Investing like Y Combinator: The famous Silicon Valley accelerator is focusing on these 6 AI trends
Everyone's talking about AI, everyone's doing something with AI, and everyone wants to invest in AI. According to Crunchbase , around €52 billion ($60 billion) flowed into AI startups worldwide in the first quarter of 2025. This represents 53 percent of all funding in that quarter.
A key figure in the AI investment scene: Ivan Landabaso, partner at JME Ventures. The Spanish early-stage VC has invested in, for example, the AI agent startup Kustomer and the AI productivity startup Rauda.AI.
Do you have a contentpass subscription but still want to view external content from third-party providers? Then click "agree" and we will integrate external content and services from selected third-party providers into our offering to improve your user experience. You can view a current list of these third-party providers at any time in the privacy policy (link to privacy). In this context, user profiles (e.g. based on cookie IDs) can also be created and enriched, even outside the EEA. In this case, your consent also includes the transfer of certain personal data to third countries, including the USA, in accordance with Art. 49 (1) (a) GDPR. Further details on data processing can be found in our privacy policy and in the privacy policy, both of which are available at any time in the footer of our offering. You can exercise your consent to the integration of external content at any time in the footer of our offering via the "Revocation Tracking" link.
Landabaso took a closer look at the business models of the AI startups in Y Combinator's Spring batch . Find out which AI trends the American accelerator is focusing on here:
AI teammates who connect to Slack, email, or Jira and process entire task chains instead of traditional dashboards. For example, "Create quarterly KPIs, build slides, inform the sales team."
They retrieve all the necessary data from CRM, BI tool or Drive, process everything and deliver the finished result back – employees no longer have to click back and forth between dozens of dashboards.
Read also
These are AI solutions that were built specifically for a particular industry – for example, medicine, manufacturing, laboratories or law.
They are not a one-size-fits-all solution, but understand the technical terms, know the regulations and rules, and are prepared for typical problems (“edge cases”) in the industry.
Sometimes hardware is also involved – robots or laboratory technology. For example, an AI that writes medical reports knows exactly what's important in a diagnosis – quite unlike a standard chatbot.
Nowadays, anyone can access good AI models. The difference is no longer in the AI itself, but in how well it is controlled, monitored, and secured.
Agent infra startups are building the infrastructure to ensure AI agents actually function in everyday life. Several factors are important.
For example, "routing": The agent decides which AI model or software is best suited for a task. For example, it uses a different AI for math than for text analysis.
Read also
AI memory is also important. The agent remembers previous conversations or tasks so it doesn't start from scratch with every question. Furthermore, it must constantly check whether the agent is delivering good results or making mistakes—this is called "evaluation" in AI terms.
And “reward tuning” means that the agent receives feedback so that it can improve over time, similar to how a human learns from mistakes.
AI can make mistakes (“hallucinate”) – in sensitive areas such as finance, law or IT, this must be identified and prevented early on.
To ensure the reliability of AI results, startups focus on various protection mechanisms. For example, startups use "tracing mechanisms" to monitor what the agent does and why.
Using benchmarks, startups can Regularly check whether the agent is delivering good results. And using "fail-safes," agents can be automatically stopped if something goes wrong.
Startups in this field are building AI agents that automatically handle complex and strictly regulated processes – for example in law, finance, healthcare or purchasing.
AI takes over the entire process: it checks contracts, obtains approvals, initiates negotiations or conducts complete audits.
Many rules and regulations apply in these areas, so agents must work with particular care and transparency. This saves a lot of time and minimizes human error.
These startups use AI agents to automate sales and marketing – everything that brings in new customers.
Read also
The agents perform tasks such as finding potential customers, sending emails and LinkedIn messages, personalizing the onboarding of new users, and adapting campaigns in real time.
The advantage of this business model: The sales pipeline grows without having to constantly hire new people. This allows companies to scale faster and more efficiently.
businessinsider
