How SpaceXSI bridged AI with real, local businesses

At 8:30 on a Monday in late October, a team from a footwear manufacturer sat down at Fórum Braga. They had brought one problem from the factory: getting shoe lasts aligned and positioned without doing it by hand.

At the tables around them were people from a cork supplier, a textile mill, a paint factory and a forestry company. Few of them would call themselves developers. But every one of them had brought a real problem from work.

They were there for the SpaceXSI Digital Transformation Lab, a two-day hackathon hosted on TAIKAI. The challenged gathered a total of 79 teams, and by Tuesday afternoon companies had a working AI prototype.

Those prototypes belong to the companies. What SpaceXSI took home, however, is a different story.

Every AI lab wants its models used inside businesses like these: the ones without an AI team of their own. The small and medium sized businesses, that are hard to reach and slow to convince. A sales team can work for months to get one of them to a pilot.

In Braga, SpaceXSI got there with 62 companies in two days. It also left with three things that are just as hard to buy: honest feedback on its product, its name in rooms tech brands rarely enter, and a reputation as the go-to lab for everything AI.

They arrived as users

The footwear team didn’t meet SpaceXSI’s tools in Braga. They had met them the week before.

To take part, every participant set up Cursor, Grok Bot and the xAI console with their work email, ahead of a Sunday deadline. By the time coffee was served on Monday, 218 people from 62 companies were already signed in.

And nobody left empty-handed. Every company walked out with a free month of SpaceXSI’s Enterprise plan and a follow-up with its Enterprise team on implementation, pricing and scale.

Even the prizes kept people on the same stack. The top three teams won 2,000, 1,000 and 500 USD in Grok Build or Cursor credits.

First place went to MAKE IT (a product R&D startup from Ílhavo) for Virtual Datasheet, a tool that turns component datasheet PDFs into searchable data. In 48 hours the team built the app and its search engine, and processed 253 datasheets. The demo ran live on real data, and some people in the room asked if it was a mockup.

Paul Dumaitre, Field Engineer at SpaceXSI, put it simply: “You upload a PDF with charts and technical information and the PDF comes to life.”

That’s how adoption actually starts inside a company: one team, one problem it cares about, and a reason to keep going. A hackathon gives you 62 of those at once.

Into rooms tech brands rarely reach

Riopele makes textiles. Cork Supply works in cork. Fábrica de Tintas 2000 makes paint.

These are not companies that follow AI launch announcements. And that kind of exposure doesn’t come from an ad campaign.

The visibility also outlasts the event. 62 project pages live on TAIKAI, each one a real use case from a sector most AI labs have no stories in yet: a forestry company tracking work per area from telemetry, a metalworks turning sheet offcuts back into stock, a flooring company drafting quotes from past proposals.

The lab that turns up

Plenty of companies say they want to help European businesses adopt AI. Saying it, though, it’s not the same as turning up in Braga on a Monday morning to do it.

SpaceXSI ran the Lab with IAPMEI (Portugal’s agency for SME competitiveness and innovation), Startup Portugal and Invest Braga behind it. That puts a global AI lab next to national and local institutions, working on local companies’ problems.

That builds a reputation no benchmark can: the lab that came and helped. In a new market, that reputation spreads through business associations and word of mouth, places ads don’t reach.

Built before anyone walked in

Most of what made the Lab work was decided weeks before Braga.

The invitation went to companies, not developers, so the people in the room were also the people who could sign a contract. The tracks read like a manager’s to-do list (speed up invoice approvals, stop building the weekly report by hand), so nobody had to guess what AI could do for them. Setup had a hard deadline the Sunday before, with the credits tied to it. And every company, winner or not, left with a trial and a follow-up.

The winners felt it from the other side. MAKE IT’s co-founders, Francisco Mendes and Pedro Magalhães, arrived with the problem already clear, which left the full 48 hours for building. Their comparison: a hackathon is like the Hyperbolic Time Chamber of Dragon Ball, where the normal rules of time don’t apply, and it only works if the team arrives prepared.

None of that depends on Grok or Cursor. Put your own model in the room and it works the same way, in Portugal or anywhere else full of businesses with real problems and no time to solve them.

The footwear team left Braga with AlignCam, a camera system that lines up shoe lasts on its own. SpaceXSI left with a new user, and 61 more like it.

Run yours with TAIKAI

TAIKAI ran the platform side of the Lab: registration, onboarding in two languages, submissions and same-day judging across four rooms. We’ve done the same across 300+ hackathons, with partners like Microsoft and the European Commission.

If your AI lab wants its next enterprise users to come from one room, talk with us and request a proposal.