Tenant controls
Scoped customers, teams, and environments so one path does not bleed into another.
Environment and usage
What you get
with Wakehold.
The Wakehold environment is how your team defines work, holds evidence, and sees what is complete — tenant controls, workflow definitions, evidence rules, visibility, support, and admin. Early access includes that workspace plus a usage pool. Each pack is included capacity for completed control-plane events. Events past the pool are additional usage.
Checkout uses Paddle (Merchant of Record). Buyer-facing charges show as Wakehold. If checkout does not complete, email [email protected] and we will set up your environment.
The Wakehold environment
Workflow definitions
The paths your work actually follows, named by your team.
Evidence rules
What must be on record before a milestone counts as complete.
Visibility
What is complete, waiting, or missing across the systems where the work happens.
Support
Help standing up the first paths and reading the record.
Admin
Owners, access, and workspace setup for the people who run it.
Usage
Tokens are prepaid credits for completed control-plane events. The default is one token per completed event. Model and LLM inference stay separate — bring your own models.
10,000 tokens ≈ 2,500 four-event control sequences.
A four-event sequence is one wake, one receipt, one packet, and one capsule. Additional events burn as later milestones, receipts, or handoffs complete. 50,000 tokens cover about 12,500 of those sequences. 250,000 cover about 62,500.
Spark
$49
10,000 tokens
- Wakehold environment included
- About 2,500 four-event sequences
- Later milestones burn additional events
Current
$199
50,000 tokens
- Wakehold environment included
- About 12,500 four-event sequences
- Room for several paths at once
Surge
$799
250,000 tokens
- Wakehold environment included
- About 62,500 four-event sequences
- Capacity for many owners and agents
Enterprise environment
The same environment — tenant controls, workflow definitions, evidence rules, visibility, support, and admin — with a custom usage pool, invoice billing, and a named contact.
What a token
records.
A token meters a completed control-plane event, not a seat and not a model call. Default burn is 1 token per completed event. The pack is included capacity; further completed events are additional usage.
Wake processed
Wake continuity
A durable wake is accepted, watermarked, and routed — including catch-up after a flapping wake path.
Proof receipt sealed
Evidence of done
A receipt is sealed with message, tracker, or system ids so a finished step is backed by proof you can inspect.
Work packet transferred
Owned handoff
Context, intent, and expected outcome move between agents or lanes as an owned work packet.
Policy capsule injected
Rules at decision time
A versioned policy capsule is injected when a decision is due — guardrails that travel with the work.
Consumption
rules.
Workflow example
Project Atlas on How it works shows one four-event sequence: wake, receipt, packet, and capsule. Later milestones, receipts, and handoffs are additional events. That is why 10,000 tokens cover about 2,500 of those four-event sequences, not 10,000 full onboardings.
Retries
A retry that successfully re-processes an event burns again. Attempts that never complete a burn event do not consume tokens.
Zero balance
Events that require tokens pause until you top up. Existing records and sealed proof remain readable.
Validity
Unused tokens don’t expire for 12 months from purchase. Beyond that, contact us. The Terms of Service and Paddle purchase terms are the commercial record.
Model costs
Wakehold Tokens do not include model or LLM inference. Bring your own models and runtimes; tokens meter Wakehold control-plane events only.
Checkout
Self-serve packs check out through Paddle. The same environment-and-usage story is on How it works.