# RightNow AI

> YC-backed GPU research lab. We ship the RightNow CUDA editor and Forge kernel agent, and publish AutoMegaKernel, AutoKernel, StreamIndex, Ouroboros, HCLSM, and TIDE on arXiv. Hardware-aware AI, custom agents with skills and MCP support, stable SSH workflows, and a clear model list across cloud and local GPU models.

## Products

- [RightNow Code Editor](https://www.rightnowai.co): AI-powered IDE for CUDA/GPU development with Nsight Compute integration, GPU emulation, multi-LLM support, and first-class Numba, CUDA Tile, and Mojo workflows.
- [Forge Agent](https://www.rightnowai.co/forge): Swarm agent system using 32 parallel Coder+Judge pairs to turn slow PyTorch into fast CUDA/Triton kernels. Up to 14x faster than torch.compile with 100% numerical correctness.

## Documentation

- [Getting Started](https://www.rightnowai.co/docs/quickstart)
- [Core Features](https://www.rightnowai.co/docs/core-features)
- [GPU Profiling](https://www.rightnowai.co/docs/profiling)
- [GPU Emulation](https://www.rightnowai.co/docs/gpu-emulation)
- [AI Coding](https://www.rightnowai.co/docs/ai-coding)
- [Forge Docs](https://www.rightnowai.co/docs/forge)
- [Forge Commands](https://www.rightnowai.co/docs/forge/commands)

## Downloads

- [Download Page](https://www.rightnowai.co/downloads): Windows, macOS (Apple Silicon + Intel), Linux. Free tier, no credit card required.

## Pricing

- [Pricing Page](https://www.rightnowai.co/pricing): Free ($0/mo), Pro ($29/mo), Enterprise (custom). Forge credits: $15/credit, volume discount at 10+.

## Research

All papers by Jaber Jaber*, Osama Jaber · 2026 · cs.LG. Full list at [rightnowai.co](https://www.rightnowai.co).

- **AutoMegaKernel: A Statically-Checked Agent Harness for Self-Retargeting Megakernel Synthesis** (arXiv:2606.09682)
  - Authors: Jaber Jaber*, Osama Jaber · 2026
  - Agent harness for synthesizing self-retargeting megakernels for LLM inference, with static checks across the generation loop.
  - arXiv: https://arxiv.org/abs/2606.09682

- **AutoKernel: Autonomous GPU Kernel Optimization via Iterative Agent-Driven Search** (arXiv:2603.21331)
  - Authors: Jaber Jaber*, Osama Jaber · 2026
  - Autonomous agent-driven search loop for iteratively optimizing GPU kernels.
  - arXiv: https://arxiv.org/abs/2603.21331
  - Code: https://github.com/RightNow-AI/autokernel

- **StreamIndex: Memory-Bounded Compressed Sparse Attention via Streaming Top-k** (arXiv:2605.02568)
  - Authors: Jaber Jaber*, Osama Jaber · 2026
  - Memory-bounded compressed sparse attention using a streaming top-k index, with a Triton implementation.
  - arXiv: https://arxiv.org/abs/2605.02568
  - Code: https://github.com/RightNow-AI/StreamIndex

- **Ouroboros: Dynamic Weight Generation for Recursive Transformers via Input-Conditioned LoRA Modulation** (arXiv:2604.02051)
  - Authors: Jaber Jaber*, Osama Jaber · 2026
  - Dynamic weight generation for recursive transformers through input-conditioned LoRA modulation.
  - arXiv: https://arxiv.org/abs/2604.02051

- **HCLSM: Hierarchical Causal Latent State Machines for Object-Centric World Modeling** (arXiv:2603.29090)
  - Authors: Jaber Jaber*, Osama Jaber · 2026
  - Hierarchical causal latent state machines for object-centric world modeling.
  - arXiv: https://arxiv.org/abs/2603.29090

- **TIDE: Token-Informed Depth Execution for Per-Token Early Exit in LLM Inference** (arXiv:2603.21365)
  - Authors: Jaber Jaber*, Osama Jaber · 2026
  - Token-informed depth execution enabling per-token early exit for more efficient LLM inference.
  - arXiv: https://arxiv.org/abs/2603.21365

## Credentials

- **Y Combinator**: RightNow AI is backed by Y Combinator. Profile: https://www.ycombinator.com/companies/rightnow
- **NVIDIA Inception Program**: Member.
- **NVIDIA endorsement**: Dave Salvator, Director of Accelerated Computing at NVIDIA, recorded a video testimonial on AI-generated CUDA kernels. Available on the [homepage](https://www.rightnowai.co).
- **Product Hunt**: Product of the Day, Product of the Week, #1 AI Tool of the Day.

## Recent Additions

- New paper on the homepage: **AutoMegaKernel** (arXiv:2606.09682) — statically-checked agent harness for self-retargeting megakernel synthesis.
- New video testimonial from NVIDIA's Dave Salvator on AI-generated CUDA kernels.
- Y Combinator backed badge now displayed on the homepage.

## Guides

- [CUDA Error Guides](https://www.rightnowai.co/guides/cuda-errors): 70+ CUDA error solutions with fixes.
- [CUDA Operations](https://www.rightnowai.co/guides/cuda-operations): 90+ GPU operation guides (matrix multiplication, memory coalescing, reductions).
- [GPU Comparisons](https://www.rightnowai.co/guides/gpu-comparison): 46+ GPU specs and benchmarks (RTX 4090, H100, A100, B200).
- [Framework Guides](https://www.rightnowai.co/guides/frameworks): PyTorch, TensorFlow, JAX, and 13+ framework integrations.

## Blog

- [Blog](https://www.rightnowai.co/blog): Technical articles on GPU optimization and CUDA development.

## Optional

- [Features Overview](https://www.rightnowai.co/features)
- [GPU Profiler](https://www.rightnowai.co/features/gpu-profiler)
- [GPU Emulator](https://www.rightnowai.co/features/gpu-emulator)
- [AI Assistant](https://www.rightnowai.co/features/ai-assistant)
- [Changelog](https://www.rightnowai.co/changelog)
- [Enterprise](https://www.rightnowai.co/enterprises)