Good answers flow
In-policy prices, promises, and refunds go out in well under a tenth of a second. Customers still get instant answers.
For Retail & Customer Service
Your agents answer customers, quote prices, and handle refunds around the clock. GaaS keeps every answer inside the lines you drew: good answers flow, unusual ones wait for a yes, off-policy ones never go out.
GaaS gives retail operators enforced AI customer service guardrails: every price the agent quotes, promise it makes, and refund it offers is checked against your policy first — allowed, fixed, held for your approval, or blocked, with a receipt for every decision.
| Time | Agent | Action | Verdict | Latency |
|---|---|---|---|---|
| 09:41:02 | cx-agent | issue_refund → $24.99 order_8812 | APPROVE | 38ms |
| 09:41:00 | support-agent | apply_discount → 25% → 15% cap | MODIFY | 28ms |
| 09:40:56 | cx-agent | issue_refund → $180.00 order_7710 | ESCALATE | 50.0s |
| 09:40:54 | support-agent | promise_delivery → unverified_date | BLOCK | 33ms |
| 09:40:50 | pricing-bot | quote_price → sku_3302 | APPROVE | 33ms |
| 09:40:48 | cx-agent | apply_discount → 10% loyalty | APPROVE | 18ms |
| 09:40:44 | support-agent | quote_price → sku_1187 below_floor | BLOCK | 31ms |
| 09:40:42 | cx-agent | issue_refund → $12.50 order_9034 | APPROVE | 36ms |
Simulated feed. Actual decisions include full audit records with hash-chain verification.
The Fear
The agent answers every customer instantly, at 2 p.m. and 2 a.m., and that's why you run it. But an agent that can answer can also quote the wrong price, promise a delivery date you can't hit, or hand out a refund it never should have given — politely, confidently, and in your store's name.
A generous chatbot is a lovely customer experience and a terrible business model. The only thing between the agent and your margin was the prompt you wrote. Hope is not a policy.
The Move
GaaS sits between your agents and the customer. Before a price, promise, or refund goes out, it's checked against your policy:
In-policy prices, promises, and refunds go out in well under a tenth of a second. Customers still get instant answers.
An offer that just needs a tweak — the right discount cap, the required terms — is approved with that fix. The customer never waits.
A refund or discount over your threshold is held for your yes — the customer sees "let me check on that," not a giveaway.
Wrong prices, impossible delivery dates, and against-policy refunds are stopped cold — before the customer ever sees them.
Every decision is kept as a receipt — what was offered, what was held, what was stopped, and the policy that did it.
Your Rules, In Your Words
No rule syntax, no config files. The conversational dashboard is powered by Claude — say the policy out loud, and it holds:
"Refunds over $50 need my approval."
"Discounts max out at 15% unless I say otherwise."
"Never promise a delivery date we can't verify."
The Low-Friction Win
A developer wires the SDK in an afternoon — Python, TypeScript or Java, with plugins for LangChain, OpenAI Agents, CrewAI, Pydantic AI, Vercel AI, Microsoft Agent Framework, and MCP. Payments and messaging context can come from your own systems through a context endpoint (the Stripe and Twilio integrations are built, not yet switched on).
For the first week, GaaS just watches: it shows you exactly which offers it would have held or blocked, while changing nothing.
When the would-have-blocked list looks right, flip to live. From then on, the policy holds on every conversation, day and night.
Pricing
Start free in Shadow Mode, no credit card. There is a free tier, plans from $99 a month, and less than a cent per governed action at scale. Nonprofits, NGOs, and veteran-owned businesses govern free for life.
FAQ
Yes. Every refund, discount, and price the agent offers is checked against your policy before the customer sees it. In-policy offers flow instantly; anything over your thresholds is held for your yes; offers that break the rules are blocked before they're made — and each catch is recorded, so you can see what the agent almost gave away.
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.
No. GaaS is for any operator running agents. Regulated teams get framework mappings; everyone else gets control over what their agents do.
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.
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.
Routine actions clear in well under a tenth of a second. Only high-stakes decisions take longer, and only because you asked them to.
No. GaaS sits outside the agent and needs no model changes and no cooperation from the agent to work.
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. See The Context Dividend.
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.
Start free in Shadow Mode — see exactly which of last week's offers would have been held or blocked, without changing a thing. No credit card, wired in an afternoon.
Not in retail? Find your door at GaaS for AI Operators.