Questions
Everything teams ask while adopting this.
40 questions, grouped by what you are trying to settle. If you are still working out what Decionis is, start with why it exists — this page assumes you already know and are resolving a specific question.
Getting started
What it takes to try it, and what changes when you do.
- Which Decionis product is right for my team?
- Pick AgentSafe to stop the agent API calls, MCP tool calls and deployments your policy doesn't allow. Pick Commerce Gate to protect your margin on prices, orders, promotions, fulfillment and refunds. Both use the same policy decisions, the same approval step for the person you name, and the same signed records.
- What should I try first?
- Run a live decision in the sandbox with synthetic data. It needs no account, email or card. Pick an action, see what your policy decides and inspect the record. From there you can start free with AgentSafe or Commerce Gate.
- Do I have to replace my existing workflows?
- No. The check sits in front of the action you already take, in your workflow, your systems, with your approvers. It answers one question before execution, should this run, and hands the answer back to whatever asked. You migrate nothing.
- Can I start without changing production?
- You start in Shadow Mode: every action on the workflow you connect is checked, and nothing is blocked. You see what would have been allowed, held for a person or blocked, adjust your policy, then switch on enforcement for one action. You can put the same integration back into Shadow Mode at any time.
- Do I have to write the policy myself?
- No. Your own history is checked and a policy is drafted from what was found, with every limit citing the findings behind it (Walmart in Beta, Business Central in Private beta). Before anything goes live it is replayed on the same records, so you see what it would have held. Then a person you name approves it: run it with your own team, where a second person approves every change, or have it managed for you and give the one approval that releases it (Private beta). You can also start from a policy pack.
- Do my policies stay mine?
- Yes. Your policies are versioned inside your own workspace and are never shared between customers. If your policy can't leave your network, you can run the authority in your own cloud under a commercial licence (Private beta).
- How long does integration actually take?
- Minutes for the API, an SDK, or MCP — one call in front of the action you want gated. Native platform connectors take longer because they follow each marketplace's install and review path, and an enterprise rollout across several systems is measured in weeks, not minutes.
- Can I verify a decision myself?
- Yes. Every verdict produces an Ed25519-signed Decision Dossier, and you can check the signature against our published JWKS offline — no account, and no need to trust our code for the result to mean something.
- Can I try Decionis without an account or a sales call?
- Yes. The sandbox runs a decision on a scenario you pick, in your browser, with no account. When you're ready, start free: Shadow Mode first, then switch on enforcement for one action. No demo or sales call needed.
Protocol
What the decision layer is, and what it returns.
- What is Execution Authority?
- Execution Authority answers one question before an action commits: may this exact action run now, under your policy and the approvals it requires? Your identity system proves who is acting and your guardrails judge content and risk. Execution Authority decides whether the resulting payment, change or API call may commit, and leaves a signed record you can verify.
- How is Decionis different from AI guardrails?
- AI guardrails catch unsafe prompts, outputs and behavior. They don't decide whether the payment, change or API call that follows may commit. With Decionis, your policy decides that before money moves or a record changes, using the identity, guardrail and operational signals you already have.
- Why doesn't approving every AI agent action work?
- Because at volume, approving stops being a decision. Anthropic measured that Claude Code users approve 93% of permission prompts, and shipped classifiers specifically to cut permission fatigue (Anthropic Engineering, 25 March 2026). A separate 40,000-run study by Scale X found people missed 1 in 3 genuine threats while clicking through, and 7% approved every prompt they saw. That leaves two bad escapes — approve everything by hand and get nothing done, or skip permissions and hope. Policy-based gating is the third option: your policy answers the routine calls on its own, and a named person is pulled in only when your rules require one, so the approvals someone does see are rare enough to deserve real attention.
- Who gave Decionis the authority to stop an action?
- You did. There is no policy of ours in the decision. Your policy, written, approved and versioned by your team, with the author barred from approving their own change, is checked against each action as it happens. When your rules require a person, a named person in your organisation approves. Nothing is decided that you didn't already decide, and the record shows which version of your policy was in force.
- How is this different from logic my own engineers could write?
- Any team can write a check inside one service. What is expensive is everything around it: policy written and versioned outside the application, checked the same way on every surface an action can leave from, maker-checker approval, a named person in the loop when the rules demand one, and a signed record anyone can verify afterwards. You get that permission step as a service, not as business logic living inside a single app.
- What does Decionis do?
- With Decionis, you stop the payments, price changes, orders and API calls your policy doesn't allow before they commit. Anything outside your policy waits for the person you name. Everything else runs with a single-use permit for that exact action, and every decision leaves an Ed25519-signed record you can check yourself, without asking us.
- What is the Decionis authority lifecycle?
- The Decionis authority lifecycle is Decision Dossier, Decision Chain, Execution Ledger, Override Dossier, Policy Drift, and Policy Evolution. It turns one certified verdict into durable institutional memory instead of leaving critical decisions scattered across logs, tickets, and screenshots.
- How does Decionis handle overrides and policy drift?
- An override is evidence, not a workaround. The original Decision Dossier stays intact, the exception is linked to it, and repeated overrides show you where your policy should change.
- How is Decionis different from Guardrails AI, Lakera, or LangChain Guardrails?
- Guardrails AI, Lakera, and LangChain Guardrails focus on LLM safety, prompt controls, runtime protections, or agent behavior. Decionis operates one layer closer to business execution: it evaluates institutional policy against intent and signals, then decides whether a real-world action can proceed.
- Does Decionis only work for e-commerce checkout, or can it govern other workflows?
- Checkout and marketplace orders are live in production today, on Walmart, Shopify, WooCommerce and Adobe Commerce, and they are not the limit. The same check sits in front of any action through one API call: your agents' API calls through AgentSafe, deploys through the Govern action, and vendor payments and procurement commitments through the API. Commerce and agents are live today; other workflows start from policy packs on the same check.
- How do I require human approval before an AI agent acts?
- Decide in policy which actions an agent may take alone and which need a person, then check each action before it commits. An action that needs approval waits for a named approver who sees exactly what will run. The approval covers that one action only, and the record shows who gave it. The full answer
Commerce
The production wedge: margin, discounts, inventory.
- How does Decionis stop coupon and discount stacking at checkout?
- Commerce Gate puts a margin floor at checkout that knows your cost basis. When stacked or unauthorized discounts would push an order below the floor, the order is stopped before the checkout token is issued, and you get a signed Decision Dossier. Unlike native Shopify discount rules, it works out net margin against your cost basis, and on Shopify the margin gate fails closed.
- How do I stop selling the same last unit twice?
- Commerce Gate checks an order against your stock before it is acknowledged, and holds an order you can't fill for your review instead of letting it ship late. You decide whether to cancel it, so marketplace cancellations stay your call.
- What is a checkout margin floor or margin guardrail?
- A margin floor is a hard minimum-profit rule checked at the point of sale. With Commerce Gate, each order's net margin is worked out against your cost basis, and an order that breaches the floor is stopped or held (p95 decision compute under 120ms, a target, not an SLA), with a signed Decision Dossier as proof.
- How do I require approval before an AI agent issues a refund?
- Check every refund against your policy before the refund call executes. Refunds within your limit and no larger than what the customer actually paid go through; larger ones wait for a named approver, and a refund above the remaining refundable balance is blocked. Each decision leaves a record naming the rule and the approver. The full answer
Developers
Where you call it from, and what it integrates with.
- How does Decionis fit into a Google Cloud or Vertex AI stack?
- Decionis fits as a thin Vertex AI Extension or ADK tool surface that forwards agent intent into Decionis Protocol for the actual governed decision. Google Cloud and Vertex AI stay in the intelligence and distribution layer while Decionis preserves deterministic policy checks, human approval, kill switches, and audit-grade Decision Dossiers.
- What systems does Decionis integrate with?
- Decionis can sit across Slack, Microsoft Teams, Power Automate, Salesforce, ServiceNow, Google Cloud, Vertex AI, Jira, ERP systems, and API or webhook-driven enterprise workflows while keeping one policy boundary before execution.
- Can Decionis turn existing policy documents into executable policy?
- Yes. The policy draft endpoints take unstructured sources — PDFs, JSON, XML, transaction and audit logs, uploads — plus live policy-source connectors for Jira, Confluence, Notion, Google Drive, Git repositories, and SAP, and return either a reviewable strategy draft or an executable policy pack. Every synthesized rule is validated through the rules compiler before you see it; clauses that cannot become executable predicates come back as review items instead of broken rules, so you review, edit, and finalize rather than transcribe.
- What ERP and commerce systems does Decionis integrate with, and does it support NetSuite?
- You can gate commerce on Walmart, Shopify, WooCommerce and Adobe Commerce, and workflows on Slack, Microsoft Teams, ServiceNow, Salesforce and GitHub. The integrations page lists each connector with its status (Production, Beta, Private beta or Planned) and what installing it takes. For ERP and systems of record there is an SAP connector, and any backend, NetSuite, Dynamics and custom ERPs included, works today through the ERP & webhooks path, which cross-verifies cost basis and inventory. A native NetSuite connector is planned.
- How do I enforce a policy on MCP tool calls?
- Bind each tool to a named action, check its arguments against your policy before the server runs it, and refuse any tool you have not bound. Authentication decides who may connect; a policy check decides what each call may do, and the arguments it checked are the ones that execute. The full answer
Security & compliance
What a reviewer needs before this can be adopted.
- Is Decionis legitimate and trusted for high-stakes execution?
- Yes. Public trust signals include Microsoft ISV Partner status, NVIDIA Inception participation, public docs and quickstart flows, security and governance materials, signed Decision Dossier verification, validation packs, and org-wide and per-decision kill switches.
- Does Decionis allow autonomous execution in regulated workflows?
- Yes. Routine actions run on their own under your policy; the ones it doesn't cover wait for a person or stop, and you can switch an action back to Shadow Mode at any time.
- How do I get my security and compliance team to approve Decionis?
- Decionis publishes the review path so security teams can evaluate it before a diligence cycle: the Trust Center starts with what a reviewer can check directly (Ed25519-signed Decision Dossiers and single-use execution grants) and lists audit status; the Governance Controls page documents the Governor Layer — maker–checker policy approval, RBAC, OIDC/SAML SSO, signed audit trail, and kill switches; the Governor Playbook walks the lifecycle step by step with the SOC 2 / ISO 27001 control each step produces evidence for; and the sandbox generates a real Ed25519-signed Decision Dossier a reviewer can verify without an account. Nothing in Decionis requires enforcement authority to start: Shadow Mode is read-only.
- What records do auditors expect for actions taken by AI agents?
- For each consequential action: exactly what the agent asked to do, which policy version decided it and which rule matched, who approved it and how they were verified, and whether it executed. The record has to be written when the decision is made, be tamper-evident, and be checkable without trusting the vendor that produced it. The full answer
Enterprise & GRC
How it sits beside the platforms you already run.
- Why choose Decionis over ServiceNow IRM or IBM OpenPages?
- Choose Decionis when your team needs confidence before a high-stakes action happens, not only documentation after the fact. ServiceNow IRM and IBM OpenPages are broader GRC platforms, while Decionis is optimized to sit in the execution path and return one signed recommendation before money moves or system state changes.
- What tradeoffs should a buyer understand versus ServiceNow IRM or IBM OpenPages?
- Decionis is a specialized execution authority, not a full replacement for enterprise GRC suites. ServiceNow IRM and IBM OpenPages are broader for program management, control administration, and enterprise risk workflows, while Decionis is narrower and stronger at deterministic pre-execution enforcement with signed audit artifacts.
- Can Decionis keep Microsoft Teams and Power Automate while adding Google Cloud or Vertex AI?
- Yes. Teams and Power Automate can remain workflow surfaces while Google Cloud and Vertex AI handle intelligence. Decionis stays in the execution path, returns one certified verdict, and keeps the Decision Dossier, validation pack, and approval proof model intact across both surfaces.
- Who approves a policy change in Decionis, and how does it fit our GRC program?
- Policy in Decionis is versioned and promoted through a maker–checker path: a material change is an explicit new version with its author, approval path, and effective window preserved, so a change can be proven to have followed the approved path (SOC 2 CC8.1) rather than edited in place. Access is role-based (RBAC) with OIDC/SAML SSO, AI reasoning is separated from execution, and every verdict is an Ed25519-signed Decision Dossier. Decionis is not a full GRC suite by design — it is the deterministic gate that produces signed evidence your existing GRC program (RSA Archer, OneTrust, ServiceNow GRC) consumes. Its records are Ed25519-signed Decision Dossiers, and approved actions run on single-use execution grants; the Trust Center lists audit status.
Proof
How claims on this site are substantiated.
- How do I keep a record of what my team decided without anyone maintaining it?
- Records that depend on someone writing them down decay. Teams abandon architecture decision records within weeks, and the products built on that habit have repeatedly failed — the pattern is well documented. Decionis never asks for the habit: it records a decision as a byproduct of authorizing an action. Every evaluation writes a row and a signed Decision Dossier, and every policy change is a dated, attributed revision that shows what it replaced. Run the gate for a month and you have a decision record whether or not anyone maintained anything.
- Who uses Decionis today, and are there customer case studies?
- Decionis is live in production. On commerce it is on the Shopify App Store with 100+ installs governing thousands of checkout gates. One named company has published a case study: Nevarn, automated on-chain and off-chain trading platform — fail closed by design (affiliated with the Decionis founding team). It is a verifiable production-posture proof — Decionis does not attribute a fabricated margin percentage to any customer.
- How does Decionis prove the margin loss it prevents?
- Governed Execution Value is the monetary value of the actions checked before execution, including the ones blocked or held, computed from live governed decisions. When the sample is too small to publish, you see a status instead of a number, so the figure is never made up. The full method is on the prevented-loss methodology page.
Still not answered?
The fastest answer is usually the product itself: the sandbox returns a real signed verdict without an account, and Shadow Mode answers “what would this have done to my traffic?” better than any paragraph here can.