Introducing AI Authoring in Tulip

Manufacturers rarely struggle to know what needs to change. The hard part is getting from knowing to running. Someone has to turn what's in the SOP and in people's heads into a working solution, then repeat much of that work for the next line, the next cell, and the next plant.

Tulip has already made this far easier: composability and no-code put solution development in your own team's hands. But the work is still assembled by hand, and it still multiplies with every site, at a time when AI has raised everyone's expectations for speed. When the effort of building and rolling out decides what actually gets built, good ideas wait.

AI Authoring changes where that work starts. Today we're announcing Projects and the Authoring Agent, which join Composable AI Agents, introduced last year, to bring AI beyond editing a single app and into planning, building, and reviewing solutions, while keeping the structure and control you rely on in Tulip. We're also previewing Code-Based Apps for teams that want to build in code.

Plan: start with a project generated with your context

Think about how a new solution usually begins. Someone remembers what's in the SOP, or opens a blank screen and starts typing. Either way, the wrong path can get picked before anything is built, and nobody sees what a change will touch until it touches it.

Projects gives you a better starting line. Come in with a problem, and Projects proposes a plan for your production system using the context you already have: your SOPs and process docs, layouts, business goals, best practices, the apps you've already built, and your enterprise and Tulip libraries. You can steer the plan through chat.

Everything tied to that problem lives on one canvas, mapped to your floor, so you can see what a change will touch before you build it. The whole solution is versioned together, and you decide when it goes live. And when it's time for the next site, it's a deployment instead of a rebuild.

Availability: Entering customer testing.

Build: let the Authoring Agent do the heavy lifting

Once you have a plan, the Authoring Agent can start building it. It's your AI partner in the Tulip Editor, and it works on the apps you've already built too. It handles more complexity than AI Composer and goes well beyond work instructions, so you can spend your time on the last 10% of refinement instead of the first 90% of assembly. It builds on the data model you've already defined, shows you its plan before it acts, and tells you what it changed afterward.

Here's what that looks like in practice:

• Hand it your SOPs, operational data, existing apps, or library examples, and let it work from real context.

• Ask it questions. It can offer answers, recommendations, and troubleshooting help.

• Take over whenever you like. Jump into the Tulip Editor and adjust things yourself.

Availability: Entering customer testing. For now, the Authoring Agent builds apps, with tables, automations, and agents to follow.

Build in code: bring your own IDE and AI coding agent

No-code remains at the center of how you build in Tulip. But some use cases call for more flexibility, like interactive visualizations, complex logic, work instructions that vary across many product configurations, or apps that pull live data from systems like PLM and ERP. For those, Code-Based Apps let you build with your own IDE and AI coding agent, then run the result in Tulip.

You aren’t starting from scratch, either. Code-Based Apps call the connectors, tables, and widgets you already use in Tulip instead of reimplementing them, so passwords, OAuth flows, and hardcoded keys stay out of your app code. The platform enforces site portability, operator attribution, and zero-trust IT security, and it carries your Git pull requests to the shop floor as frozen, fully audited versions designed for GxP environments.

Availability: Previewing now, and we’re accepting design partners.

Review: get a first pass from Composable AI Agents

Building faster is great, but only if review keeps up. Composable AI Agents are custom-built agents that test, document, and evaluate solutions, so reviewers get a first pass done for them and can spend their time on judgment. Tailor each one with your own deployment context and enterprise best practices, and quality stays consistent no matter who built the app.

A few to get you started:

• Validation Guide Generator drafts a GxP validation guide for your apps.

• Solution Evaluator scores a solution on simplicity, focus, human-centricity, and more.

• Table Architecture Reviewer gives you data schema recommendations and answers your schema questions.

These are just starting points. Because agents are composable, you can shape them to match how your organization actually reviews and approves work.

Availability: Open beta.

Composability and no-code put solution building in your team's hands. With AI Authoring, planning, building, and reviewing get faster, so your team can spend more time on what matters most: continuously improving your operations.