WISPAL. Monetization engine for creators.

2025 - 2026 · wispal.me · Founder

Wispal is a platform for private, one-to-one paid interaction between creators and their audience: real-time chat, video and audio calls, paid content, and subscriptions, with preference-based matching to connect the two sides. Payments, translation, and AI assistance are built in. The idea first emerged during the pandemic; Wispal is the product it became.

A client sets preferences and gets matched to creators, opens a private thread, and pays per message or per minute of call. Wispal sits in the middle: it holds the money, translates across languages in real time, and settles earnings to the creator. Built and running in production across seven languages.

INSIGHT

85% of people are unhappy with their 9-5 jobs. New generation of experts and creators are looking for more flexible and rewarding ways of work. Plenty of them want to earn online, but the accessible paths are brutal: making content or streaming is a real skill most people lack. One-to-one is something almost anyone can do, yet little exists to monetize it well. The harder wall is money and trust. User-to-user payments are difficult to run compliantly across borders, and neither side moves first: the client will not pay a stranger, the creator will not work unpaid. Wispal sits in the middle, handle payments and matching, and a conversation becomes income.

WHAT I BUILT

• Real-time chat and messaging
• One-to-one video and audio calls
• Per-minute call billing
• Preference-based matching system
• Wallet and payments system
• Paid content and paid conversations
• Automatic message translation
• AI writing assistant
• AI conversation memory
• Content moderation for user content
• Recurring subscriptions
• Referral and affiliate programs

HOW IT'S BUILT

Architecture, split across several repositories:

• Consumer app: a Vue 3 progressive web app for creators and clients
• Backend: a Node.js and Express API with a Socket.IO real-time layer and scheduled jobs, on MongoDB Atlas
• Panel: a separate B2B dashboard for referral and affiliate programs
• Business-context repo and technical-documentation repo: the written source of truth the AI agents build against
• Real-time calls over Daily.co (WebRTC); Firebase for auth and push; KYC and card/payment rails integrated
• AI on Google Gemini via Vertex AI, with DeepL as a translation fallback: writing copilot, conversation memory (vector search), moderation, transcription, and translation

How it is built, an AI-augmented workflow:

• Agents build against the business-context and documentation repos, so they work with full product and domain knowledge, not guesses
• Guardrail hooks hard-block irreversible or dangerous actions and route production changes to explicit approval
• A money-flow sanity suite runs before every deploy, exercising the paths where money moves end to end
• Browser-driven visual checks open the real app as each persona and verify screens, so UI is tested, not assumed
• It started mostly manual; today most of the build runs through this agent workflow

DECISIONS

AI to close real gaps, not for hype. Watching beta testers surfaced concrete problems: long conversations were hard to remember, people did not know what to say, and users wanted to talk to creators who did not speak their language. Each became a feature: conversation memory, an AI writing assistant, automatic translation. The cost is real AI infrastructure, latency, and moderation overhead, which I only took on because each feature answered a problem users actually hit.

Human-verified over AI avatars. The rest of the market races to sell AI avatars you pay to chat with. I bet the opposite: in a world flooded with synthetic everything, a verified human is what carries a premium. Creators must record a voice message live in the browser, not upload a file, to prove a real person is there. The tradeoff is onboarding friction and turning away pure-AI supply, in exchange for trust I can price.

Moderation as a real safety layer. User content and user-to-user payments carry heavy compliance obligations, so every message runs through an AI review that flags risk in layers. Clear violations are blocked automatically; the most serious, anything involving a minor, is blocked and reported to authorities; ambiguous cases go to human review. The cost is false positives and review load, accepted because a single leaked violation is the kind of thing that ends a company.

OUTCOMES

• Raised $35,000 in angel capital to build it at a $1.2M valuation
• Built the full product end to end: app, backend, studio panel, payments, AI, and moderation
• Tested with around 30 creator beta testers, with positive feedback from relevant industry people
• Inbound interest in acquiring the technology

LEARNINGS

• A tool is easier to launch (you win one side) but worth less; a marketplace is worth more but hard to match. Either can win, OnlyFans is a tool that brings no clients, but only if it is 5-10x better or removes a real blocker.
• Whether tool or marketplace, start in one vertical: the needs and the matching are clearer, and you grow from there.
• Marketplaces are cash-hungry to bootstrap. Without investment or a 10x wedge the cold-start economics are brutal, and I have built one before and hit exactly that.

STATUS

Parked. The product is built and the technology works, but active development has stopped.