We Hate Waste & Defects.

New

• Continue in chat from a workflow run. When a workflow finishes, you can now jump into a new chat seeded with the run output and keep iterating — from a single step or the whole run.

• Model multipliers in the model selector. Each model now surfaces its credit usage multiplier so you can make speed/cost/capability tradeoffs at a glance, including a callout for models on promotion.

• Generated files now attach to chats and workflows automatically. Charts, exported documents, transcripts, screenshots — anything Hatz creates appears reliably in your conversation or workflow output. Images render inline, other files appear as clickable cards you can open or download.
• The new Sources Modal now separates files you uploaded from files generated by AI, so users can reopen, download, or reference files from the current conversation more easily.

• Upload code, config, and legacy Office files. You can now bring almost any working file into Hatz — Python, JavaScript, TypeScript, Java, Go, Ruby, Rust, C/C++, C#, Swift, Kotlin, Scala, PHP, SQL, shell, and PowerShell code; JSON, XML, YAML, TOML, INI, .env, HTML, SVG, and log files; legacy Microsoft Word and PowerPoint (.doc and .ppt); and ZIP archives that automatically unpack into the code execution sandbox. Roughly 60 file extensions are now supported.

Improved

• We rebuilt the ConnectWise Manage integration from a 9-tool, tickets-only implementation. It now covers tools spanning time entries, agreements, configurations, schedule entries, projects, opportunities, documents, and members — plus workflow shortcuts.
• Faster Workshop loads. Lookups powering the Workshop are now substantially faster.
• Smarter Auto fallback. If you've selected a model in a workflow, app, or agent and it's been rate-limited, your run automatically falls back to Auto Model Selection — so the work continues with the next best option.
• Admins can now update their billing email and address from the billing settings page.

Fixed

• Setting your organization's default LLM to Auto now consistently saves all three modes (Lite, Performance, Turbo) as discrete options instead of locking you into the last variant you used personally.
• Generated PDFs now open in your browser with a proper title in the tab, and workflow-generated files more reliably open and download.
• Improvements to LMS reporting accuracy and user mapping.

New

• Hatz now supports audio file uploads, making it easier to bring meeting recordings, voice notes, and interviews directly into your workflows. Supported formats include mp3, mp4, m4a, wav, webm, ogg, flac, and aac. Uploaded audio is automatically transcribed and processed like any other file — available for search, context, and AI use across chat, agents, and workflows.

• 4 new integrations. can be found on the Connections tab.

Pylon - Connect Pylon to give Hatz access to your B2B support issues, accounts, and message history — so you can search and analyze tickets, identify trends across customers, and take action without leaving your workflow.

Jam - Connect Jam to give Hatz direct access to your bug recordings, console logs, network requests, and user event timelines — so you can drop in a Jam link and get straight to diagnosing and fixing the issue.

Sentry - Connect Sentry to give Hatz visibility into your application errors, stack traces, and performance data — so you can investigate issues, understand root causes, and get fix suggestions grounded in your live production monitoring.

Postman - Connect Postman to give Hatz access to your collections, environments, and API specifications — so you can run tests, generate client code, and manage your API workflows through natural language.

Improved

• Fixed an issue where later workflow steps could not access files saved by earlier steps if the original sandbox was no longer active; files are now stored more durably and reliably passed between steps.
• Custom MCP labels are now clearer when multiple are connected, reducing the chance of selecting the wrong tool.
• Corrected an error in the workflow editor where a disconnected integration could not be removed.

New

• Introducing Auto Model Selection modes: now when you use Auto Model Selection, you can choose how the system balances speed, cost, and capability—without picking a fixed model for every message.

Lite keeps things lightweight and fast for everyday questions and quick back-and-forth.

Performance is the balanced default: smart routing that matches the right level of model to each task so you get solid answers when you need depth and efficiency when you don't.

Turbo unlocks maximum intelligence for complex, high-stakes work where you want the product to be able to reach for its strongest options.

Together, the three modes make Auto more predictable for you and more flexible for how you actually work.

• Admins can now manage using custom roles. This gives teams flexibility to control who can view, edit, or collaborate within the workshop based on their specific role

View Community Workshop Items: Lets people browse and open agents and workflows that have been shared to the community (when community is enabled for your organization).

Publish Community Workshop Items: Lets people share their own eligible agents and workflows into the community so others can discover them (when community is enabled for your organization).

Improved

• Chat pages now show the conversation name in the browser tab, making it easier to keep track of multiple open chats.

• Custom Roles can now control how a user can use AI. You can turn Auto Model Selection on or off, and choose whether they may use Lite, Performance, or Turbo. You can also turn Memory, Personalization, and Sub-agents on or off for that role. Your company’s top-level rules still apply, but each role can be made stricter so different teams get different limits. This functionality can be used to further limit functionality - not expand it.

New

• analyzes each message you send in real-time and routes it to the most capable and cost-efficient model based on task type, complexity, files, and tools involved. Operates per-message, so a quick question and a complex coding task in the same conversation can each go to the right model. Set Auto-LLM as a personal default, or an MSP or tenant can set it as the default for the entire team. As new models are added, routing improves - you benefit automatically. Now available in chat, agents, workflows and apps.

• 8 new integrations. can be found on the Connections tab.

Netdata - Connect Netdata to give Hatz visibility into your infrastructure health and real-time performance metrics-so you can investigate anomalies, summarize system behavior, and draft remediation guidance grounded in live monitoring data.

Read AI - Connect Read AI to give Hatz visibility into your meeting summaries, transcripts, and engagement data-so you can review key takeaways, surface action items, and stay aligned on conversations without rewatching recordings.

Fellow - Connect Fellow to give Hatz visibility into your meeting notes, agendas, and action items-so you can track follow-ups, summarize discussions, and keep your team aligned without losing context between meetings.

Granola - Connect Granola to give Hatz visibility into your AI-generated meeting notes and transcripts-so you can surface decisions, extract action items, and reference conversation context without manually reviewing recordings.

Miro - Connect Miro to give Hatz visibility into your collaborative boards and visual workspaces-so you can summarize brainstorms, extract structured insights, and carry whiteboard context into your broader workflows.

Circleback - Connect Circleback to give Hatz visibility into your meeting notes and CRM-synced follow-ups-so you can track commitments, summarize outcomes, and ensure nothing falls through the cracks after every conversation.

AirOps - Connect AirOps to give Hatz visibility into your AI workflows and data pipelines-so you can review automation logic, summarize pipeline activity, and get structured guidance on optimizing your AI-powered operations.

Klaviyo - Connect Klaviyo to give Hatz visibility into your email and SMS marketing activity-so you can review campaign performance, summarize audience segments, and draft on-brand messaging grounded in real customer data.

• New LLM Gemma 4 added to the model selector for chat, apps, agents, and workflows.

- Photo provided by Google

• New LLM Gemini 3.1 Flash Lite added to the model selector for chat, apps, agents, and workflows.

Gemini 3.1 Flash-Lite, our fastest and most cost-efficient Gemini 3 series model. Built for high-volume developer workloads at scale, 3.1 Flash-Lite delivers high quality for its price and model tier. - Google

- Photo provided by Google

Improved:

• Each Chat Tool Template shows a small badge for how many tools it uses giving users a clearer indicator that ties the prompt to the tools it leverages.
• The chat screen adds a “Browse workflows” action that jumps straight into workflow discovery in the Workshop—so new and returning users have an obvious next step from chat into automation.

• Improvements to Python Code Execution - clearer language for what’s actually running.

Fixed

• A correction to an error that occasionally did not display prompt text after saving in the Workflow editor

New

• Users in the Admin Dashboard can now access Lark support directly within the platform. Chat access is limited to admin areas and automatically includes user and account context to reduce back-and-forth.

• Stop active workflow runs directly from the live run view, run history, and run summary. Safely halt in-progress work without waiting for timeouts or support intervention.

• You can now create multiple tenants at once by uploading a single CSV file, cutting onboarding time from hours to minutes. Built-in validation catches errors before submission, ensuring clean tenant data every time.

• 7 New Integrations can be found on the Connections tab.

Hunter - Connect Hunter to give Hatz visibility into contact and email data-so you can find verified professional email addresses, enrich leads, and draft outreach without manually hunting down contact details.

Jam - Connect Jam to give Hatz visibility into bug reports and user-submitted issues-so you can triage problems faster, summarize reported errors, and generate developer-ready context for quicker resolution.

Intercom - Connect Intercom to give Hatz visibility into customer conversations and support activity-so you can summarize open tickets, identify trends, and draft consistent, on-brand responses grounded in real customer context.

Sentry - Connect Sentry to give Hatz visibility into application errors and performance issues-so you can investigate incidents, summarize error patterns, and draft remediation guidance grounded in live monitoring data.

Calendly - Connect Calendly to give Hatz visibility into your scheduling availability and booking activity-so you can review upcoming meetings, identify scheduling gaps, and manage appointment workflows without leaving your workspace.

Postman - Connect Postman to give Hatz visibility into your API collections, environments, and request history-so you can summarize endpoint behavior, generate documentation, and get structured guidance on API design and testing.

Pylon - Connect Pylon to give Hatz visibility into your B2B customer support activity and account health-so you can surface open issues, summarize customer sentiment, and draft consistent responses grounded in enterprise account context.

• Claude Opus 4.7 is now available in Hatz.

Opus 4.7 is a notable improvement on Opus 4.6 in advanced software engineering, with particular gains on the most difficult tasks. Opus 4.7 handles complex, long-running tasks with rigor and consistency, pays precise attention to instructions, and devises ways to verify its own outputs before reporting back. The model also has substantially better vision: it can see images in greater resolution. It’s more tasteful and creative when completing professional tasks, producing higher-quality interfaces, slides, and docs.