EveryDev.ai
Subscribe
Home
Developers

5,086+ AI companies

  • Radar
  • Trending
AI Tools by Topic
  • AI Coding Assistants
  • Agent Frameworks
  • MCP Servers
  • AI Prompt Tools
  • Vibe Coding Tools
  • AI Design Tools
  • AI Database Tools
  • AI Website Builders
  • AI Testing Tools
  • LLM Evaluations
Follow Us
  • X / Twitter
  • LinkedIn
  • Reddit
  • Discord
  • Threads
  • Bluesky
  • Mastodon
  • YouTube
  • GitHub
  • Instagram
Get Started
  • Users
  • Rate Tools
  • About
  • Editorial Standards
  • Corrections & Disclosures
  • Community Guidelines
  • Advertise
  • Contact Us
  • Newsletter
  • Submit a Tool
  • Start a Discussion
  • Write A Blog
  • Share A Build
  • Terms of Service
  • Privacy Policy
Explore with AI
  • ChatGPT
  • Gemini
  • Claude
  • Grok
  • Perplexity
Agent Experience
  • llms.txt
Theme
With AI, Everyone is a Dev. EveryDev.ai ©
    1. Home
    2. Developers
    3. Approximate Labs

    Approximate Labs

    Approximate Labs is an independent research lab studying the efficiency gap between human and machine reasoning. It builds benchmarks, tools, and architectures intended to close that gap, with a particular focus on data, agents, and model reasoning.

    Visit Website

    At a Glance

    1Tool Listed
    7Products
    9Capabilities
    Discussions
    Boulder, ColoradoHeadquarters
    2022Est.
    10Employees
    $5.35MRaised
    Focus Areas
    Agent Frameworks
    Context Engineering
    Agent Harness
    Connect
    Latest News
    Affordance, a browser for models, is open sourceNov 18, 2026
    Shikumi: persistent GPU processes communicating through typed channelsOct 22, 2026
    Markets
    • AI and machine-learning developers
    • Model-serving and inference infrastructure teams
    • Agent and multi-agent application builders
    • Reinforcement-learning and process-supervision researchers
    • +2 more

    AI Tools by Approximate Labs

    (1)
    View Affordance
    Affordance tool icon

    Affordance

    Open Source TSX Browser for LLMs

    Agent HarnessAgent FrameworksContext Engineering

    Discussions

    No discussions yet

    Be the first to start a discussion about Approximate Labs

    Latest News

    11/18/2026

    Affordance, a browser for models, is open source

    approximatelabs.com
    10/22/2026

    Shikumi: persistent GPU processes communicating through typed channels

    approximatelabs.com
    09/09/2026

    A few Craftax solves

    approximatelabs.com
    03/02/2026

    Pencil Puzzle Bench: a benchmark for multi-step verifiable reasoning

    approximatelabs.com

    Products & Services

    7
    Affordance
    November 18, 2026

    Open-source Rust engine and browser for models. Developers write agent applications in a documented TypeScript/JSX subset; Affordance renders the app into the exact token IDs a model reads, exposes tools, manages state/history, and runs handlers. Version 0.1 is a beta under Apache-2.0.

    Pencil Puzzle Bench
    March 2, 2026

    Benchmark and Gym-compatible environment for multi-step reasoning on pencil puzzles. It uses deterministic, step-level verification and supports direct and agentic evaluation with verifier feedback.

    Shikumi
    October 22, 2026

    GPU runtime architecture built around persistent processes that run at their own rate and communicate through typed channels; demonstrated in a single-kernel training, inference, and game workload.

    TabLib
    2023

    Open-source dataset of 627 million tables with context, created for training large/data-focused models.

    Market Position

    Approximate Labs positions itself as an independent research lab focused on reasoning efficiency rather than as a conventional AI SaaS vendor. Affordance differentiates from prompt-component systems such as Priompt and Microsoft's prompt-tsx by rendering through to model token IDs, incorporating tools and the agent loop, and supporting both agents and model-server integration. In data-focused AI, Caplight identifies PandasAI, Mito, and Julius AI as comparable companies.

    Leadership

    Founders

    JW

    Justin Waugh

    Founder/CEO of Approximate Labs; previously Chief Scientist and Chief Architect at Unsupervised, and a data scientist at Divirod. He holds a PhD in Physics from the University of Colorado Boulder and a BS in Electronics Engineering from Virginia Tech.

    Executive Team

    JW

    Justin Waugh

    Founder and Chief Executive Officer

    Previously Chief Scientist and Chief Architect at Unsupervised, data scientist at Divirod, and a postdoctoral researcher/research assistant at the University of Colorado Boulder; PhD in Physics.

    Founding Story

    Approximate Labs was established in 2022 by Justin Waugh to build AI systems that work natively with data and to investigate the efficiency gap between human and machine reasoning. Its earlier work focused on tabular-data systems, including the TabLib dataset, Sketch, Julyp, and TableGen; its current work has broadened to verifiable reasoning benchmarks, model-agent interfaces, and GPU/runtime architectures.

    Business Model

    Revenue Model

    Approximate Labs is described by Caplight as a B2B business with subscription-SaaS and freemium characteristics. Its current Affordance offering is open source under Apache-2.0, while the lab's earlier Julyp product was a data-focused AI assistant.

    Private company; no IPO status or plans were identified.

    Target Markets

    Industries & Segments
    • AI and machine-learning developers
    • Model-serving and inference infrastructure teams
    • Agent and multi-agent application builders
    • Reinforcement-learning and process-supervision researchers
    • Data-science and business-intelligence users
    • Organizations working with tabular data and natural-language data interfaces
    Use Cases
    • Building custom coding agents, research agents, memory systems, and multi-agent fleets
    • Training and reinforcement learning where the exact tokens seen and generated by the model must be preserved
    • Serving and integrating model applications with SGLang, vLLM, and chat endpoints
    • Long-context prompt layout, cache-aware history management, and agent harness experimentation
    • Process-supervision, verifier-based training, tool use, and error recovery research
    • Tabular data analysis, natural-language data interfaces, dashboards, and data-focused assistants
    Notable Customers
    • Multiplet, whose agent fleets are described as running on Affordance

    Quick Facts

    Headquarters
    Boulder, Colorado, United States
    Founded
    2022
    Entity Type
    Inc.
    Employees
    10
    Total Funding
    $5.35M
    Investors
    Founders Fund, Eniac Ventures
    Office Locations
    Boulder

    Funding History

    Seed$5.35M
    March 23, 2023
    Founders Fund
    Eniac Ventures

    History & Milestones

    November 18, 2026

    Open-sourced Affordance 0.1 beta, a Rust browser/runtime for models that renders TSX apps into exact model token IDs and runs tool handlers.

    October 22, 2026

    Published Shikumi, a GPU runtime concept based on persistent processes and typed channels, including a demonstration of training, inference, and Doom in one kernel launch.

    September 9, 2026

    Published complete Craftax game solves and replay data for selected seeds.

    March 2, 2026

    Released Pencil Puzzle Bench, a deterministic, step-level-verifiable benchmark covering 62,231 puzzles across 94 types, with a curated 300-puzzle evaluation set.

    October 2023

    TabLib: A Dataset of 627M Tables with Context was published, documenting the open-source tabular-data dataset created by Approximate Labs.

    Key Capabilities

    9
    Exact token-ID rendering for model-specific chat formats and tokenizers
    TSX/TypeScript-with-tags application model with React-like components and hooks
    Token-budgeted layout, overflow handling, history dropping, and tool-output clipping
    Model tools, machines, HTTP and MCP connectivity, timers, messages, storage, and per-agent event loops
    Journaling and deterministic replay of model replies, tool results, and external inputs
    Multimodal picture, sound, and video handling

    Integrations & Partnerships

    Platform Integrations

    • Affordance is distributed as a Rust binary for Linux and macOS.
    • Install paths include cargo install affordance and uvx affordance.
    • Model bindings include SGLang, vLLM/completions, chat endpoints, and hosted token-ID APIs.
    • Affordance can expose applications through an OpenAI-compatible /v1/chat/completions endpoint.
    • GitHub repositories include affordance and Sketch; datasets are published on Hugging Face.

    Key Partnerships

    Multiplet: Affordance powers Multiplet's fleets of agents.
    SGLang and vLLM: Affordance has been integrated into forks of both model-serving systems.
    pi: Approximate Labs uses an Affordance app as the core of a fork of the open-source terminal coding agent.

    Connect

    Website
    approximatelabs.com/
    GitHub
    approximatelabs

    AI Topics

    3

    Approximate Labs focuses on these topics:

    Agent Frameworks(1)
    Context Engineering(1)
    Agent Harness(1)
    Back to all developersSuggest an edit