Chat Matt Rickard 
AIM by Matt Rickard is an open-source browser experiment that runs an older language model locally through WebGPU, storing conversations in local browser storage. Developers should check hardware compatibility, understand the model is dated, protect local data, verify code and outputs and avoid production reliance.
Starting price Free
Tool Information
AIM by Matt Rickard is an open-source browser experiment that runs an older language model locally through WebGPU, storing conversations in local browser storage. Developers should check hardware compatibility, understand the model is dated, protect local data, verify code and outputs and avoid production reliance.
Use authorized inputs and least-privilege integrations. Configure privacy, retention, sharing, accessibility, disclosure, export, moderation and spending controls. Test representative cases, verify facts, calculations, citations, messages and media, preserve originals and retain accountable human approval before publication, outreach, deployment or operational action.
The MIT-licensed web experiment is available free and displays no paid plan. A compatible desktop Chrome browser and GPU are required; hosting and browser support may change.
AI output can be inaccurate, biased, derivative, insecure, incomplete or misleading. Review consent, copyright, training and retention terms, renewals, refunds, platform rules and applicable law. Education, health, finance, legal, marketing and customer-facing workflows require qualified human review.
F.A.Q (3)
Pros and Cons
Pros
- Runs a language model entirely in the browser
- Uses WebGPU for local inference
- States that conversation data is not sent to a server
- Caches conversations in browser local storage
- Caches the downloaded model after first use
- Can work without repeated model downloads
- Uses the Vicuna 13B model
- Uses Apache TVM and the MLC Relax runtime
- Provides a nostalgic instant-messenger interface
- Lets users create multiple conversations
- Lets users edit the system prompt
- Lets users name a conversation
- Provides settings and conversation controls
- Chat Matt Rickard is open source on GitHub
- Uses an MIT-licensed application codebase
- Demonstrates private client-side LLM inference
- Can serve as an educational example for WebGPU and browser AI
- Avoids per-message API fees
- Lets users customize persona behavior through a visible system prompt
- Provides source links to the model and code
Cons
- The application explicitly targets Desktop Google Chrome 113 and may not work in other browsers or mobile devices
- Vicuna 13B is an older model with limited current factual and reasoning quality
- The model download is large and can take several minutes
- Local inference requires substantial memory and a compatible GPU
- WebGPU drivers can crash; hang; or consume significant power
- Browser local storage is not encrypted by the application
- Anyone with access to the browser profile may read cached conversations
- Clearing site data can erase the model cache and chat history
- Private browser execution does not protect against malicious extensions; screen capture; or compromised devices
- No server processing also means no cloud backup or cross-device synchronization
- The model can hallucinate; produce unsafe content; and follow malicious prompts
- Vicuna's training data and license conditions need review for commercial use
- The nostalgic SmartestChild persona intentionally uses lowercase; emojis; and old internet abbreviations; which is unsuitable for many tasks
- No current-web search or retrieval is built into the visible interface
- Performance varies dramatically across Macs; Windows PCs; GPUs; and browser versions
- The page appears to be a technical demo rather than a supported production service
- There is no service-level agreement or enterprise support
- The system prompt can be overridden or produce unexpected behavior
- Local models still expose information typed into them through browser memory and logs on the device
- The application cannot match current frontier-model capabilities
- Users must verify whether the WebGPU implementation receives security updates
- A model cached by an old browser version can become incompatible after updates
- The name Chat Matt Rickard may suggest chatting with a person; but the visible persona is an AI named SmartestChild
- The tool should not be used for high-stakes advice
- Users should export or copy any important conversation before clearing browser data
Reviews
You must be logged in to submit a review.
No reviews yet. Be the first to review!
Before you subscribe
- Check the currency, billing period, automatic renewal, usage limits and seats at checkout — prices and promotions change.
- App-store pricing can differ from website pricing by country, device and tax.
- Share only authorised data, and read the provider's encryption, retention and model-training terms first.
- AI output can be incomplete or wrong. Verify facts, citations, calculations and code before you publish or ship.
- Commercial usage rights vary by plan, provider terms and jurisdiction.
- None of these tools replaces a qualified professional for medical, legal or financial decisions.
Promote
Share this Tool
Similar Tools
Chat Thing creates support agents trained on websites, documents, FAQs and connected sources, with tool calling, multiple channels, analytics and human takeover. Operators should validate sources, restrict actions, protect customer data, monitor conversations and retain accountable escalation.
August 20, 2026
Free + from $14/mo
Chatworm is an open-source web, Android and Windows client for directly accessing multiple model providers, switching models, voice conversations, image generation and image analysis. Users should secure API keys, review provider privacy and spend, verify outputs and isolate untrusted files or generated code.
August 20, 2026
Free + API costs
Chat.AI offers general AI conversations for writing, editing, education, software development, data cleanup and translation, with personal and shared team interactions. Users should verify outputs, protect confidential prompts, inspect sharing settings and understand that it is a third-party service using external models.
August 20, 2026
Paid plans