Of the 7 MTONGA steps from Pavel Durov, the first 4 are shipped (Catchain 2.0, fee cuts, Telegram validator, Gram rebrand). The remaining 3 are officially undisclosed, but the public TON Foundation 2026 roadmap names two concrete projects that probably land in steps 5–7: AgenticKit and Teleport.
This piece is a technical deep dive into both, based on public TON Foundation documents, Telegram patents, and dev-team interviews.
AgenticKit: a framework for AI agents on TON
What AgenticKit is
AgenticKit is an SDK + framework for building autonomous AI agents with their own TON wallets. Each agent:
- Has a unique TON address and private key
- Decides based on on-chain and off-chain data
- Executes Gram transactions (send, swap, stake)
- Interacts with other agents (agent-to-agent payments)
- Plugs into Telegram Mini Apps for the UI layer
Tech stack
From public commits and docs:
- Core: TypeScript/Python SDK
- Wallet layer: TON Connect 2.0 for wallet binding
- Execution: agent loop with access to an LLM API (OpenAI/Anthropic/local)
- Memory: long-term + episodic memory (RAG-based)
- Tools: predefined adapters for STON.fi, DeDust, EVAA, Tonkeeper, Telegram-API
Potential use cases
- AI trading agents — automated DEX algotrading on TON
- AI content moderation — channel moderation via AI with Gram payments for work
- AI customer service — autonomous agents accepting Gram payments
- AI NFT artists — generate NFTs, sell on Fragment, distribute profit in Gram
- AI financial advisors — personal AI finance helpers
- AI game NPCs — TON-game NPCs with their own economy
Competition
- Eliza-OS (Solana/Ethereum/Base) — most mature, $1B+ market cap
- Virtuals Protocol (Base) — $400M market cap at 2024 peak
- Olas Network (Gnosis/multi-chain) — $200M market cap
- AgenticKit for TON — under development; potential advantage: 1B Telegram-user distribution
What AgenticKit gives Gram
If AgenticKit takes off:
- Permanent micro-transactions between agents in Gram (Catchain 2.0 sub-second finality makes it feasible)
- Demand for Gram from agent developers (Gram tokens to pay for operations)
- Mainstream crypto onboarding via Telegram Mini Apps with AI
Risks
- AI-agent space is highly speculative with no proven product-market fit as of 2026
- Eliza-OS already captured significant market share; TON may lose first-mover advantage
- Technical complexity (LLM cost, key management for agents, security)
- Regulatory risk (AI-agent-managed money is a novel category for regulators)
Teleport: trust-less Bitcoin bridge
Why Teleport matters
Bitcoin is the largest crypto by market cap (~$1.5T), but BTC holders don’t have an easy way to use BTC in other L1s’ DeFi. Existing solutions:
- WBTC — centralized custodian (BitGo), $5B+ TVL
- tBTC — multi-party custodian, risks at small validator sets
- renBTC — shut down in 2022 after the Alameda collapse
- Native BTC bridges — on specific L1s (Stacks, Liquid Network)
All existing solutions require trust in someone: custodian, federation, multi-sig. Teleport promises a trust-less approach.
How a trust-less BTC bridge works (in theory)
Building blocks:
- BTC light client in TON validators — each TON validator verifies BTC block headers (via partial Merkle proofs)
- ZK proofs — proof that a BTC transaction occurred, without keeping full BTC state in TON
- MPC (multi-party computation) for signing BTC transactions — distributed control of BTC funds with no single point of failure
- Threshold signatures — TON validators collectively sign a BTC withdrawal (2/3+ votes required)
BTC → TON flow
- User sends BTC to a multi-party-controlled address (controlled by TON validators via MPC)
- A ZK proof of the BTC transaction is submitted to TON
- After confirmation, a bridged-BTC jetton mints on TON (e.g.,
bBTC) - User uses bBTC in DeFi (lending, swapping, perps)
TON → BTC flow
- User burns the bBTC jetton on TON
- TON validators collectively sign a BTC transaction (via MPC)
- BTC is sent to the user’s Bitcoin address
- Confirmation in the Bitcoin blockchain
What Teleport gives the TON ecosystem
- Liquidity influx: the $2T BTC ecosystem can push hundreds of millions in BTC deposits into TON DeFi
- New asset classes: BTC perps on Storm Trade, BTC lending on EVAA, BTC LP pools on STON.fi
- Brand positioning: TON becomes one of the few ‘real’ BTC bridges, alongside Stacks, RSK
- Regulatory edge: trust-less = less regulatory risk (no custodian to sue)
Technical risks
- MPC complexity: coordinating 300+ TON validators on BTC signing is challenging
- BTC fee economics: Teleport ops on the BTC side may be expensive during high-fee periods
- Latency: BTC confirmations need ~30–60 minutes for finality
- Smart-contract risk: bridge bugs historically = the biggest hacks in crypto (Ronin $625M, Wormhole $325M, Nomad $190M)
Competition
- tBTC v2 (Threshold) — diminished, ~$200M TVL
- Babylon Chain — BTC restaking for the security of other chains
- Bitcoin Layer 2s: Stacks, RSK, Liquid Network
- Spiderchain — claims trust-less BTC finality
If Teleport is actually trust-less and works, it can lead, because TON has Telegram distribution.
Possible timeline (our estimate)
Base case: both products folded into MTONGA, with shifted timing:
| MTONGA step | Possible product | Date | Confidence |
|---|---|---|---|
| 5 | Premium staking / Telegram integration | July 2026 | 70% |
| 6 | AgenticKit launch | Aug–Sep 2026 | 50% |
| 7 | Teleport mainnet | Oct–Nov 2026 | 40% |
Alternatively, MTONGA may include only product steps (Premium, Stars conversion, ad payments), and AgenticKit/Teleport become separate TON Foundation projects without formal MTONGA tag.
What it means for Gram holders
If AgenticKit takes off
- Demand for Gram from agent developers (to pay for operations)
- Transaction velocity rises
- Possible announcement pump of 30–50%
If Teleport ships
- BTC capital starts flowing into TON DeFi
- TON TVL could grow 2–3× within 6 months after launch
- TON brand shifts from “messenger blockchain” to “settlement layer for BTC”
- Very bullish — but requires successful execution and avoiding bridge exploits
What could go wrong
- AgenticKit may not gain traction (Eliza-OS already captured the market)
- Teleport may slip or face security bugs (history of bridge hacks)
- MTONGA may shift focus to other priorities (Telegram product integration over DeFi innovation)
Sources
- TON Foundation 2026 roadmap (
ton.org/roadmap) - TON Foundation blog posts (April–May 2026)
- Telegram patents filed at USPTO 2025–2026
- TON Connect 2.0 documentation
- Pavel Durov public statements 2024–2026
Further reading:
Top comments (0)