thingsidoatwork.co.uk

What It Is

Things I Do At Work presents a collection of AI-native app and tool projects rather than a single standardized product. The content covers ThingsIDoAtWork Agent Builder, MCP Browser, AI Compare, Home Learning Hub, Save Receipts, AI Stories For Kids, and more, spanning multi-agent LLM workflows, a macOS AI browser, cross-model chat comparison, educational visual recognition, receipt scanning, and children’s story generation.

Core Capabilities

Its AI capabilities are mainly built on mainstream model ecosystems, including ChatGPT, Claude, Gemini, Grok, as well as GPT-4, DALL·E, Google Gemini Vision, Stable Diffusion/ComfyUI, and others. Agent Builder highlights a SwiftUI / Mac Catalyst node canvas for creating multi-agent LLM workflows, with support for MCP tools. MCP Browser turns a native macOS browser into a Model Context Protocol server, offering 40+ tools for AI agents to control a real WKWebView. AI Compare provides a unified chat interface across multiple providers, side-by-side comparison, model fetching, and token usage tracking.

Use Cases and Integrations

For vertical applications, Home Learning Hub uses Gemini Vision to extract curriculum-related skills from photos of children’s homework; Save Receipts uses Gemini Vision to extract merchant names, totals, taxes, currencies, and categories; Tailored Tales uses GPT-4 and DALL·E to generate personalized illustrated children’s stories. On the integration side, the content mentions Claude Desktop, Codex auto-registration, Firebase Functions, StoreKit, OpenCV, ComfyUI, and more, indicating a focus on combining native apps with AI toolchains.

Pricing and Privacy

The publicly available content does not provide complete pricing, free quota, or trial policy details. It only states that Save Receipts has a StoreKit credit system and that the children’s story app includes StoreKit IAP. As a result, its actual value for money cannot be assessed. Privacy information is also insufficient, especially given the involvement of sensitive data such as children’s homework photos and receipt images; there is no disclosure of data storage, encryption, training use, or deletion mechanisms.

Pros, Cons, and Who It’s For

The strengths are its broad project coverage and modern tech stack, making it especially suitable for developers or early adopters interested in native macOS/iOS AI tools, multi-model comparison, MCP, and vision AI applications. The limitations are that the information reads more like a portfolio introduction, with missing details on download access, service SLA, pricing, privacy, and Chinese-language support. Access from China cannot be determined from the content; meanwhile, because it depends on services from OpenAI, Anthropic, Google, xAI, and others, network connectivity, account registration, and payment in mainland China may be uncertain. Alternatives to consider include Dify, LangGraph, Flowise, n8n, Claude Desktop, TypingMind, ComfyUI, and others.