We introduced AI app generation in ToolJet before. It showed us what was possible. It also showed us the limits of our first approach.

The AI worked for simpler applications, but it was not reliable enough for the complexity of real business software. It understood only a small part of the ToolJet platform. As applications grew, users often had to complete manually what AI had started.

We could have kept patching it. Instead, we rebuilt the entire experience.

  • MCP-based implementation; connect with other MCPs to import designs, migrate non-ToolJet apps, etc.  
  • ToolJet AI is rebuilt from scratch; supports building complex use cases, uses less tokens, & faster. 
  • 100+ data source integrations are now supported (compared to 10). 
  • ToolJet database is now supported; build apps without bringing your own data layer.
  • New governance MCP; manage permissions & other admins stuffs from your AI agents.

Build from the coding agent you already use

We are also launching the new ToolJet MCP.

ToolJet MCP allows Claude Code, Codex, Cursor and other MCP-compatible agents to create, inspect and modify ToolJet applications directly. This means developers do not have to leave their preferred coding agent to build on ToolJet. They can also use the model subscriptions and token limits they already have through their own Claude Code, Codex or other accounts.

Because MCPs can work together inside the same agent, entirely new workflows become possible.

For example, an agent can read a design from Figma using Figma MCP, understand the design system and screens, and then build the working application in ToolJet using ToolJet MCP.

The same model applies to other tools and sources available to the agent.

ToolJet no longer dictates where the building must happen. It provides the structured application platform and production runtime underneath whichever AI agent your team prefers.

An entirely new AI Builder

The new AI Builder can create complete business applications – including pages, interfaces, queries, data connections, and complex application logic.

More importantly, AI now works across all 100+ integrations supported by ToolJet.

It can build applications using the databases, APIs and SaaS tools your business already depends on, rather than generating an isolated demo with sample data.

You can describe what you need, let ToolJet build it, and continue refining it through natural language. At any point, you can switch to the visual builder. 

Governance for agent-built software

Giving AI more power to build software also creates a new problem: how do companies retain control?

Our new Governance MCP brings ToolJet’s governance capabilities into agent-driven workflows.

Organizations can adopt AI-assisted development without creating a separate layer of ungoverned applications outside their existing security and operational controls. Applications built through ToolJet AI or an external coding agent can still use ToolJet’s production foundation:

Authentication, permissions and role-based access
SSO and audit logs
Development and production environments
Git-based workflows
Secure connections to business data
ToolJet Cloud, VPC and self-hosted deployment options

A different way to build business software

For developers, it means building from the coding agent they already use, with the model account they already pay for.

For enterprises, it means allowing more people and agents to build software while maintaining governance, security and deployment control.
For a small business, the new ToolJet means being able to describe an internal tool and start using it without assembling a dedicated development team.

For business and operations teams, it means turning workflows and ideas into working software without waiting in an engineering backlog.

For designers, it creates a path from a Figma design to a functioning application.

Everything does not have to be built through the same interface anymore.

Build with ToolJet AI. Build from your preferred coding agent. Refine visually. Extend with code. Govern and deploy through ToolJet. Explore the new ToolJet.

Developers can also install the new ToolJet MCP.

When we started ToolJet in 2021, our goal was to make internal software easier to build.

AI has fundamentally changed who – and what – can build software. A builder can now be a developer using Codex, a designer working in Figma, an operations leader describing a workflow, or an AI agent working across several systems.

The new ToolJet is built for all of them.

Navaneeth
Founder & CEO, ToolJet