The rules leave the prompt
Your policies live in GaaS, not pasted into every system prompt. The context window they used to occupy goes back to the actual task.
The Context Dividend
Governance that pays you back.
Moving governance outside the AI agent cuts the context it spends on governance from about 14,700 tokens per routine governed action to 0–2,470, and on a 200K-token model hands back the 60% of the window self-governance fills after about 28 governed actions. That returned capacity is the Context Dividend.
org_demo · blocked before they cost
Simulated
| Time | Agent | Action | Verdict | Latency |
|---|---|---|---|---|
| 09:41:02 | cx-agent | send_email → customer_2209 | APPROVE | 39ms |
| 09:41:00 | support-agent | issue_refund → duplicate_refund | BLOCK | 28ms |
| 09:40:56 | content-agent | post_to_social → claim_softened | MODIFY | 33ms |
| 09:40:54 | billing-agent | process_refund → $1,450.00 | ESCALATE | 50.7s |
| 09:40:50 | research-agent | query_database → pii_trimmed | MODIFY | 37ms |
| 09:40:48 | ops-agent | push_to_prod → v2.6.0 | APPROVE | 25ms |
| 09:40:44 | cx-agent | send_email → wrong_account | BLOCK | 33ms |
| 09:40:42 | buyer-agent | create_po → $120 supplies | APPROVE | 19ms |
GOVERNING
33,461 decisions
Simulated feed. Actual decisions include full audit records with hash-chain verification.
01 · The Reframe
Governance is not overhead. It is a dividend.
You have been told that adding governance costs you speed and money. When governance lives inside the agent, that is true; the rules ride along in every prompt and the model pays for them on every call. GaaS moves governance outside the agent, and that flips the math: it hands back context and tokens you were spending to keep the agent in line.
02 · The Mechanics
Where the dividend comes from.
Bad actions never run
GaaS blocks a wrong action before it executes, so you stop paying tokens for the retry, the apology, and the cleanup that used to follow it.
The model carries less
The agent no longer reasons about compliance on every call; prompts get lighter, faster, and cheaper, and the model's attention stays on the work.
03 · The Numbers
Guardrails in the prompt vs. GaaS.
Figures published by GaaS on gaas.is; the pricing calculator models your own.
| Per governance cycle | Guardrails in the prompt | GaaS |
|---|---|---|
| Tokens spent governing | About 14,700 per routine governed action | 0 (framework plugin) or about 2,470 (MCP) |
| Context window reclaimed | None | 30% of a 200K window after about 14 governed actions, 60% after about 28 |
| Real-world context on the decision | None | From your own systems, through a context endpoint; 29 vendor integrations built, none switched on yet; missing context is treated as risk |
| At 500,000 governed actions a month | ~$73,000 in self-governed tokens at Opus 5.5 list prices (~$56,000 with prompt caching) | ~7.4 billion tokens returned (plugin), ~6.1 billion (MCP) |
The last row is GaaS's own modeled example; your figures depend on volume, and the calculator computes them.
04 · The Through-Line
Go Deeper
The full white paper.
White Paper · PDF
The Context Dividend
Edition 5, September 2026: what self-governance costs an agent's context, measured against the live system.
White Paper · Web edition
The Context Dividend — Edition 5
The full paper on the web, with every figure tagged by the kind of evidence behind it.
FAQ
Every objection, answered.
Doesn't governance cost me tokens?
The opposite. A self-governing agent spends about 14,700 tokens of context on every routine governed action; with GaaS it spends 0 through a framework plugin (74 when an action is blocked), or about 2,470 through MCP. On a 200K-token model, self-governance fills 30% of the window after about 14 governed actions and 60% after about 28. That returned capacity is the Context Dividend.
I already have prompt guardrails. Why GaaS?
Prompt guardrails live inside the model, get re-read on every call, and can be argued away. GaaS is external and enforced; the agent cannot talk it out of a block.
What is GaaS, in one sentence?
An external layer that checks what your AI agents are about to do and allows, fixes, holds, or blocks it against your rules, keeping a tamper-evident record of every decision.
Will it slow my agents down?
Routine actions clear in well under a tenth of a second. Only high-stakes decisions take longer, and only because you asked them to.
Do I have to change my agent or my model?
No. GaaS sits outside the agent and needs no model changes and no cooperation from the agent to work.
What does it cost?
Start free in Shadow Mode, no card. There is a free tier, then plans from $99 a month, and under a cent per governed action at scale. Nonprofits, NGOs, and veteran-owned businesses govern free for life. See pricing.
Is it hard to set up?
No. Start in Shadow Mode with just an email; it runs the full pipeline on real actions without enforcing anything, so there is zero operational risk. A developer wires the SDK in an afternoon, and you author policies in plain language.
Do I have to be a regulated business?
No. GaaS is for any operator running agents. Regulated teams get framework mappings; everyone else gets control over what their agents do.
Collect the dividend.
Start free in Shadow Mode — real actions, real verdicts, zero enforcement, no credit card. See what your agents are doing, and what they'd get back, before you decide to go live.