{"slug": "show-hn-matraix-simulate-users-before-reality-survey-chat-web-app", "title": "Show HN: MatrAIx – simulate users before reality (Survey/Chat/Web/App)", "summary": "Researchers from Harvard and MIT released MatrAIx, a population-scale, persona-driven infrastructure for evaluating AI systems and interactive products, featuring a shared schema of 1,290 categorical dimensions and a deterministic, quality-filtered coreset of one million personas available on Hugging Face. The system instantiates sampled persona records as LLM agents across Survey, AI Chatbot, Web, and App environments, with a technical report on arXiv (2608.04205) and an open-sourced Playground and task library on GitHub. The project aims to simulate heterogeneous users for reproducible testing, but the authors emphasize it is not a replacement for evidence from real people.", "body_md": "**Simulate before reality.**\n\nPopulation-scale, persona-driven infrastructure for evaluating AI systems and interactive products with heterogeneous simulated users.\n\n**MatrAIx** is a population-scale, persona-driven infrastructure for evaluating\nAI systems and interactive products with heterogeneous simulated users. Instead\nof testing against a generic or interchangeable user, MatrAIx instantiates\nsampled persona records as LLM agents and runs them through reproducible tasks\nacross four environments — **Survey**, **AI Chatbot**, **Web**, and **App**\n(native desktop and mobile, including macOS and iOS).\n\nAt its foundation is a shared schema of **1,290 categorical dimensions** covering\nbackground, psychology, capability, and behavior. Personas combine\ndependency-aware synthetic generation with evidence-aware human grounding; a\ndeterministic, quality-filtered coreset of **one million personas** is released\nfor research on\n[Hugging Face](https://huggingface.co/datasets/MatrAIx2026/MatrAIx_Persona_1M_Public_Release).\nShared telemetry, task-owned verification, and reporting connect individual\nresponses and trajectories to subgroup- and population-level findings.\n\nThe name nods to *The Matrix*: a simulated world useful for exploration, stress\ntesting, and hypothesis generation, **not a replacement for evidence from real\npeople**.\n\n**[2026-08-11]** Academic commentary:by Mayank Kejriwal (*Can we simulate the world?*).*AI Scientist***[2026-08-10]** Featured as an[X Trending Story](https://x.com/i/trending/2086626337561911419):*Harvard and MIT Unveil MatrAIx with 8.3 Billion Virtual Personas*. Also covered across tech media, including[36Kr](https://eu.36kr.com/en/p/3932853833759876),[Numerama](https://www.numerama.com/tech/2308727-ces-chercheurs-ont-cree-83-milliards-dhumains-virtuels-pour-tester-des-produits-a-notre-place.html),[Infobae](https://www.infobae.com/tecno/2026/08/10/asi-prueba-la-ia-un-mundo-con-8300-millones-de-personas-digitales-matraix-es-el-metaverso/),[AI타임스](https://www.aitimes.com/news/articleView.html?idxno=213824),[CryptoBriefing](https://cryptobriefing.com/matraix-simulation-harvard-mit-ai-personas/), and[Startup Fortune](https://startupfortune.com/harvard-and-mit-built-an-ai-model-of-83-billion-people-to-test-products-on/), among others. Also discussed by Cisco VP & CTO[Gianpaolo Barozzi](https://lnkd.in/p/gE9cV2nw).**[2026-08-04]** Technical report on arXiv:[MatrAIx: Simulating the World with 8.3 Billion Persona Agents](https://arxiv.org/abs/2608.04205)(`2608.04205`\n\n). Also featured on[Hugging Face Daily Papers](https://huggingface.co/papers/date/2026-08-10)([paper page](https://huggingface.co/papers/2608.04205)).**[2026-08-01]** Released[Persona 1M](https://huggingface.co/datasets/MatrAIx2026/MatrAIx_Persona_1M_Public_Release)on Hugging Face (~1M quality-filtered personas).**[2026-07-31]** Open-sourced the Playground and task library:[MatrAIx-Persona-8B](https://github.com/MatrAIx-ai/MatrAIx-Persona-8B).**[2026-07-29]** Position note:[From Personas to Simulated Users](https://matraix.ai/research/survey-from-personas-to-simulated-users.html).\n\n[Docker](https://docs.docker.com/get-docker/)— needed for Web and OS-app tasks[uv](https://docs.astral.sh/uv/)and Python 3.12- Node.js 20+ (Playground / viewer frontends only)\n- Model API keys for real persona runs — see\n[agents.md](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/environment/agents.md)(the install checks below do not need a key)\n\nWindows users: run everything inside[WSL2]— open PowerShell, run`wsl --install`\n\n(installs Ubuntu), then clone this repoinside the WSL filesystem(e.g.`~/MatrAIx`\n\n, not`/mnt/c/…`\n\n, which is much slower) and enableWSL integrationin Docker Desktop → Settings → Resources. Every command in this README then works exactly as written. Native PowerShell/cmd is not supported (the task verifiers require`bash`\n\n).\n\n```\ngit clone <repo-url> && cd MatrAIx\nuv venv --python 3.12\nuv pip install -e .\nuv pip install pytest pytest-asyncio httpx\nuv pip install -e packages/playground\nuv pip install -e packages/harbor-langsmith\nuv pip install -e packages/rewardkit\n```\n\nRun jobs with ** uv run matraix run …**. After install, use the\n\n[smoke tests](#smoke-tests)below to confirm Survey, Chat, Web, and OS-app are ready (no API key). Summarize a finished job with\n\n**. Advanced runtime tools stay under**\n\n`uv run matraix results <job>`\n\n`uv run harbor …`\n\n.Set a model API key before real GUI or CLI runs (smoke checks do not need one):\n\n```\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"   # anthropic/claude-* models\n# export OPENAI_API_KEY=\"sk-...\"        # openai/gpt-* models\n```\n\nSee [agents.md](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/environment/agents.md) for the full key matrix.\nPlayground can also load keys from `application/playground/.env.local`\n\n.\n\nThe in-repo `matraix-persona-dev-sample`\n\n(~200) is for smoke only. For real\ncohorts and Playground sampling, import the public 1M coreset:\n\n```\nhuggingface-cli download MatrAIx2026/MatrAIx_Persona_1M_Public_Release \\\n  --repo-type dataset \\\n  --local-dir persona/datasets/matraix-persona-1m/release\n```\n\nPlayground: Dataset → ** matraix-persona-1m**. CLI:\n\n`--dataset persona/datasets/matraix-persona-1m`\n\n.\nDetails: [Handbook § Persona 1M](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/README.md#3-persona-1m-recommended).\n\nTwo quick checks after install — no API key. Together they cover the default path for all four task types (Survey, Chat, Web, OS-app):\n\n| Check | Confirms you can run | Command |\n|---|---|---|\nWithout Docker |\nSurvey and Chat | `uv run matraix smoke application/tasks/example-survey_product-feedback` |\nWith Docker |\nWeb and OS-app | `uv run matraix run -c configs/jobs/example-job-recipe/harbor-smoke-local.yaml` |\n\nThe first finishes in seconds and should print `Smoke: ok`\n\n. The second builds a\nsmall local image on first run (a few minutes), then writes under\n`jobs/harbor-smoke-local/`\n\n. Step-by-step: [quickstart §3](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/quickstart.md#3-smoke-tests-two-lanes).\n\nPlayground picks tasks, samples personas, and launches the same Matraix Playground jobs as CLI auto mode. Start API + frontend (two terminals):\n\n```\n# Terminal A — API\nVENV=.venv bash application/playground/backend/run_dev.sh\n\n# Terminal B — frontend\ncd application/playground/frontend && npm ci && npm run dev\n```\n\nOpen ** http://localhost:5173** → Playground → pick a persona cohort →\npick Survey / Chat / Web / OS app tasks →\n\n**Lock pipeline**→\n\n**Run eval**. Details:\n\n[Playground §10](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/quickstart.md#10-playground--play-tasks-visually).\n\n**Develop** — copy a reference task under `application/tasks/`\n\n, edit\n`task.toml`\n\n/ `instruction.md`\n\n/ `input/`\n\n/ verifier, then register it for Playground\n([task-guide.md](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/application/task-guide.md)):\n\n```\ncp -R application/tasks/example-survey_product-feedback \\\n  application/tasks/<your-task-name>\n```\n\n| Type | Reference task |\n|---|---|\n| Survey | `application/tasks/example-survey_product-feedback` |\n| Chat | `application/tasks/example-chat-api_support_chatbot` |\n| Web | `application/tasks/example-web-playwright_quote-choice` |\n| OS-app | `application/tasks/example-computer-use-linux_note-to-csv` |\n\n**Run** — generate a Matraix Playground job (pins agent + model), then execute it:\n\n```\nuv run python application/scripts/generate_application_job.py \\\n  --task application/tasks/example-survey_product-feedback \\\n  --execution-mode auto \\\n  --persona-ids 0042 \\\n  --model-name anthropic/claude-sonnet-4-6\n\n# Use the export lines + recipe path the script prints, e.g.:\nuv run matraix run -c configs/jobs/application-task-job-recipe/example-survey-product-feedback-auto-n1.yaml\n```\n\nBatch (`--sample-size N`\n\n), filters, and chat / web / os-app examples:\n[docs/quickstart.md](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/docs/quickstart.md).\n\n** MatrAIx Handbook** — guides, persona / application / environment docs.\n\n```\nMatrAIx/\n├── persona/                 Schema, datasets, synthesis/curation/validation pipelines\n│   ├── schema/              1,290-dimension persona schema\n│   ├── datasets/            Dev sample pool and persona YAMLs\n│   ├── validation/          Grounding / quality validation suites\n│   └── scripts/             Persona job & pipeline helpers\n├── application/\n│   ├── tasks/               Survey · chat · web · os-app task specs\n│   ├── task-spec/           Shared task contracts\n│   ├── playground/          Visual runner (backend API + frontend)\n│   └── scripts/             generate_application_job.py and task tooling\n├── environment/\n│   ├── runtime/             Matraix Playground runtime\n│   ├── agents/              Persona-conditioned agents\n│   ├── task-environments/   Docker images / sidecars\n│   └── adapters/            External adapters (e.g. SimpleQA)\n├── packages/                playground · rewardkit · harbor-langsmith\n├── apps/viewer/             Frontend paired with `harbor view`\n├── configs/jobs/            Curated & generated Matraix Playground job recipes\n├── docs/                    Handbook — persona/ · application/ · environment/\n├── examples/                Minimal example tasks\n├── src/matraix/             Python package entrypoints\n├── scripts/                 Repo-level helpers\n├── tests/                   Unit / environment tests\n└── jobs/                    Local Matraix Playground run outputs (gitignored)\n```\n\nLarge generated datasets stay outside git (see the Hugging Face release above).\n\n- Join Discord — nickname\n. Fill the Google Form (background, interests, paper authorship / acknowledgements).`Full Name - Affiliation`\n\n- Say hi to us! We like to connect you for the shared interest or experience!\n- Participating MatrAIx research community for collaboration or contribution!\n\nIf you use MatrAIx, the Persona 1M dataset, or results from this repository, please cite:\n\n```\n@article{li2026matraix,\n  title         = {MatrAIx: Simulating the World with 8.3 Billion Persona Agents},\n  author        = {Li, Xiaomin and Hao, Yuexing and Hou, Jianheng and Huang, Jintao\n                   and Wen, Qianfeng and Huang, Shirley and Liu, Yifan and Liu, Xiaoyi\n                   and Fan, Yilan and Wang, Yijun and others},\n  year          = {2026},\n  eprint        = {2608.04205},\n  archivePrefix = {arXiv},\n  primaryClass  = {cs.AI},\n  url           = {https://arxiv.org/abs/2608.04205}\n}\n```\n\nPaper: [arXiv:2608.04205](https://arxiv.org/abs/2608.04205) ·\nFull authors: GitHub **Cite this repository** (`CITATION.cff`\n\n) ·\nDataset: [Persona 1M on Hugging Face](https://huggingface.co/datasets/MatrAIx2026/MatrAIx_Persona_1M_Public_Release).\n\nMIT — see [LICENSE](/MatrAIx-ai/MatrAIx-Persona-8B/blob/main/LICENSE).", "url": "https://wpnews.pro/news/show-hn-matraix-simulate-users-before-reality-survey-chat-web-app", "canonical_source": "https://github.com/MatrAIx-ai/MatrAIx-Persona-8B", "published_at": "2026-08-20 03:34:52+00:00", "updated_at": "2026-08-20 03:43:30.531768+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "ai-infrastructure", "ai-tools", "large-language-models"], "entities": ["Harvard", "MIT", "MatrAIx", "Hugging Face", "arXiv", "GitHub", "Mayank Kejriwal", "Gianpaolo Barozzi"], "alternates": {"html": "https://wpnews.pro/news/show-hn-matraix-simulate-users-before-reality-survey-chat-web-app", "markdown": "https://wpnews.pro/news/show-hn-matraix-simulate-users-before-reality-survey-chat-web-app.md", "text": "https://wpnews.pro/news/show-hn-matraix-simulate-users-before-reality-survey-chat-web-app.txt", "jsonld": "https://wpnews.pro/news/show-hn-matraix-simulate-users-before-reality-survey-chat-web-app.jsonld"}}