Applied AI where it changes the product rather than the pitch: the model plumbing – inference, agents, tooling – and the interface around it, shipped inside products with real users on them.
AI agents & in-product assistants
Agents that act inside the product, not a chat bolted onto it.
Assistants that act on a user's portfolio
Agent flows wired into the product's own actions
An in-product assistant that adapts to the user's behaviour
Portfolio advice and market forecasts
In-the-moment trading coaching
Model plumbing
Inference, agents and tooling – the part behind the interface.
Self-hosted LLMs – Llama or DeepSeek family
A vector base over the product's own data: trades, charts, user activity
Analytics models trained on on-chain data
The interface around the model, built by the same team
Token generation from a prompt
The NOTAI flow: from an idea in chat to a live token.
Pick a category, describe the idea – the flow returns the concept and the visual identity
A no-code ERC-20 or BEP-20 contract written for you
One-tap launch with automatic deployment to Pump.fun on Solana
An AI token launcher inside a trading terminal
Auto-sniping of new launches, in-app swaps with automatic pricing
A PnL dashboard for every token launched
On-chain & social analytics
Analytics on every screen, not a separate tab.
Market trends, AI analytics and asset discovery inside the terminal
On-chain wallet data and top-trader tracking
Social analytics next to on-chain data, split by category
Discovery across 23 chains and 80+ decentralised venues
Asset scoring & verification
The AI layer of an ADGM-regulated RWA platform.
Analysis and verification of what backs an asset
Scoring before capital: AI analytics, then investor validation, then a DAO vote
Insight across 12 asset classes, from real estate to music and film
On-chain data sync for tokenized assets
Modules you add
Each of these runs in production somewhere in the portfolio – switch them on where the product needs them.
Model marketplace & agent hub
A marketplace of tokenized AI models
A launch hub for AI agents
API keys for deploying agents
Fraud detection & audience scoring
Scores that survive contact with bought followers.
AI scoring of KOL and campaign performance
Reward-pool campaigns run against real audiences
Fraud detection on reach and engagement metrics
Scoring and ranking of performers
Data labelling & ZKML
A working annotation network, still running.
Onboarding lessons and skill-based tasking
Leaderboards, approver roles and DAO-driven quality control
On-chain task tracking
Specialist and Expert ranks, with an academy to qualify for harder work
Contributors in 100+ countries
ZKML data labelling – the model the network started on
Telegram superapp delivery
Consumer scale through the app 950M people already have.
A Telegram mini app on TON
Chain abstraction: the Telegram handle is the wallet, no seed phrase
