Mohsin Khan Shaik

Local AI · Python · Windows

HAYAT: a local-first AI assistant for Windows

HAYAT is an open-source AI assistant created by Mohsin Khan Shaik. It brings conversation, voice, public-web research and permissioned computer tools into a shared local runtime.

By · Source review:

Read the source on GitHubGetting started

What problem does HAYAT explore?

HAYAT explores how a desktop assistant can keep its language model and memory on a user's computer while connecting useful tools to one conversation. Instead of treating voice, research and app control as separate assistants, the project documentation describes a common runtime with shared permissions, state and verification.

The repository is the primary source for current setup instructions, feature status and changes. These notes summarize its published architecture; they do not establish that every integration works on every machine.

Creator
Mohsin Khan Shaik
Primary language
Python
Primary platform
Windows desktop
Model interface
Local Ollama models, according to the repository documentation
Project source
mohsinkhanshaik/HAYAT

How the assistant is organized

The architecture document places configuration, providers, conversation, voice, tools, permissions and memory in one runtime. Voice is an input/output adapter rather than an independent agent. The documented action sequence separates interpreting a request from authorizing, executing and checking it.

  1. Request
  2. Plan
  3. Permission check
  4. Tool action
  5. Observe and verify
  6. Record outcome
Responsibilities described in the architecture
ComponentResponsibility
Model and intent routingChoose a response or tool path using configuration, privacy, modality and resource constraints.
Tool registry and permissionsKeep tool execution behind explicit policies and runtime authority checks.
ResearchRead public sources, track evidence and treat retrieved content as untrusted data.
MemoryStore task state and conversation context for later retrieval.
VerificationCheck observed outcomes before reporting an action as completed.

These boundaries are useful engineering ideas to inspect in the source. Documentation describes intended behavior; security and reliability still depend on configuration, implementation and testing.

What is needed to run it?

The README targets Windows 10/11, Python 3.11 or newer and Ollama. Model downloads require substantial disk space, and RAM or GPU requirements vary with the selected model. Consult the current quick start before installing. A smaller model may be more appropriate for a smaller machine.

The project also links a troubleshooting guide and a local diagnostic script. Installation steps and models can change, so the repository remains the setup reference.

Scope and limitations

  • Local model inference does not make every action offline. Web research, downloads and configured connectors use the network.
  • Voice, vision, browser automation and media tools depend on local hardware, installed models and supporting software.
  • The README's test counts and performance claims are not reproduced here because this source review did not rerun the suite or benchmark a desktop installation.
  • The platform focus is Windows. Other operating systems should be checked against the current roadmap rather than assumed supported.

Questions about HAYAT

Who created HAYAT?

HAYAT was created by Mohsin Khan Shaik. The repository credit and licence section names him as its creator.

Does HAYAT require a paid cloud model?

The README describes local Ollama models as the default and a policy that blocks paid fallbacks. Check the selected provider and project configuration; optional network features still access external services.

Where should bugs and project questions go?

Use GitHub Issues for bugs and Discussions for questions and ideas.

Sources and review scope

This page is a source-linked overview, reviewed on November 6, 2026. It is not a new runtime benchmark or certification. The repository may advance after this review.