One key that proves what an agent is allowed to do.

We compile the rules into a cryptographic key. We call it a mandate. The agent cannot sign outside it, and anyone can verify it with one standard signature check. No onboarding, no platform to join.

// pre-settlement mandate

blisk.mandate.compile({

governs : ["cash", "funds", "collateral"],

actors : ["treasury", "agent_ops_01"],

window : "pre_settlement",

privacy : "private",

})

# authorize → valid instruction + evidence

# fails    → no valid instruction exists

For agent platforms

Your agent negotiates, sources, and commits. The counterparty can't tell if it's real, what it may commit, or whether it will pay, so deals stall in credit desks and prepayment demands. A mandate makes the commitment verifiable in milliseconds.

For institutions

Your controls live in custody consoles, treasury policies, signer lists, and approval queues. A mandate compiles them into one key that execution cannot exceed, enforced before value moves.

tokenized cashfundscollateralpre-settlement controlscontrol evidenceprivate mandatestokenized cashfundscollateralpre-settlement controlscontrol evidenceprivate mandates

Market shift

Money is becoming programmable. Buyers are becoming agents.

Both sides of every transaction are turning into software. The rules that govern them are still paperwork.

01

Agents that commit

Procurement agents award POs. Negotiation bots close deals. Treasury automations move funds. Software now signs for real money.

02

Money that settles itself

Stablecoins, deposit tokens, and tokenized assets create 24/7 settlement legs. Funds, T-bills, receivables, and collateral move as digital records.

03

The missing layer: provable authority

Every rail got faster. Nothing proves what the thing signing is allowed to do, or who pays if it overreached. That layer is the mandate.

Problem

Institutions have controls. They just do not travel as one mandate across the workflow.

Settlement can be atomic, but institutional authority is still stitched together manually.

01

Treasury policy

Risk appetite and approval limits.

02

TMS / ERP

Payment instructions and account rules.

03

Bank / custodian

Wallet, account, and settlement controls.

04

Transfer agent

Eligibility, register, subs, redemptions.

05

Compliance stack

KYC, sanctions, jurisdiction, counterparty.

06

Smart contracts

Public on-chain enforcement.

07

Reports

After-the-fact evidence.

Seven surfaces.
One workflow.
No shared mandate.

Product

BLISK governs whether the agent is allowed to make this commitment at all, before the rail settles it.

The mandate is compiled into authority that can authorize a transaction only when the institution's conditions hold. The sensitive logic stays private; approved parties receive control evidence.

INPUT

Proposed instruction

A transaction an institution, employee, or autonomous workflow wants to send.

BLISK MANDATE

private rules
executable authority
control evidence

fails → no valid instruction

OUTPUT

Settlement rail

Bank rail, tokenized deposit network, custodian, CSD, transfer agent, smart contract, broker, or RWA venue.

BLISK does not replace the bank, custodian, CSD, transfer agent, or chain. It gives them a mandate boundary to consume.

The supplier or counterparty joins nothing. One signature check plus a read of a public registry. That's the whole integration.

Interactive constructor · Demo

Assemble the mandate. Compile it. Delegate it.

Pick the rules, set their parameters, and compile. The result is a single cryptographic key that only authorizes instructions the mandate allows.

Pick and set the rules

Each rule is a live control with parameters, not a checkbox.

Compile into one key
Delegate without losing control

// live mandate expression

mandate :=

access(agent_ops_01

∧ 2of3(treasury)

∧ compliance_officer)

∧ eligible(WL-EU-07)

∧ risk(venues, caps, tenor≤90d)

∧ liquidity(window, buffer)

∧ derisk(oracle → safe_mode)

5 rules set  ·  monotone policy  ·  ready to compile

Capabilities

One private construction carries the controls normally scattered across teams and vendors.

in-keyroles, quorums, revocation
attestedoracle-signed facts as conditions
collapseevents rewire authority, zero interaction

01

Access

Who may initiate, approve, delegate, revoke. Roles and quorums, native to the key.

02

Eligibility

Investor, beneficiary, account, instrument. Proven by attested credentials, not promises.

03

Compliance

KYC, sanctions, jurisdiction, transfer rules. Signed screening results become signing conditions.

04

Risk

Exposure, tenor, venue, counterparty limits. An oracle attests the bound holds; the key requires it.

05

Liquidity

Settlement windows, buffers, cash availability. Time and treasury state enter as attested facts.

06

Collateral

Haircuts, margin, concentration, substitution. Collateral state feeds in via oracle attestation.

07

De-risking

On an attested breach, authority collapses to the safe branch. No meeting, no interaction.

08

Audit

Every signature carries proof a valid rule was satisfied. Show it to auditor, insurer, counterparty.

This is why BLISK is more than policy. It is the executable rulebook for institutional asset movement.

Landscape

The question is not whether someone can enforce a rule. It is whether the full mandate stays private, pre-settlement, and portable.

Approach
Best at
Private
Pre-settle
Portable
Gap
Custodian policy
Account / wallet controls
Yes
Yes
No
Perimeter-bound
Smart contract
On-chain enforcement
No
Yes
Partial
Public and chain-scoped
MPC custody
Key authorization
Yes
Partial
Partial
Protects the key, not the mandate
TEE
Private compute
Yes
Yes
Partial
Hardware attestation trust
BLISK
Private mandate boundary
Yes
Yes
Yes
Requires adapters

Honest scope: BLISK raises bypass cost and shrinks the trusted surface. It is not trustless and does not replace MPC vendors, auditors, or SLAs.

Research foundation

We invented BLISK, an original cryptographic primitive.

The mandate layer is based on a construction authored by our team and published as a public cryptography preprint, with an open reference implementation.

Public research record

IACR ePrint 2026/088

We are onboarding our first design partners.

Work directly with the authors of the construction. Shape the mandate compiler, adapters, and evidence APIs around your workflow.

First design partners get priority and hands-on integration support. No spam.