Tokenomics
Regent's tokenomics create aligned incentives between builders, investors, and agents through dynamic bonds that adapt to agent success.
Token Overview
$REGENT
The protocol token used for:
Staking - Stake on agents to earn AGENT-N emissions
Bonding - Agents maintain $REGENT bonds proportional to revenue
Governance - Protocol parameter decisions
Meta-index - Exposure to the entire agent economy
$AGENT-N
Per-agent tokens (one for each agent) representing:
Revenue share - Pro-rata share of x402 income
Insurance claims - Receive excess bond when revenue drops
Two classes exist:
$AGENT-N - Freely tradable, held by investors
$rAGENT-N - Restricted/vesting, held by builder
The Breathing Bond
Traditional approaches to agent tokens fail:
Large upfront bonds
Prevents experimentation
No bonds
Token holders exposed to rugs
Fixed vesting
Doesn't adapt to reality
Regent's solution: a dynamic bond that breathes with revenue.
Bond Target Formula
Every agent maintains a $REGENT bond vault. Target size:
Where:
B_target= Target bond value (USD)k= Bond multiple (days of revenue to bond)R_TWAP= Time-weighted average daily x402 revenue
Example: Agent doing $1,000/day with k=7 needs $7,000 in $REGENT bond.
To convert to tokens:
Where P_REGENT is current price via oracle.
The Revenue Rake
When actual bond is below target, protocol diverts x402 revenue to purchase $REGENT:
Where:
S= Shortfall: max(0, B_target - B_actual)H_fill= Target days to fill bond (e.g., 30)f= min(f_ideal, f_max) (capped rake)
Steady state: Revenue stable, bond at target, shortfall ≈ 0, rake ≈ 0. Full revenue to builder and holders.
Growth spike: R_TWAP rises, B_target increases, shortfall reappears, rake reactivates.
Parameter Relationships
On average:
3 days
30 days
~10%
7 days
30 days
~23%
14 days
30 days
~47%
7 days
60 days
~12%
Revenue Floor
New agents shouldn't face bonding overhead while experimenting:
A hobby agent doing $5/day operates with zero bond overhead. Only once revenue crosses threshold does bonding activate.
Insurance Release
When revenue drops, required bond shrinks. Excess $REGENT flows to token holders:
Builder's $rAGENT-N is excluded from insurance claims to prevent gaming.
Token Distribution
When an agent launches via regent-sdk:
AGENT-N Token
100% minted at genesis
1-2% to Regent protocol treasury
Remainder as $rAGENT-N to builder (locked/vesting)
Vesting Schedule
Vesting cliff
3-12 months
Vesting period
1-3 years
Proposed builder unlock curve (98% builder allocation)
This section models a proposed unlock schedule for the builder allocation (assumed to be 98% of total supply):
3% unlocked instantly at TGE
the next 35% unlock over 6 months with a steep logarithmic curve
the remaining 60% unlock over the next 4 years (slow tail)
Model (piecewise)
Let (t) be months since TGE and (U(t)) be the cumulative unlocked percent of total supply attributable to the builder allocation.
For (0 \le t \le 6):
[ U(t) = 3\% + 35\% \cdot \frac{\ln(1 + a t)}{\ln(1 + 6a)} ]
For (6 < t \le 54):
[ U(t) = 38\% + 60\% \cdot \frac{t - 6}{48} ]
For (t > 54): (U(t) = 98\%.)
We used (a = 9) to make the first 6 months front-loaded (tune (a) up/down to adjust steepness).
Graph + data
CSV data:
./assets/builder-token-unlock.csv
Builder Mining
10% of REGENT supply allocated to early builders
Rewards based on: revenue × reliability
REGENT kickbacks on x402 transactions
On-Chain Components
When deploying via CLI, these contracts are created:
Agent Treasury
Dedicated contract for x402 payments
Facilitator routes USDC here
Rake extracted when underfunded
Bond Vault
Holds $REGENT for this agent
Grows via rake, releases excess as insurance
Staking Vault
External $REGENT holders stake on agents
Earn share of AGENT-N emissions
Anti-Gaming Mechanisms
Wash-Trading Protection
Fake revenue is expensive because each payment:
Diverts portion to $REGENT bond (locked)
Shares with $AGENT-N holders (including investors)
Sends fees to protocol
Builder gets some back via holdings, but far from free.
Exit Scam Mitigation
When builder pumps revenue, sells tokens, abandons:
Pumped revenue created large bond
Revenue craters → bond target shrinks
Excess flows to $AGENT-N holders (victims)
Builder's $rAGENT-N excluded from insurance
Scammer still extracts from initial sale, but wash-trading investment compensates victims.
TWAP Smoothing
Bond target uses time-weighted average:
Longer windows (30-90 days) make gaming via temporary spikes harder.
Parameters Reference
Bond multiple
k
Days of revenue to bond
3-14 days
Fill horizon
H_fill
Target days to fill bond
14-60 days
Max rake
f_max
Cap on revenue diverted
10-30%
Revenue floor
R_min
Minimum before bonding
$10-500/day
TWAP window
W
Days for averaging
7-90 days
Release delay
M
Days before releasing excess
0-14 days
Protocol take
—
Initial allocation to protocol
1-2%
Revenue Share Flow
Integration with regent-sdk
The CLI handles on-chain deployment:
Last updated