LIVE NEWS
  • ‘Slug sleuth’ farmers in England help develop prediction tool to cut back on pesticide use | Business
  • Ten years on, Brexit still divides Britain and casts a pall over its economy
  • Ethereum’s Jaredfromsubway MEV bot drained after approving its own $7.5M theft
  • ‘Popa’ Botnet Linked to Publicly-Traded Israeli Firm – Krebs on Security
  • Here is how SpaceX could turn up in your current investments : NPR
  • US-Iran delegations arrive as talks begin in Switzerland | Conflict News
  • The business secretary knows about jobs, and seems pretty sure Keir is out of one | John Crace
  • Australia and Europe at the centre of coalition against hybrid threats
Prime Reports
  • Home
  • Popular Now
  • Crypto
  • Cybersecurity
  • Economy
  • Geopolitics
  • Global Markets
  • Politics
  • See More
    • Artificial Intelligence
    • Climate Risks
    • Defense
    • Healthcare Innovation
    • Science
    • Technology
    • World
Prime Reports
  • Home
  • Popular Now
  • Crypto
  • Cybersecurity
  • Economy
  • Geopolitics
  • Global Markets
  • Politics
  • Artificial Intelligence
  • Climate Risks
  • Defense
  • Healthcare Innovation
  • Science
  • Technology
  • World
Home»Artificial Intelligence»What Europe’s AI Startups Are Building For The Enterprise Era
Artificial Intelligence

What Europe’s AI Startups Are Building For The Enterprise Era

primereportsBy primereportsJune 20, 2026No Comments8 Mins Read
Share Facebook Twitter Pinterest LinkedIn Tumblr Reddit Telegram Email
What Europe’s AI Startups Are Building For The Enterprise Era
Share
Facebook Twitter LinkedIn Pinterest Email


What Europe’s AI Startups Are Building For The Enterprise Era

Europe’s AI opportunity is becoming clearer at the enterprise layer, where adoption depends less on abstract model power and more on solving the practical problems that decide whether AI can actually be used.

Enterprise AI is moving into the hard parts

The easy version of enterprise AI is a demo. The hard version is everything that happens after the demo works.

That is where companies run into permissions, legacy software, incomplete data, user interfaces, compliance requirements, integration costs, security risks, and workflows that look simple from the outside but are full of exceptions. This is also where many of the most interesting AI companies are now positioning themselves.

Stay Ahead of the Curve!

Don’t miss out on the latest insights, trends, and analysis in the world of data, technology, and startups. Subscribe to our newsletter and get exclusive content delivered straight to your inbox.

At HumanX, one of the clearest enterprise AI signals came from the companies building around practical bottlenecks rather than abstract hype. For the European companies and founders Elena spoke with, the pattern was clear: they were not trying to copy the biggest U.S. labs. They were building around specific enterprise problems.

H Company focused on computer-use agents for legacy workflows. Malt looked at talent, permissions, and human supervision. Neuralk AI focused on tabular foundation models for company data. Twelve Labs focused on video intelligence as a missing layer of enterprise AI.

Together, they point to a more grounded version of the AI market. The next enterprise wave will not be won only by the biggest general model. It will be won by companies that understand where work actually gets stuck.

The enterprise map

CompanyEnterprise problemAI layer
H CompanyLegacy workflows across old software, disconnected tools, and interfaces without APIsComputer-use agents
MaltMatching talent, agentic work, permissions, and human supervisionAI-enabled talent marketplace
Neuralk AIPrediction from tabular enterprise data without classic ML pipelinesTabular foundation models
Twelve LabsSearching, analyzing, and understanding video at enterprise scaleVideo intelligence / multimodal AI

The legacy software problem

H Company offered perhaps the clearest example of that practical focus.

Its argument is that much of the enterprise world still runs on software that was not designed for agents, APIs, or clean automation. Companies have old systems, disconnected tools, manual processes, and workflows that stretch across Salesforce, SAP, email, PDFs, internal portals, and industry-specific interfaces.

Gautier Cloix described the problem bluntly: humanity is still working on legacy software that does not have APIs and does not have clean data. The traditional answer has been migration. But migration is slow, expensive, and often outdated by the time it is finished.

H Company’s answer is computer use: agents that operate software through the same human interfaces employees already use. Cloix described workflows where a salesperson, customer service agent, purchaser, nurse, or back-office worker completes 40 steps across five or 10 different tools. Instead of rebuilding all the systems underneath, the agent learns to operate the interface above them.

That is why computer use matters. It is not glamorous research. It is the practical problem of clicking, typing, scrolling, reading screens, and moving across systems that were never meant to talk to each other.

In Cloix’s words, the common customer profile is not a specific sector. It is whether a company has ‘a software stack with more than five tools’ and at least one of those tools lacks APIs.

The recent release of Holo3.1 reinforces that direction, with H Company positioning the model family around web, desktop, mobile, and business workflow automation. The broader signal is that computer-use agents are becoming a serious enterprise category, not just a demo of a model operating a browser.

The human layer of agentic work

If H Company is focused on the software interface, Malt is focused on the human and organizational interface.

Claire Lebarz, Malt’s CTO, described the company as Europe’s largest platform for independent experts and freelancers. That gives Malt a specific view of how work changes because freelancers often react to new technology faster than large companies do.

According to Lebarz, talent was already talking about agents before demand had fully caught up. Now Malt is seeing a 600% increase in demand for agentic skills over just three or four months.

That matters because enterprise AI adoption is not only about buying tools. It is about whether companies have people who can translate messy business needs into workflows, supervise agents, and adapt automation to company context.

Lebarz’s most interesting phrase was “humans over the loop.” In her view, tomorrow’s work will involve agents doing more tasks, but humans will still be needed above the process: training, supervising, orchestrating, and adapting agents to real company environments.

That is a useful correction to the usual automation story. The question is not whether agents replace people in a simple one-to-one exchange. The question is how work gets packaged: which parts go to agents, which parts need experts, and which parts require humans who understand the context well enough to supervise several systems at once.

Malt’s perspective also shows why Europe may have a different AI opportunity. The region has deep enterprise clients, talent marketplaces, regulatory awareness, and a workforce transition problem that cannot be solved by hype alone. If agentic work needs trust, permissions, identity, evaluation, and context, then the human layer becomes part of the product.

The data enterprises actually use

Neuralk AI brought the conversation down to one of the most common but under-discussed forms of enterprise data: tables.

The company’s pitch is simple and ambitious. Its founder described Neuralk as doing for tabular data what foundation models did for text. Instead of requiring every customer to build a separate machine-learning pipeline, the company is building foundation models that can make predictions from rows and columns through an API endpoint.

That matters because most enterprises are not operating on clean, internet-scale text. They run on structured data: customers, transactions, inventory, financial records, operational metrics, risk scores, and internal histories. These tables are often the core of the business, but they are not data that can simply be scraped from the web.

The founder explained that tabular data is the core data of every company, which is precisely why companies will not freely give it away. Neuralk’s approach uses synthetic tables during training so the model can learn statistical patterns and then use labeled context samples at inference time to make predictions on client data.

This is a very different enterprise AI problem from chat. It is about statistical inference, prediction, data quality, and deployment without asking every company to maintain the full machinery of classic ML operations.

If it works, it points to an enterprise trend: the AI stack is moving closer to the data structures companies already depend on.

The missing video layer

Twelve Labs added another missing layer: video.

The company started from the view that understanding video is not the same as transcribing dialogue or detecting objects in frames. Video requires temporal understanding, sound, dialogue, scene context, motion, and the ability to decide what matters and what does not.

Its Marengo model powers semantic search across video, image, audio, and language. Pegasus is a video-language model that can analyze scenes, summarize video, generate metadata, and support structured outputs.

That is important because enterprises already sit on massive video archives: studios, sports leagues, news broadcasters, production companies, public-sector organizations, security teams, and data providers. Much of that video is valuable, but hard to search, curate, monetize, or turn into workflows.

The Twelve Labs conversation also connected video to a larger physical-AI story. One representative described video as fundamental for robotics and automotive systems because machines need to make sense of the real world. They described the ambition as becoming a kind of ‘visual cortex for machines.’

That phrase helps connect the company to the broader HumanX theme. Enterprise AI is not only text, code, or databases. It is also visual, temporal, multimodal, and eventually spatial.

Europe’s opportunity is specificity

The shared pattern across these companies is specificity.

H Company is not trying to build a universal chatbot. It is trying to operate legacy software. Malt is not only talking about AI jobs in the abstract. It is looking at how talent, agents, and supervision get packaged for enterprises. Neuralk is not trying to make another general-purpose language model. It is building around tabular data. Twelve Labs is not treating video as a side feature. It is treating video understanding as a foundation layer.

That specificity may be where Europe can compete. The enterprise AI market does not only need bigger models. It needs companies that understand workflows, industry constraints, sensitive data, labor markets, and the last mile between capability and adoption.

That was the more interesting signal from HumanX. The European AI story is not only about whether Europe can produce a frontier lab to rival the U.S. It is also about whether European companies can turn AI into deployable systems for the messy, regulated, operational world enterprises actually live in.

The answer may come less from spectacle and more from the boring places where work really happens: old software, private data, recruitment workflows, video archives, and the humans supervising agents from above the loop.


Featured image credit

Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
Previous ArticleMombasa ocean summit drives marine protection worth $6.4 billion
Next Article Push for electrification finally takes centre stage in pre-Cop31 climate talks | Cop31
primereports
  • Website

Related Posts

Artificial Intelligence

Crawlee for Python: Build a Web Crawling Pipeline with Robots Handling, Link Graphs, and RAG Chunk Export

June 21, 2026
Artificial Intelligence

Telegram Ban Explained: What Happened and How Can You Access in India

June 21, 2026
Artificial Intelligence

Researchers grow a hypothesis tree for AI coding agents

June 21, 2026
Add A Comment
Leave A Reply Cancel Reply

Top Posts

Paxton’s win over Cornyn sets up high-stakes Texas clash with Talarico

May 28, 202616 Views

Global Resources Outlook 2024 | UNEP

December 6, 202510 Views

Texas Democrat Talarico claims voting laws are rigged ahead of Paxton race

May 28, 20269 Views
Stay In Touch
  • Facebook
  • YouTube
  • TikTok
  • WhatsApp
  • Twitter
  • Instagram
Latest Reviews

Subscribe to Updates

Get the latest tech news from FooBar about tech, design and biz.

PrimeReports.org
Independent global news, analysis & insights.

PrimeReports.org brings you in-depth coverage of geopolitics, markets, technology and risk – with context that helps you understand what really matters.

Editorially independent · Opinions are those of the authors and not investment advice.
Facebook X (Twitter) LinkedIn YouTube
Key Sections
  • World
  • Geopolitics
  • Popular Now
  • Artificial Intelligence
  • Cybersecurity
  • Crypto
All Categories
  • Artificial Intelligence
  • Climate Risks
  • Crypto
  • Cybersecurity
  • Defense
  • Economy
  • Geopolitics
  • Global Markets
  • Healthcare Innovation
  • Politics
  • Popular Now
  • Science
  • Technology
  • World
  • About Us
  • Contact Us
  • Privacy Policy
  • Terms & Conditions
  • Disclaimer
  • Cookie Policy
  • DMCA / Copyright Notice
  • Editorial Policy

Sign up for Prime Reports Briefing – essential stories and analysis in your inbox.

By subscribing you agree to our Privacy Policy. You can opt out anytime.
Latest Stories
  • ‘Slug sleuth’ farmers in England help develop prediction tool to cut back on pesticide use | Business
  • Ten years on, Brexit still divides Britain and casts a pall over its economy
  • Ethereum’s Jaredfromsubway MEV bot drained after approving its own $7.5M theft
© 2026 PrimeReports.org. All rights reserved.
Privacy Terms Contact

Type above and press Enter to search. Press Esc to cancel.