Changelog
New updates and improvements to Planhat
This suite of upgrades allows you to fully define your own models in Planhat, alongside system models like Company, End User, and Conversation. Your models get their own fields, associations, role-based permissions, previews and profiles, enabling many new use cases.
Custom associations: link a record on one model to a record on another. Primary associations make a record belong to its parent. They build hierarchies, prevent orphaned records, and pass portfolio permissions down chains that reach a Company.
Junction models: a model with two association fields, the pattern for many-to-many relationships (partners, deal contacts, stakeholder mapping), with metadata such as Role or start/end dates on the relationship itself.
Joined Tables: show a parent model’s fields as live columns in the child model’s tables, and group by them.
Cross-model formula fields: aggregate across associations, system or custom. Count, sum, or roll up associated records.
Integration: Salesforce, Spreadsheet imports and API are available currently.
To automate processes with agents, Planhat needs as much structured context as possible. Expanded Custom Models let you replicate your exact business structure. You can:
Give AI and agents clean, well-structured, associated data to reason over
Mirror a Salesforce object (Workspaces, Tenants, Customer Requests) into Planhat and have any changes reflect in either source
Roll KPIs up a hierarchy for segment or management reporting
Model many-to-many relationships and add metadata on each relationship
The number of models you can create scales with your plan, with add-ons for higher volumes.
You can now enable Human-in-the-Loop review on any step in an Automation sequence, AI-powered or not. When enabled, designated users or roles review the output before it proceeds to the next step. Available to all customers.
A few highlights:
Reviews can approve or reject the output of a step before the Automation continues
Assign review to specific users or roles, so the right person sees what needs attention
Configure timeouts and fallback behavior if no action is taken
Automations run in the background and surface only the steps that need review, useful for high-stakes actions like sending an AI-drafted email or updating a key data field.
You can now upload .md files to create Planhat Documents, and export any Planhat Document to .md format. Available to all customers now.
What this means:
Markdown is the native format for agents and LLMs, enabling better flows between Planhat and tools like Claude and ChatGPT
Supports building more performant knowledge bases on top of Planhat Documents
Works in both directions: import .md files as Documents, or export existing Documents as .md
You can now read and create documents in Planhat directly through MCP, so all users can move work between Planhat and tools like Claude or ChatGPT and bring the output back in for the team to access and collaborate on.
Use it to:
Build enablement material and pipeline analysis across new business, expansion, and upsell
Profile portfolio health, including churn risk and ICP fit
Surface trends across recent reported issues and feature requests
Analyse or brainstorm in Claude or ChatGPT, then pipe the result straight back into Planhat
OAuth connections to MCP now apply the highest allowed permission level automatically, so all users can connect with minimal setup while admins retain control over the maximum scope.
A few highlights:
Users can set up any OAuth connection without configuring permissions manually
Admins define the maximum scope, which OAuth enforces on every connection
Re-connect in Claude or ChatGPT with your existing setup to apply this
The Salesforce integration now supports formula fields alongside lookup fields for populating associations between records.
Use it to:
Map associations between records that lookup fields alone could not support
Bring across the most important data from Salesforce into Planhat with fewer workarounds
Expand what you can import from Salesforce without restructuring your field configuration
Automations have been updated with several UX improvements that reduce manual setup and make steps easier to configure correctly.
You can now:
Set specific fields when using a GET single step, giving you more control over what data is returned
Add, replace, or delete values in Team Members fields directly within a step
Smart-fill field values from earlier steps in the same Automation, including outputs from AI steps
Quick-fill IDs based on your trigger:
companyIdauto-populates in Get Company requests when the trigger is not a Company, and the record ID auto-populates in Update steps when the model matches the trigger model
You can now leave start and end dates unset on tasks in Workflow templates, so ad hoc tasks without a fixed deadline don't skew project timelines or trigger false delay flags.
How it works:
Set 0 in Start After Days / Duration to schedule a task for the same day
Leave Start After Days / Duration empty to create a task with no set dates (similar to "Someday" on standalone tasks)
Existing Workflows and templates are unaffected (previously empty or 0 values continue to be treated as same day)
Monthly scheduling is now available for Automation triggers, letting all users with access to App Center run automations on a chosen calendar day.
For example:
Run end-of-month pipeline reviews automatically on the last working day
Generate monthly summaries of feature requests or upcoming renewals
Trigger recurring AI analysis of customer data on a fixed cadence each month
All users can now sort columns and adjust which columns appear in a Data Table widget from the Dashboard itself, without editing the widget or cloning the page.
How to use:
Click any column header in Explore mode to sort ascending or descending
Right-click a column header to add, remove, or reorder columns
Save your sorting order and column selection as the default for that widget, the same way table preferences work elsewhere
Personalize a shared Dashboard view without creating duplicate pages
