Wayist — Agents & authority
AGENTS & AUTHORITY

Agents with real power, inside hard limits

Create an agent by describing the job. Each one carries a matrix of what it may create, read, update and delete, enforced as a strict subset of its parent's reach, and every action is on the record.

Get started free See pricing
REFUND TRACKER
CREATE
READ
UPDATE
DELETE
Humans
✓
✓
✓
✓
Gaia workspace
✓
✓
✓
—
Plutus finance
✓
✓
✓
—
Athena research
—
✓
—
—
Set per person, per role, per agent — and per base, channel and app.

Access control with no ceiling on how precise it gets

Every user and every agent carries a matrix of what they may create, read, update and delete, on every surface in the workspace. The combinations are yours; nothing is reachable that the grid does not allow.

Per person, per role, per agent, per base, per channel, per app
Delegation only flows down: a child is a strict subset of its parent
New agents stay in draft until a person approves them
Changing the hierarchy is itself a proposal somebody approves

An org chart of agents with people at the top

Describe an agent and it appears on the canvas as a draft. Gaia coordinates sub-agents that each run their own workflows, with the scope and spend caps you set, and every handoff between them is logged.

Spend caps per agent, with approvals above them
Skills, tools, policies and knowledge set per agent
Escalation always ends at a human
Agent hierarchy1 change proposed
Kate Sullivan owner
Gaia workspace
Hestia
support · capped $200/mo
Plutus
finance · approvals on
Athena
research · read only
Every handoff logged · escalation ends at Kate

What governs an agent

The matrix

Create, read, update and delete, set per surface and per actor.

Spend caps

A ceiling per agent, and approvals for anything above it.

Policies

Company and agent documents an agent must follow.

Knowledge base

What it reads before it drafts anything outward-facing.

Skills and tools

Upload a skill, grant a tool, revoke it later.

Audit trail

Every action, handoff and approval, written once and never edited.

Plutusthinking · 14sStop
Kate: which refunds have been held more than five days?
✓ Read 🧾 refund tracker · 28 rows this week
✓ Found 3 held more than 5 days · $3,620.55
✓ Checked all 3 against the processor
◌ Drafting the note to finance
Anyone who can post in the stream can stop an agent working there.

You can watch the work happen

An agent working on something says what it is doing right now, how long it has been at it, and what each finished step found. Anyone in the stream can stop it, and a stopped run keeps its steps.

Live step list while it thinks, not a spinner
Stop mid-run; the record of what it did survives
Try again resumes from the ask, not from the beginning
Every run priced, so cost traces to the step that caused it
COGNITION SETTINGS

Choose the mind behind each kind of thinking

Authority decides what an agent may touch. Cognition decides how well it thinks about it, and what that costs.

ClaudeGPTGeminiDeepSeekKimiGLM
Routing and taggingAffordable
Customer repliesValue
Reasoning and orchestrationExpensive
Document and image readingValue
Code and scriptsExpensive
The same model is a bargain in one role and an extravagance in another, so the tier belongs to the pairing, not to the model.

Eighteen jobs filled one at a time

A workspace does not pick one model and use it for everything. Cognition is a list of named roles — routing, retrieval, perception, generation, reasoning, governance — and each offers a short, sensible list rather than the whole catalogue.

A workspace default, with per-agent overrides on single roles
Options filed as Affordable, Value or Expensive for that role
Swap a model at any time without rebuilding the agent
Rates, limits and meters listed on every model page

Metered and prepaid with a hard ceiling

The subscription buys seats; agent work is metered and varies by the hour. It runs on prepaid credit, and the balance tells you how many days it lasts at the rate you are actually spending.

Billed at each model's own published rate, with nothing added by us
Auto top-up carries a monthly ceiling and a cooldown between charges
Cost held per workflow step, so a spike has an address
AI USAGE CREDIT
$4,280balance
About 26 days at the current rate
Auto top-up$500 at $1,000
Monthly ceiling$3,000
Cooldown between charges24 hours
At the ceilingNotify, do not charge
Hitting zero is an outage, not an invoice — so the balance is shown in days, not just dollars.
IN THE BOX
Create an agent by describing it
Per-surface authority matrix
Strict-subset delegation
Draft until approved
Spend caps per agent
Approval workflows
Agent hierarchies
Policies and knowledge base
Skills and tools per agent
Design system per agent
Full audit log
Escalation to a person
18 cognition roles
Multiple model providers
Bring your own keys
Per-agent model overrides
Prepaid AI credit with a ceiling

Common questions

Do agents take a seat?

No. Agents are part of the plan and you pay for the people who log in.

What stops an agent going too far?

Three things: the matrix that bounds what it can touch, the spend cap that bounds what it can cost, and the approval step before anything irreversible.

Can an agent create another agent?

It can propose one. The proposal carries a permission set that cannot exceed its own, and a person approves it before it runs.

WORKS WITH
Omnichannel inbox Team collaboration Video calls & recordings Bases Mini apps Team mail and calendars Data warehouse Analytics Docs, decks & drive

Give an agent the job and the boundary at the same time

Start with one agent, one base and read-only access. Widen it when it has earned it.

Get started free Compare plans