Changelog

Structured Output from AI Steps

No more messy, unstructured data back from your LLMs. This feature lets you control exactly what Planhat’s AI step outputs — in a structured format (JSON). Define fields, the expected data type for each, and whether they’re required to ensure consistent results every time.

Add field descriptions to guide the AI with clear context for what each field should contain — helping it generate more accurate and reliable output. The resulting JSON can be used for branching logic, or to create and update data directly. Support for using your existing field configurations from Planhat is coming soon — so you won’t need to define them manually.

Accurate AI with Markdown KV Formatter

When sending CRM-style input to an LLM — like account details, end-user information, or conversation metadata — formatting it as Markdown KV pairs leads to better results. It reduces hallucinations and improves accuracy by giving each data point a clear label for context. This formatter is now available in Automations to help structure your prompts more effectively.

Assign External Users to Tasks and Notes in Portals

External users can now be assigned as Task owners in Portals, enabling shared accountability in project planning and onboarding workflows.

Add new Fields from apps and Integrations

The ability to add new fields directly from field pickers has been extended to apps, app-templates, and integrations. Whether you are integrating with a new source, or building an automation to enrich your records, you can create and map your fields in the same motion.

Dynamic Dates in Formula Fields + Performance Improvement

You can now use formula fields to calculate differences between dates, expressed in days, weeks, months, quarters, years (including fiscal periods). This works for both same and cross-model calculations.

With this, you can attribute events and milestones to defined time periods, like counting the number of tickets from a customer so far this year. You can also track targets, like summing up opportunities closing this quarter, using calculated metrics to track how a rep’s target attainment develops over time.

You can read more here.

Upgrades to User Management

Planhat has made various upgrades to user management in the application.

You will now see a concept of User Types in Planhat; Core, Productivity and View, as well as Service Seats for customers with Service Delivery available in your plan.

User Type and Seat are set on Role and User to control eligibility of Roles for Users. The number of available Core Users, Productivity Users, View Users, and Service Seats are governed by your plan.

You can read more here.

Customise Automation Templates

On most Automation Templates, you can now press "Customise" to open the raw automation editor.

Great to customise general templates after your specific case (eg, prompt), or to build the flow out based on a core automation sequence.

Create custom fields from field pickers

Now you can update/manage your data model "from the edge".

This creates a more natural flow when you want to add a column, create an automated flow, or integrate - and can directly adapt your data model. Small but useful improvement. Over next week, adding this across all field pickers.

Run Automations Directly from Tables

Configure automations to be triggered manually, and run them on records of your choosing, directly from data tables.

You can choose whether to run automations one-by-one (using automation buttons) or in bulk.

The Action Title and Confirmation Message that you defined when setting up the Automation, both show up when applying the automation.

Automation Buttons

The Buttons are available in all data tables, be it in the Data Explorer, a data table page, or a widget.

Configuring Visibility of Buttons

In most data tables, you can choose whether to hide automation buttons in column preferences. In data table widgets, however, you configure this in the customisation section.

Running Automations in Bulk

When running automations in bulk, simply select the records you want to include, and run the automation from bulk actions.

Bonus Feature: Fetching Multiple full Email/Ticket/Chat threads in Automations

Another enhancement to Automations, unrelated to the above buttons and bulk-actions. You can now fetch the full thread, or the latest entry, for up to 10 conversations at a time, in automation steps. This is ideal when constructing automations that require a lot of conversational context, e.g. when feeding the data to an LLM.

Analyse Conversation content in Automations

Include the full Email, Chat and Ticket thread content in GET Conversation steps (vs. previously only accessing a short snippet). Paired with AI Steps, allows for everything from intelligent risk flagging, finding competitor mentions, to identifying decision-makers or project blockers.

Actionable Notifications in Automations

Notifications are now fully actionable — customise the title and where the notification leads (a Conversation, a Company) to help your team execute faster.

Introducing Formatters in Automations

We’ve also introduced a formatters in our Automations, allowing you to transform data into JSON, Markdown, HTML, or CSV on the fly. Great for preparing the content for AI steps, rich text fields, documents or external tools.

Improved UX for selecting and sorting GET requests

We’ve replaced free text fields for with dropdown pickers when configuring sorting and selecting GET steps in Automations. This makes setup faster and more intuitive.

Share Issues on Portals [Beta]

End users can now access and collaborate on Issues shared within their Portal, enabling transparent resolution and collaboration directly in the Portal.

Multiple Enduser Roles on Portals [Beta]

Create multiple Enduser Roles with granular permissions to control exactly what each type of Portal user can access and do, tailoring the experience for different stakeholders.

View Portal as an End User [Beta]

Test and debug your Portal experience from your customers' perspective by impersonating any end user from Portal settings. This makes it easy to verify permissions and keep data secure.

Introducing Claude Sonnet 4.5 in AI Model Hub

Claude Sonnet 4.5 is now available in our AI Model Hub and AI workflows. Thanks to its enhanced quantitative reasoning, sustained task performance, your automations can tackle more complex logic and deliver deeper reasoning with less oversight.

Improved Widget Drill-Down

There's now deeper capabilities for filtering and customizing columns in widget drill-downs, to understand the underlying data more in depth.

Improved UX for Custom Connections

Expand input fields for lengthier parameter setting when using an integration in Automations, e.g. for better UX when connecting to your preferred LLM.

Portal Engagement Metrics [Early Access] 

Analyze Portal engagement with new activity metrics (End User Portal Activity and End User Added Comment), and the Active Portal User system metric.