ROMA + OML

ROMA and OML are @SentientAGI most advanced technologies that enable Sentient Chat to reach — and even surpass — the level of closed corporate AIs like ChatGPT.

HOW?
To understand this, we need to look at the main problems of open-source AI and how ROMA and OML solve them.

Problems
As we know, open-source AI faces major challenges such as monetization, model theft, pirated use, and low efficiency/functionality (1 model = 1 function).

Solution
Various specialized agents, models, datasets, and tools connect to the GRID ecosystem, which is accessible through Sentient Chat.

ROMA
When you use Sentient Chat, your request can be handled not by a single AI model but by several — sometimes even dozens — of models that can work sequentially or, when possible, in parallel.

This solves the problem of low functionality and efficiency.

OML
→ Piracy
OML solves the problem of pirated use of open AI models through cryptographic control. Each model includes built-in authorization mechanisms — before execution, the system verifies a digital signature confirming that the user has permission to use the model. Without this authorization, the model simply won’t respond.

→ Theft
Stealing another AI model becomes impossible because OML embeds multiple fingerprints (key-response pairs) directly into each AI model. This allows anyone to verify ownership through a special challenge-response test. So far, Sentient has successfully embedded nearly 25,000 fingerprints into an LLM without any performance loss — enough to ensure the model survives fine-tuning, compression, and merging, making theft practically impossible.

→ Monetization
OML records virtually all interactions occurring within Sentient Chat — which user, which models were used, when they were used, and how many requests were made. All this data is stored off-chain, while the key usage records are written into an on-chain ledger.
This ensures that every contributor receives a fair reward, and anyone can verify this fairness by checking the public ledger on the blockchain.

Other factors also contribute to AI agent monetization — you can read more about them here