

🗓️ 2025-09-08

🚀 Features

Added drag-and-drop file upload from computer to file tree.

🌟 Improvements

Moved Instance status indicator to right side of header next to user menu.

🛠️ Bug fixes

Fixed ascend view connection sample command failing with pyarrow.lib.ArrowInvalid error when PostgreSQL tables contain range type columns. Command now converts range types to string representations.
Fixed incremental reads not honoring start values during backfill phases.

🗓️ 2025-09-01

🚀 Features

MySQL and PostgreSQL databases hosted by AWS can now authenticate using AWS IAM roles.

🛠️ Bug fixes

Fixed Databricks SQL connector compatibility issue due to renamed API. System now handles both API versions with fallback logic.
Fixed partition_template functionality in blob Write Components. Datetime templates are now properly parsed and interpolated.

🗓️ 2025-08-25

🚀 Features

🤖🐐 New Otto capabilities: Otto can now be used as an Action in Automations to respond to Flow events, analyze data patterns, and execute actions across multiple platforms:

automation: name: otto-notifications enabled: true triggers: events:

sql_filter: json_extract_string(event, '$.data.flow') = 'transform-demo' types:
- FlowRunError
- FlowRunSuccess

actions:

type: run_otto name: run-otto config: prompt: >-

Instructions

You are being invoked at the end of a flow run. Your job is to assess changes in flow run behavior, and send a Slack message to keep other users updated. You cannot ask for help or clarification; you must do this end-to-end on your own based on your own judgement. Your last action should always be to post a message to Slack.

Flow run analysis rules

Analyze flow success / failure by fetching the maximum number of flow runs.
1. If the flow run was a success and the previous flow run was also a success, send a funny message congratulating the team on another smooth run.
2. If the flow run was a success and the previous flow run was not, send an exciting and congratulatory messages celebrating the flow is now fixed, and congratulating whichever user did the fixing (check git history).
3. If the flow run was a failure and the previous flow run was also a failure, send a sad message imploring somebody to fix it.
4. If the flow run was a failure and the previous flow run was not, send a big alert message. You should try to identify who broke the flow, and even @ mention them if you can. You should also try to analyze the changes and suggest fixes if possible.
Additionally, you should make heavy use of git history and file reading calls to help track down details that may be relevant.

Slack message and notification

At the end of your analysis, you must post a message to Slack in the appropriate channel.

Your slack post message should:
- include a link to the flow run: https://[runtime.link_url]/flows/[flow_name]/[run_name]
- be formatted for Slack rendering, which is different than markdown
- when @ mentioning users, look up their profile so you can reference their actual id

### 🌟 Improvements

- [Application](/concepts/application/application.mdx) Components can now access Flow parameters during build time, enabling conditional Component generation based on Profile and Flow parameter values.
- Enhanced Automation configuration UI with improved layout on small screens and delete buttons on Sensor, Event, and Action cards.

## 🗓️ 2025-08-18

### 🚀 Features

- Added [Ascend-managed Repositories](/how-to/setup/repository/managed.mdx) for creating repositories directly within the platform.
<Video controls={true} playing={true} muted={true} url="https://storage.googleapis.com/ascend-docs-assets-bucket/whats-new/ascend-managed-repo.mov" />

### ⚠️ Breaking changes

- Fixed Oracle connection configuration validation to prevent runtime failures. Added model-level validation to ensure mutually exclusive Oracle Connection parameters (`dsn`, `database`, `service_name`) are handled correctly.
- Users with existing Oracle connections using both `database` and `service_name` fields must choose one Connection method.

### 🛠️ Bug fixes

- Fixed landing table duplicate data issue in incremental persist. Added de-duplication logic to prevent `UPDATE/MERGE must match at most one source row for each target row` errors.

## 🗓️ 2025-08-11

### 🚀 Features

- Ascend now supports DuckDB via DuckLake as a Data Plane. [Follow our guide](/how-to/setup/data-planes/duckdb/duckdb.mdx) to get started.

### 🌟 Improvements

- Database incremental reads now process up to 50% faster through concurrent download and upload operations using threading.

### 🛠️ Bug fixes

- Fixed `Failed to list resources` errors when browsing Connection data containing variable placeholders in file paths.
- Fixed incremental merge queries failing with syntax errors when column names contained special characters or reserved keywords. Column names are now properly quoted.
- Fixed job failures and schema inconsistencies when running Snowflake DML operations concurrently or with metadata drift. Added protection from concurrency limits and error detection for schema inconsistencies.

## 🗓️ 2025-08-04

### 🌟 Improvements

- Settings interface updated with unified design and consistent card styling.

<Video controls={true} playing={true} muted={true} url="https://storage.googleapis.com/ascend-docs-assets-bucket/whats-new/settings.mp4" />

### 🛠️ Bug fixes

- Fixed PostgreSQL Read Component incremental strategy memory issue where Components were loading entire tables instead of empty tables during initial runs. Predicates from min/max queries are now properly applied to data fetching queries.

## 🗓️ 2025-07-28

### 🚀 Features

- Added [PostgreSQL Write Component](/how-to/write/databases/postgresql.mdx) with support for multiple write strategies: snapshot, full, partitioned, and incremental.

### 🌟 Improvements

- Added light/dark mode toggle in General settings.

### 🛠️ Bug fixes

- Fixed Incremental Read Components with full-refresh flag returning empty results. Full-refresh now properly resets both output and metadata tables.

## 🗓️ 2025-07-21

### 🚀 Features

- 🤖🐐 New Otto capabilities:
- Otto now provides code completion suggestions in the UI, powered by context from rules and files.
  ![Otto code complete](https://storage.googleapis.com/ascend-docs-assets-bucket/whats-new/code-complete.gif)
- Otto can now be configured with Automations to send activity summary emails. See our [how-to guide](/how-to/otto/summary.mdx) for instructions.
- Files panel now supports drag-and-drop for existing files.
- Hold option/alt key while dragging to create copies.
- Files panel right-click menu now includes `Duplicate` option.

### ⚠️ Breaking changes

- Default Write Component strategy changed from Smart (partitioned) to Simple. Smart Components are more scalable but more complex for simple datasets.
- Restore previous behavior by explicitly setting `strategy: partitioned`. See [write strategies documentation](/concepts/component/write/strategies.mdx) for details.

### 🛠️ Bug fixes

- Fixed SSH public key input box appearing for all secret fields instead of SSH-specific ones only.
- Fixed date indicators in Flow runs timeline view being unstable due to CSS issues.
- Fixed BigQuery Transform queries failing when referencing newly created columns in the same query. Column references are now automatically moved to a separate CTE.

## 🗓️ 2025-07-07

### 🚀 Features

- 🤖🐐 New Otto capabilities:

- Otto now supports [custom agents](/how-to/otto/agents.mdx) for creating specialized behaviors tailored to your workflows. Define agents in markdown with custom traits and tool access.

  - Tool inclusion supports wildcards: `[category].*` for all tools of a specific type, or `*` for everything. Example:

  ```md title="otto/agents/custom_agent.md"
  ---
  otto:
    agent:
      name: Standup Reporter
      mcp_servers:
        - slack
      model: gpt-4o
      model_settings:
        temperature: 0.1
      tools:
        - "*"
  ---

  # Standup Reporter

  You are a professional activity reporter specialized in summarizing Ascend platform work.

  Your job is to create comprehensive weekly summaries of my Ascend activities and share them with my manager via Slack.
  ...

Otto introduces custom rules for adding special instructions for chat interactions.
Otto now connects to external MCP servers via configuration files.
Otto's configuration now uses otto.yaml file in a dedicated otto directory for centralized agent and MCP server configuration.

🌟 Improvements

Write Components for blob storage (ABFS, GCS, S3) now support configurable chunk size using the part_file_rows field to optimize performance and avoid OOM errors:

write_custom_chunk.yaml
component:
  write:
    connection: write_s3
    input:
      name: my_component
      flow: my_flow
    s3:
      path: /some_other_dir/
      formatter: json
    part_file_rows: 1000

🛠️ Bug fixes

Fixed database Read Components (PostgreSQL, MySQL, Microsoft SQL Server, Oracle) converting SQL null values to the literal string 'None' when loading query results into PyArrow arrays. Null values are now properly preserved.
- This change affects all pipelines using database Read Components across all Data Planes.
- If you previously added cleanup steps to handle 'None' string values, you can remove them.

🗓️ 2025-06-23

🚀 Features

🤖🐐 New Otto capabilities: Otto now renders tool calls, showing which tools are called, their inputs, and outputs.

🌟 Improvements

Write Components now support improved performance for large uploads through chunked blob storage:
- Paths ending with a filename (e.g., .parquet) → single file output (unchanged)
- Paths ending with a trailing slash (/) → chunked output with part_<chunkid> format
- Partition writes now always produce chunked output files
Flows can now be run from Deployment history pages when on the latest Project build.

🛠️ Bug fixes

Fixed action failures in Automations interrupting the entire workflow. Individual action failures are now isolated.

🗓️ 2025-06-16

🚀 Features

Private Python package repositories now supported for AWS CodeArtifact and GCP Artifact Registry via pyproject.toml files. See how-to guide for setup instructions.

🌟 Improvements

Instance header now displays error when Instance backend is unhealthy:
Improved error messages for OOM process terminations:

Flow runner process for my_flow/fr-01977f3c-845b-72a1-b225-4327835f8434 exited with code -9.
Exit code -9 usually means the kernel's Out-Of-Memory (OOM) killer terminated the process after it exceeded its memory limit.
Consider increasing the runner size or optimizing the workflow's memory usage.

🛠️ Bug fixes

Fixed build history calendar days outside the current month not being selectable.

🗓️ 2025-06-09

🛠️ Bug fixes

Fixed Workspace settings not clearing old run history when switching Projects.
Fixed status bar on Explore tab showing outdated information. Now updates in real-time.
Fixed Workspace settings URLs not navigating to correct pages.
Fixed Workspace or Deployment size settings not being respected.
Improved "Flow runner not found" error messages to include detailed pod status information (OOM, timeout, unexpected states, API/connection errors).
Fixed Flow runners staying active when idle. Now automatically shut down when not in use.
Fixed Git operations hanging indefinitely when repositories became unresponsive. Now timeout automatically.
Fixed Application parameters not working with Incremental logic in SQL transforms.
Fixed partition filter and partition value analyzers interfering with each other during concurrent backfills. Now run independently.
Fixed text parser crashing when given string file paths in Arrow and pandas reads.

🗓️ 2025-06-02

🚀 Features

🤖🐐 Otto can now provide SQL linting information via SQLFluff.
Command-K search now supports navigation to settings pages.

🌟 Improvements

Improved UI rendering of errors and warnings.
Added Polars DataFrame support as input/output format in Python transforms.
Parquet processing now normalizes column names to avoid case conflicts.
Automation sensors now support timezone configuration.
Automation failed status now shown when there is no associated Flow run.

🛠️ Bug fixes

Fixed column case sensitivity handling for Python Read Components on Snowflake.
Fixed Profile name, Project UUID, and path not being persisted into build info.
Fixed Transformations with upstream Incremental Read Components not resolving to correct table name.
Improved Unicode and emoji handling.
Fixed Component/Run state carrying over from current builds into historical ones.
Added timezone information to timestamps across the UI.
Fixed record columns not being resizable in Explore view.

🗓️ 2025-05-26

🚀 Features

Added syntax highlighting for INI, TOML, and SQLFluff files.
AWS Managed Vault now supported as Instance or Environment Vault.
🤖🐐 New Otto capabilities:
- Otto can now validate YAML files and test Connections.
- Otto automatically configures tools based on environment context (Workspace vs. Deployment).

🌟 Improvements

Streamlined Python interfaces in Simple Applications for easier migration from regular Flows.
Added copy button to Connection test errors.

🛠️ Bug fixes

Fixed multi-line comment parsing in SQL and SQL Jinja files.
Fixed compatibility issues between .yaml and .yml file extensions.
Optimized error stack traces to be more concise.
Fixed PostgreSQL Connections SSL verification and empty port configuration handling.
Fixed clickable error links in Component cards within Deployments.
Fixed arrow keys not working when renaming files.

🗓️ 2025-05-19

🚀 Features

🤖🐐 New Otto capabilities:
- Project file health checks including YAML linting and fixes
- Connection testing and issue resolution
- Connection listing and exploration
- Flow and Component execution, monitoring, and troubleshooting
Components can now have non-data graph dependencies to any other Component.
Added Git status indicators to file browser and tab bar.
Retry logic now configurable for all Component types.
Added pause-all functionality for Deployment Automations.

🌟 Improvements

Runs table column widths now remember user preferences.
Improved Automation form workflow.
Component build errors now provide more specific information about which Component failed.

🛠️ Bug fixes

Fixed Databricks Connections not reconnecting when encountering "Database Invalid session" errors.
Fixed file refresh not updating cached and open files.
Fixed repo save button state not reflecting current state correctly.
Improved branch listing operation reliability.
Fixed zooming on individual nodes in expanded Application Component graph tab.
Fixed Databricks Connections catalog reference handling.
Fixed UI state consistency issues during rapid save, build, and run operations.
Fixed MySQL Connections with SSL=True throwing exceptions.
Fixed table references not using full names in merge operations.
Improved detection and handling of build failures caused by OOM conditions.
Fixed empty files causing Project builds to fail.

🗓️ 2026-02-20​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-02-13​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-02-05​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-01-29​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-01-22​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-01-14​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-01-07​

🚀 Features​

Worksheets (Public Preview)​

Regex-based retry clauses​

Notification center​

Custom and high-memory runner sizes (Private Preview)​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-01-01​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-12-25​

🗓️ 2025-12-18​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-12-11​

🌟 Improvements​

🗓️ 2025-12-04​

🚀 Features​

🛠️ Bug fixes​

🗓️ 2025-11-27​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-11-20​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-11-13​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-11-06​

🛠️ Bug fixes​

🗓️ 2025-11-03​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-10-27​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-10-20​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-10-13​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2025-10-06​

🚀 Features​

🌟 Improvements​

🛠️ Bug fixes​

🗓️ 2026-02-20

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-02-13

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-02-05

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-01-29

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-01-22

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-01-14

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-01-07

🚀 Features

Worksheets (Public Preview)

Regex-based retry clauses

Notification center

Custom and high-memory runner sizes (Private Preview)

🌟 Improvements

🛠️ Bug fixes

🗓️ 2026-01-01

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-12-25

🗓️ 2025-12-18

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-12-11

🌟 Improvements

🗓️ 2025-12-04

🚀 Features

🛠️ Bug fixes

🗓️ 2025-11-27

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-11-20

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-11-13

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-11-06

🛠️ Bug fixes

🗓️ 2025-11-03

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-10-27

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-10-20

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-10-13

🚀 Features

🌟 Improvements

🛠️ Bug fixes

🗓️ 2025-10-06

🚀 Features

🌟 Improvements

🛠️ Bug fixes