Order The Right Parts, Faster

‍

‍

Transforming CO₂ into Valuable Chemicals

‍

Again.bio, founded in 2020, is on a mission to turn CO₂ waste into sustainable chemical products. After 15 years of R&D, and a successful demonstration plant in Copenhagen, Again is now moving into commercial-scale production.

‍

Now, Again is quickly building out their first-of-its-kind facility in Texas City. With detailed design complete, the project has moved into its next phase: equipment procurement.

‍

Thousands of Parts Required for Construction

‍

To build a new chemical plant, thousands of pieces of equipment need to be selected, purchased, and arrive on time — from simple gaskets to custom reactors. With so many pieces, even the most detail-oriented human can accidentally mis-size a valve, select the wrong material, or forget to add something to the equipment list!

‍

Small errors cause critical path delays, pushing projects back by weeks or months.

‍

In April 2025, Again and Alkali teamed up to show how AI could streamline equipment selection for their Texas City facility.

‍

With ProcessMate, Alkali’s project engineering platform, we used a two-stage approach:

‍

1. P&ID Digitization: With modern vision-language AI models, we created a digital representation of the entire plant, directly from the plant schematics.

‍

2. Automated Part Selection: Using curated vender catalogues, and Alkali's sourcing experience, ProcessMate recommends and configures compatible makes, models, sizes, and materials for key equipment pieces.

P&ID Digitization

‍

‍

Before AI can support the design of a chemical facility, it first needs an AI-compatible representation of the plant. Expert engineers are trained to understand how Process & Instrumentation Diagrams (P&IDs) represent a chemical plant. AI – for now – is not.

‍

‍To help with equipment selection, AI must understand what it's looking at.

‍

ProcessMate ingests P&IDs using a modern vision-language pipeline and a custom domain-specific language to build a precise, digital model of the plant.

‍

As it digitizes the facility, ProcessMate performs a series of automated design checks, for example:

‍

·     Does this line size match the specified fitting?

‍

·     Is this the right material for the component?

‍

At the end of the day, human engineers are responsible for these critical decisions. But who wouldn’t want to have their work double checked? Like the safety of lane-assist when driving – you don’t depend on it, but it can save you!

Equipment Selection

‍

‍

Once P&ID digitization is complete, AI starts scanning through curated manufacturer catalogues, and past Alkali vender notes, to select and configure the right parts for the project.

‍

Based on Alkali’s past vender interactions, ProcessMate provides a range of compatible makes and models for the targeted piece of equipment.

‍

When you confirm the make and model of the part, ProcessMate will automatically configure the size, connection types, and materials. After human approval, the parts are added to the bill of materials. These lists are then sent to venders for rapid quotes.

Outcomes

‍

Alkali’s ProcessMate helped Again select 75 pieces of equipment with their associated make, model, size, actuation, material compositions, and connection type. The result: Again was able to get quotes from venders weeks earlier, and move their project faster.

Closing Notes

‍

Artificial intelligence in real-world infrastructures projects is still in its infancy. Nevertheless, modern AI systems can still contribute through:

‍

·     Design checks: Catch design errors early, big and small. This minimizes slow redesign cycles, or equipment mis-orders, that could derail project schedules and budgets.

‍

·     Creating Request for Quotes: Get RFQs out the door to more venders, faster. Earlier RFQs means you can get the parts you need sooner.

‍

For now, before designs are Issued For Construction, or equipment procured, senior human engineers (either at Alkali or in-house) still need to approve.

‍

‍