RightNow AI

Description

RightNow AI is a YC-backed GPU research lab that builds CUDA and GPU development tools focused on hardware-aware editing, profiling, emulation, remote GPU workflows, and benchmarked performance analysis. Its products help engineers close the gap between AI models and the hardware they run on, going from model to optimized production inference faster.

The RightNow CUDA editor integrates NVIDIA Nsight Compute CLI for GPU profiling and surfaces real-time performance metrics such as SM efficiency, memory throughput, occupancy, and cache hit rates. It provides AI-powered CUDA code completion with context-aware suggestions, automatic GPU architecture detection, inline kernel profiling with one-click execution, and remote GPU execution via SSH for cloud instances and clusters. It supports NVIDIA GPUs from the GTX 1060 to the H100 and multiple LLM providers including OpenRouter, Ollama, and vLLM.

Beyond its tools, RightNow AI conducts research on model-hardware co-design, GPU kernel optimization, and inference infrastructure, publishing work such as AutoMegaKernel and AutoKernel. It is free to use with unlimited profiling and benchmarking, and available on macOS, with a Pro plan adding GPU emulator access and AI credits.

RightNow AI's Core Features

• CUDA editor with hardware-aware AI code completion

• NVIDIA Nsight Compute integration for GPU profiling

• Real-time metrics: SM efficiency, memory throughput, occupancy, cache hit rates

• Automatic GPU architecture detection

• Inline kernel profiling with one-click execution

• Remote GPU execution over SSH for cloud and clusters

• Multi-LLM provider support (OpenRouter, Ollama, vLLM)

• GPU emulator access on the Pro plan

How to use RightNow AI?

1. 1

   Install the editor: Get the RightNow CUDA editor, available on macOS.

2. 2

   Detect your GPU: Let it automatically detect your GPU architecture, or connect a remote GPU over SSH.

3. 3

   Write CUDA: Use AI-powered, context-aware code completion as you edit kernels.

4. 4

   Profile: Run inline kernel profiling with one-click execution and review real-time metrics.

5. 5

   Optimize: Use benchmarked analysis and, on Pro, the GPU emulator to compare and tune kernels.

RightNow AI's Use Cases

• Optimize CUDA kernels
• Hardware-aware coding
• Remote GPU workflows
• Benchmark and compare

FAQ from RightNow AI

• What is RightNow AI?

  RightNow AI is a YC-backed GPU research lab and software company building CUDA development tools and researching model-hardware co-design and inference infrastructure.

• Which NVIDIA GPUs are supported?

  It supports all NVIDIA GPUs with CUDA Toolkit 11.0-12.5, including RTX 40/30/20 series, GTX 16/10 series, Quadro RTX, Tesla, A100, and H100.

• How much does RightNow AI cost?

  It is free with unlimited profiling and benchmarking. RightNow Pro is $29 per month and adds GPU emulator access, multi-GPU comparison, and 1,000 AI credits per month.

• Can I use it on macOS?

  Yes. It is fully available on macOS (Apple Silicon and Intel), where you can use remote GPUs for free or the built-in GPU emulator.

• Does it support remote GPUs?

  Yes. It supports remote GPU execution via SSH for cloud instances and clusters.

• Which LLM providers does it support?

  It supports OpenRouter (200+ models), Ollama, vLLM, and the managed RightNow Pro option.