Table of contents

Release

Forest

Feature

Forest Everywhere: your internal data and actions, in the tools each of your teams works in

Guillaume Rigal

0 min read

You built your data, actions and permissions in Forest. Forest Everywhere brings them to the tools each of your teams works in: the terminal, AI assistants, agents, internal tools and the helpdesk.

Your team builds its operations in Forest: data from every source in one place, the Forest actions that run your processes, the workflows that chain them, and the roles and permissions that decide who can do what.

Yet your users don’t all work in the same tool. Admins live in the terminal, often with Claude Code or Codex. Ops managers ask Claude or Dust about a case. Support answers customers in Zendesk. Product and ops people build internal tools with Lovable or Cursor. AI agents run checks around the clock.

Our vision is that the interface follows the user, and the controls stay with the data. Your team defines its operations once in Forest: the data, the actions, the workflows and who can use them. Each person, and each agent, then reaches them from the tools they already work in, under the same rules and in the same audit trail.

Forest Everywhere puts that vision into the product. It brings what you built in Forest to the primary tools of each team, with the same roles, permissions and activity logs. It rests on four capabilities:

  • CLI for admins. Configure Forest from the terminal, alone or through a coding agent.

  • MCP for AI assistants and agents. People work through their assistant as themselves, and autonomous agents work under service accounts.

  • Gateway API for builders. Engineers, product and ops people build the internal tools they need on Forest data and actions.

  • Apps and connectors for support. Forest works inside Zendesk and n8n, and Intercom, Pylon and Pappers data sits next to the rest of your data in Forest.

UI everywhere: one set of controls behind all your teams’ tools

Each new tool usually starts a new integration project: an API to expose the data, permissions to recode, an audit trail to rebuild. The teams we work with describe internal tooling as six-month projects that eat their product roadmap.

With Forest Everywhere, that work happens once. Your data, Forest actions and workflows stay in Forest, on your backend, in your infrastructure. Every tool reaches them through the Forest Gateway, and every call runs as the person or the service account that made it. The Forest backend checks the role, applies the permissions and writes the activity log, whichever tool sent the request.

CLI: admins configure Forest from the terminal, in minutes

In a large team, admins click through the same configuration screens for every role, team and environment.

The Forest CLI gives them the full configuration from the terminal:

  • Onboarding. npx forest-cli@latest start gives you Forest on your own data: a demo project, a standalone backend, or Forest inside an existing app.

  • Branches, environments and deployments. Work on a branch, push layout changes to a review environment, then deploy to production. Create, reset or delete environments.

  • Roles, permissions and layouts. Export every role’s permissions to one file, edit them for every team at once, and apply. Copy permissions between environments, or one team’s layout to another.

Admins run it alone, or hand the task to Claude Code or Codex: describe the change, and the agent edits the permissions file and runs the commands. Every change shows before it goes live, and the committed file goes through pull request review.

How to set it up: Forest CLI.

MCP: AI assistants and agents act on live data, and every action is traced

Ops teams already ask Claude or Dust about their cases, and our most advanced customers run autonomous agents for compliance checks. The Forest MCP Server, served by the Forest Gateway, lets both read records, run Forest actions and start workflows.

The first customers to adopt it show where this goes. Taking a sample of 10 customers who have deployed it in production, 84% of the changes agents made to live data in October went through a Forest action, under the same rules as their teams. At the most advanced, agents and the built-in assistant already account for up to 28% of all data changes.

We updated authentication to work for humans and agents alike:

  • People sign in as themselves. An analyst connects their AI assistant to Forest with their Forest account. The assistant sees what their role allows and runs only the actions their permissions grant.

  • Agents get a service account. An autonomous agent gets its own service account, with the role and teams you assign. Its actions carry its name.

Our tech team also updated the activity logs to match. The new Activity Logs page shows the work of people, assistants and agents across every collection, with field-level before and after values. Filter it by author, activity type, date range or free text.

Your security team decides how far access goes:

  • MCP tools. Pick which tools an assistant can see. New tools stay off until you enable them.

  • Client apps. Allowlist the client applications that may connect.

  • Tokens. Shorten access and refresh token lifetimes.

  • Files. Files an agent sends to a Forest action go to an upload URL, outside the model’s context.

How to set it up: MCP Server, Service accounts, Audit trail.

Gateway API: product and ops people build the internal tool they need

Ops leads know the screen they want: a review queue for one type of case, a panel inside the CRM, a portal for a partner team. Until now, they wrote a ticket and waited for an engineering sprint.

The Gateway API, served by the same Forest Gateway, gives those tools your records and Forest actions. Every write is checked against the caller’s role and logged. Engineers can build on it, and so can product and ops people:

  1. Build. Describe the tool in Lovable, Cursor or Claude Code. The OpenAPI document gives the coding tool the real routes to build against.

  2. Test. Point the tool at a test environment and try it on test data.

  3. Ship. Once the team has reviewed it, switch it to production. Each user signs in with their Forest account and sees only what their role allows.

How to set it up: Forest Gateway, Standalone Gateway.

Apps and connectors: support works with Forest inside the helpdesk

Without Forest, one support ticket means four screens: the helpdesk, an internal tool to find the customer, a registry website to check the company, and another screen to run the fix. Agents copy data by hand between them.

With Forest, the customer record, the company data and the actions sit in one place. The new Intercom, Pylon and Pappers connectors bring in conversations and tickets, issues and accounts, and company data as Forest collections, next to the rest of your data. The Pylon connector adds actions to create, reply to and close an issue.

It gets better when Forest comes to the helpdesk itself:

  • Zendesk app, new version. The app now runs on the Gateway API. On each ticket, it matches the requester to their Forest records and shows related data in the sidebar. Agents run the Forest actions their role allows, get AI-suggested actions, and generate a reply draft once the action has run. Records and matching never go to the AI service.

  • n8n node. Get records and run Forest actions from an n8n workflow. Forest events, such as a new record or an executed action, can start n8n workflows.

How to set it up: Zendesk app, n8n node, Intercom, Pylon.

One case, end to end, at a fictional fintech

Take a fictional fintech that offers business accounts. Every new company goes through a KYB review before it can send payments. Its support team works in Zendesk, its compliance analysts use an AI assistant, and an autonomous agent screens incoming documents.

The fintech’s admin set up Forest from the CLI, with the roles for every team in one reviewed permissions file. The compliance lead built a KYB review queue on the Gateway API with Lovable, tested it on staging, then moved it to production.

A new business customer writes in: their account still can’t send payments.

  1. The support agent opens the ticket in Zendesk. The Forest app matches the company and shows the KYB case, the missing document and the company data from the Pappers connector.

  2. The agent runs the Forest action the team built to request the document, then edits and sends the AI reply draft.

  3. When the document arrives, the screening agent, connected through MCP under its own service account, starts the KYB workflow on the record.

  4. A compliance analyst asks their AI assistant, signed in as themselves, for the case history, then approves the case in the review queue. The Forest backend checks their role and logs the decision.

When an auditor asks how the account was opened, the record’s Activity tab shows every step, filterable by author: the support agent, the screening agent and the analyst, each under their own name.

Ready to try it? Get started with the Forest CLI and set up Forest on your own data from the terminal.

November Changelog

What’s new in Forest

CLI and headless onboarding

  • npx forest-cli@latest start sets up Forest on your data from the terminal.

  • Export, apply and copy role permissions and team layouts across environments from the CLI.

MCP

  • The Forest Gateway serves the MCP Server and the Gateway API from one mount.

  • Restrict or expand the MCP tools an assistant can see.

  • OAuth improvements: allowlist client applications, shorter token lifetimes.

  • MCP file tool: files sent to Forest actions go to an upload URL, outside the model’s context.

Gateway API

  • A REST API for your own interfaces, with per-user sign-in, service accounts and an OpenAPI document.

  • Standalone Gateway (npm and Docker) for Ruby and Rails backends.

Apps and connectors

  • New version of the Zendesk app, running on the Gateway API.

  • n8n node: get records, run Forest actions, start n8n workflows from Forest events.

  • New connectors: Intercom, Pylon and Pappers.

Workspaces

  • Copy workspaces across teams.

  • Workspace projection, for faster loading.

  • Templating in the Workflow component, and other workspace improvements.

  • Ruby v1 speed-ups.

Actions

  • Redesigned actions dropdown.

  • “Select all” when approving actions.

Billing and licensing

  • New billing page, with a pricing recap.

  • The Forest backend packages are no longer under the GPL v3 licence.

Docs: Forest CLI · MCP Server · Forest Gateway · Zendesk app · n8n node · Workspaces

Forest is the operational layer for regulated fintechs. Connect your core banking, payment and compliance providers as datasources, build the actions and workflows your team runs every day, and keep permissions and a full audit trail on every call. Runs on your own infrastructure.

LEVEL UP YOUR OPS GAME

Every action traced, for every human, AI agent, BPO, LLM, and workflow.

LEVEL UP YOUR OPS GAME

Every action traced, for every human, AI agent, BPO, LLM, and workflow.

LEVEL UP YOUR OPS GAME

Every action traced, for every human, AI agent, BPO, LLM, and workflow.

The operational infrastructure regulated companies grow on

Copyright © 2026 Forest

Design by Alasta & Built by Reiya Studio