🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

YC W24's Roe AI is a new data warehouse that specializes in the kind of unstructured data that other data warehouses can't handle. They make complex data analysis quick and easy with natural language. ycombinator.com/launches/KRH-r… Congrats on the launch, Richard Meng &

🚀 Big News! Roe AI (getroe.ai) launches VolansDB! 🎉 Dive into advanced unstructured data analytics with our customizable AI Data Agents & VolansDB 📊 🔹 Analyze documents, videos, & more with SQL 🔹 AI-powered data processing 🔹 Multi-modality semantic search

One Y Combinator company previously spent days scraping Google Maps to find all 1-3 location restaurants with decent Google reviews. Now, they can do it within five minutes using Openmart. Get yourself a demo at openmart.ai

x.com/i/article/1841…

x.com/i/article/1846…

Announcing new open-source Python package: aisuite! This makes it easy for developers to use large language models from multiple providers. When building applications I found it a hassle to integrate with multiple providers. Aisuite lets you pick a "provider:model" just by

Today OpenAI announced o3, its next-gen reasoning model. We've worked with OpenAI to test it on ARC-AGI, and we believe it represents a significant breakthrough in getting AI to adapt to novel tasks. It scores 75.7% on the semi-private eval in low-compute mode (for $20 per task

Future of AI tooling: instead of asking for user prompts, infering intentions based on the context and provide answers directly.

For early stage startups, no one shall replace founders’ responsibility to translate business problems into solutions. Not an AE, not an engineer, not even the customers themselves. The customers has the problem, but it’s on us to find out the solution. This is the path to find

Since the very beginning of Roe AI, we have always been thinking about, when foundation models become more and more powerful, what values can we provide in addition to that? Pure document parsers/extractors have no future in competition with model advancement. The value and power

OpenAI just released a powerful deep research. But hallucinations remain a challenge. If it’s just numbers, the errors might be easier to spot. But as AI tackles more complex queries, verifying its outputs becomes harder. What happens when the human user lacks the expertise to

Roe AI is hiring founding engineers in San Mateo, CA! 🚀 Join us in building the next-gen data warehouse that powers enterprise-level, multi-modal AI insights. If you’re experienced in Generative AI, full-stack development (Next.js/Django/FastAPI), infrastructure (AWS, k8s) and

Another avenue of opportunity brought by LLMs is controlled subjectivity. Each person form opinions from their life experiences which they cannot control and humans sometimes have strong emotions. However, we can control what information are used to train/fine-tune AI models. By

scaling from 0 to 1 versus 1 to 100 requires a completely different sales skill set—one wrong hire or founders not figuring it out themselves can lead to disaster and burn $$$ later down the road

This is a fantastic post from Anthropic that should inform folks on what a production-grade multi-agent architecture looks like. There are three points worth calling out: 1️⃣ Not every use cases is suitable for multi-agents: “Further, some domains that require all agents to

Wine and SaaS Kathryn Wu Richard He Shuangling Yin Jason Wang Jisong Roe AI Openmart camelAI (YC W24) InspectMind AI (YC W24) BeFreed