The platform uses an ideation module that reviews historical lessons to rank new experiments by budget and expected return, while an independent verifier evaluates the outcome. Lessons are stored in a Git-based repository complete with application conditions, evidence trails, and credence scores. Because it is model- and coding-agent-agnostic, the software can automatically deploy successful configurations directly into target environments like Docker, Modal, BentoCloud, or LangGraph.
Key points
Kapso improves enterprise data processing by validating designs through continuous empiri…: Kapso improves enterprise data processing by validating designs through continuous empirical testing rather than pre-planned architectures.
The system prevents repetitive errors by storing accumulated institutional knowledge insi…: The system prevents repetitive errors by storing accumulated institutional knowledge inside an auditable Git-based lesson bank.
Automated deployment capabilities allow empirically verified solutions to transition smoo…: Automated deployment capabilities allow empirically verified solutions to transition smoothly onto infrastructure like Docker or LangGraph.
Sources checked by our analysis pipeline: 1
Hey HN! We're Alireza and Majid building Leeroo (
https://leeroo.com/). Previously we’ve built AI systems for products with millions of users, including at LinkedIn and Meta. We’re now focusing on building AI systems for enterprise data. Today, we're open-sourcing Kapso (
https://github.com/Leeroo-AI/kapso), an engine that runs experiments, deploys the best solution, and learns from every run.
The nature of data work is iterative: investigate, form hypotheses, experiment, and improve based on what you learn. This bottom-up process is the only way to deliver meaningful outcomes quickly in large, messy enterprise data environments. Consultants gathering context in meetings or teams defining a business ontology upfront to fit into the data has no chance of execution, in quarters and sometimes years. An AI-native system should automate this process, compounding learning to each new investigation and implementation.
For instance, one of our enterprise projects was moving a company off its old data estate onto a lakehouse with more than 30 source-system families, thousands of tables and transformations, and hundreds of BI reports. No individual or AI agent knew how that data estate was…
A Google researcher working on an independent project sparked controversy after incorrectly announcing an unmapped archaeological find in Mexico. The location had actually been documented in an academic journal the previous year, prompting domain experts to raise alarms over the lack of proper peer review. The episode highlights the growing friction surrounding amateur technologists applying automated scanning systems to specialized academic fields without collaborating with specialists.
This upgrade expands the platform's core capabilities by automating model transitions seamlessly during active use. By delegating workload management to an automated system, users can leverage specialized model strengths without manually restarting or adjusting their current workflows. Access to these operational features remains reserved for the platform's top tier of users.
The integration of the Gemini 4 Hydrogen model represents a major architectural leap designed to interpret convoluted, multi-part inquiries and natural thought patterns. By executing multi-step agentic workflows significantly faster, the system aims to reduce search friction and surface deeper web results that standard queries often miss. This global rollout modernizes the core search experience while extending high-tier processing to enterprise developer tools.