SYNTOSHI
AI Agent Powered by On-Chain Data
Syntoshi transforms live blockchain activity into real-time intelligence — tracking de-pegs, hacks, and emerging trends across chains. Powered by Synternet Protocol. Built for alpha.
Start interacting with Syntoshi
Syntoshi needs fuel to function — and that fuel is $SYNT.
To keep the AI agent running and support the Synternet economy, we’re calling on all holders to contribute. For now, each message costs just 5 $SYNT — trivially cheap for alpha access.
FAQ
How long are my purchased credits valid?
Your purchased credits remain valid for the lifetime of the Syntoshi Telegram channel, which we expect to operate for the foreseeable future.
Why isn’t Syntoshi responding, even though I have unused credits?
- Your query might be outside the scope of Syntoshi’s current capabilities.
- The system could be undergoing maintenance.
- If the issue persists, please reach out to the team for support.
Can I get a refund if I purchased more credits than I need?
Credits are non-refundable. However, if there’s an exceptional circumstance, you’re welcome to contact the team and we’ll review it on a case-by-case basis.
Will the price per credit change in the future?
Prices may be adjusted over time based on usage patterns, system costs, and demand.
What is Syntoshi?
Syntoshi is an AI agent deeply integrated with the Synternet Protocol and Synternet Insights. It transforms live on-chain activity into structured intelligence — flagging de-pegs, monitoring collateralization levels, detecting hacks, and surfacing anomalies across chains in real time.
Its core focus is on cross-chain wrapped assets and systemic risk, with rapid signal generation during critical response windows. Using retrieval-augmented generation (RAG), Syntoshi continuously improves its understanding of market behavior, protocol health, and liquidity dynamics.
Change log
v0.3.2: - Reduced message throttling, reducing response times. - More concurrency for agents, which means even faster response times on top. - Loosened message filters (less ignored messages if the topics are very similar) - Misc personality updates
v0.3.1: - A lot less message duplication - Improved (sub-)agents concurrency, ensuring fast and proper data flow - (sub-)agents now will call data sources more often for more accurate data - Minor data ingestion fixes
v0.3.0: - Market Data Integration: Direct access to more market data via Synternet Protocol RPC functionality (two-way communication), providing comprehensive historical data for virtually any token - Price Predictions: Integration with a new two way data stream enables latest and historical price predictions for BTC, SOL, and ETH (5m and 8h timeframes) - Enhanced News Coverage: Improved historical news system queries both more recent (7-day timeframe) news and older historical significant events - Better Message Formatting: Syntoshi can now use formatting for cleaner, more readable Telegram messages - Reduced Repetition: Significantly improved data stream patterns to minimize redundant information - Various Formatting Fixes: Resolved several data presentation issues for clearer communication - Minor personality improvements: Resolved Known issues (which are already being worked on for the next version): - In some cases, Syntoshi may still send duplicate messages - More sub-agents means it takes longer to reach consensus, so Syntoshi might take longer to respond if it has to ingest a lot of data - Some sub-agents do not see your conversation history with Syntoshi, so be sure to either include relevant context if your request needs new data, or use Telegram reply-to feature, which provides previous Syntoshi message as a context for sub-agents.
v0.2.0: - Long term memory of multiple data sources - Ability to query relevant data on demand - Multiple agents behind the scenes to process even more data - More dynamic architecture, eliminating the need of 50% of agent restarts for minor changes - Ability to see full telegram response context, direct responses to agent do not need a mention - Syntoshi can now choose to ignore spam or super irrelevant off-topic discussions (while still being able to have fun)
v0.1.2 - Minor data ingestion fixes
v0.1.1 - Updated personality: - Less quantum and neural - More friendly - Less smirky - Other small improvements - More stream data - Update to bot structure to enable near future improvements