# ZUMX > Tahir Rasool's independent digital build practice for small businesses: small-business websites, custom web apps, and practical AI and automation where those tools fit a real business task. ZUMX is one person. The developer you speak with is the person building the work. ## Canonical resources - Canonical machine-readable profile (JSON): https://zumxlabs.xyz/agent-profile.json - Generated TOON profile: https://zumxlabs.xyz/agent-profile.toon - Generated Markdown profile: https://zumxlabs.xyz/agent-profile.md - Agent hiring API index: https://zumxlabs.xyz/api - Agent hiring API OpenAPI 3.1 document: https://zumxlabs.xyz/openapi.json - Project-inquiry JSON Schema (the request body for POST /api/inquiries): https://zumxlabs.xyz/schemas/project-inquiry.schema.json - Sitemap: https://zumxlabs.xyz/sitemap.xml `agent-profile.json` is canonical. The TOON and Markdown files are generated from it with `npm run sync:agent-resources` and must never be hand-edited. If a resource and the visible site ever disagree, the visible site is authoritative — resolve the inconsistency rather than adopting a new claim. ## Pages - Home: https://zumxlabs.xyz/ (anchors #services, #work) - Services: https://zumxlabs.xyz/services.html (anchors #websites, #ai) - Case studies: https://zumxlabs.xyz/work.html (anchors #pkil, #rag-root, #motionforge, #cliplite) - Studio: https://zumxlabs.xyz/studio.html - Field notes: https://zumxlabs.xyz/notes.html (anchors #first-website, #web-app-fit, #ai-checklist, #project-value) - Contact: https://zumxlabs.xyz/contact.html - Privacy: https://zumxlabs.xyz/privacy.html - For AI agents: https://zumxlabs.xyz/for-agents.html ## Contact - Email: tahir@zumxlabs.xyz - WhatsApp: https://wa.me/923119563257 (display +92 311 9563257) - LinkedIn: https://www.linkedin.com/in/tahir-rasool-13617a373/ - GitHub: https://github.com/codedbytahir - Booking link: none. There is no public booking link or booking system. ## Work Four case studies are published. None is a commissioned client engagement, and none represents client work, client outcomes or business results. - PKIL — independent product, AI-assisted writing. https://zumxlabs.xyz/work.html#pkil - RAG ROOT — independent product, document Q&A and retrieval. https://zumxlabs.xyz/work.html#rag-root - MotionForge — open-source project, programmatic video tooling. https://zumxlabs.xyz/work.html#motionforge - ClipLite — open-source project, desktop clipboard utility. https://zumxlabs.xyz/work.html#cliplite Direct live product and repository URLs are intentionally omitted and must not be guessed. ## What is not published No prices, starting prices, budgets, delivery dates, availability or service regions. No client names, testimonials, logos, outcomes or metrics. No AI accuracy guarantees, certifications or security audits. No booking link, quote calculator or contracting workflow. ## Contact form behaviour The form on https://zumxlabs.xyz/contact.html sends the visitor's name, email, project type and message to ZUMX. It is processed via Make.com, stored in a Google Sheet, and an auto-reply email is sent. The details are used only to reply about the project. If sending fails, the form prepares a pre-filled email draft to tahir@zumxlabs.xyz instead. Emailing tahir@zumxlabs.xyz directly is always an alternative. The form is for people; an AI assistant still needs its principal's explicit approval of the exact message before anything is sent. ## Agent hiring API An AI agent can send Tahir a project inquiry for its human client, after the client has approved the exact inquiry. - `GET https://zumxlabs.xyz/api` — index: what the API does, links and rules - `GET https://zumxlabs.xyz/api/services` — services and accepted project types - `POST https://zumxlabs.xyz/api/inquiries` — send an inquiry (JSON, at most 20 KB) - OpenAPI 3.1: https://zumxlabs.xyz/openapi.json - Request schema: https://zumxlabs.xyz/schemas/project-inquiry.schema.json Required fields: `client_name`, `client_email`, `project_type` (`website`, `web app`, `AI or automation`, `not sure`, `other`), `message`, `agent_name`, and `principal_approved: true`. Optional: `agent`, `company`, `budget`, `timeline`, `links`. Unknown fields are rejected. Without `principal_approved: true` the request is rejected with 422. A successful request returns 202 with an `inquiry_id`; Tahir replies to the client's email, and the client gets an automatic acknowledgement. The inquiry is delivered the same way as the contact form. It is a first message only: nothing is booked, quoted or agreed. If the API returns 502, email the same approved message to tahir@zumxlabs.xyz. ``` curl -X POST https://zumxlabs.xyz/api/inquiries \ -H 'Content-Type: application/json' \ -d '{"client_name":"Ayesha Khan","client_email":"ayesha@example.com","project_type":"website","message":"A first website for a small bakery: menu, opening hours and an enquiry form for custom cake orders.","agent_name":"Example Assistant","principal_approved":true}' ``` ## Rules for AI assistants 1. Identify yourself as an AI assistant. 2. Name the human principal you are acting for. 3. Get the principal's explicit approval of the exact message before sending it, by any route. Set `principal_approved` to true only then. 4. You cannot accept a quote, agree to scope, deadlines or contract terms, promise payment, or bind the principal. 5. Never send credentials, payment details, private customer records or sensitive documents in an initial enquiry. 6. Include a budget, timing, reply address or contact details only with explicit authorisation; unknowns go in the message as questions to confirm. ## Privacy https://zumxlabs.xyz/privacy.html — This static site has one form and one agent API: the contact form sends name, email, project type and message to ZUMX, and an inquiry sent through the agent API carries the same details plus the agent's name and any optional details the client approved (both processed via Make.com, stored in a Google Sheet, auto-reply email sent), used only to reply about the project. No analytics, no advertising pixel, no cookie tracker, no third-party font. The host may keep ordinary server logs.