Solutions/Use cases

Nine places where AI policy has to hold

Every use case below uses the same controls and produces the same evidence. Pick the one your security review is stuck on, and start there.

Govern production AI agents Protect sensitive data Control models and providers Govern MCP tools Enforce organisation-wide AI policies Build a live vendor and processor register Prepare evidence for audits and reviews Control AI spend and usage Extend governance to coding assistants
01 Agents

Govern production AI agents

An agent that can act is an actor in your organisation. It needs an owner, a risk rating, a permitted set of tools and a record of what it did.

Where it breaks today
Nobody can name every agent running in production, or who owns it.
Agent permissions live in application code, so changing them needs a deploy.
When an agent takes an action it should not have, there is no single record of it.
What obstruo does
Applies policy by agent identity, so the same key can behave differently per agent.
Restricts each agent to a permitted set of models, tools and actions, changed without a deploy.
Keeps an evidence record per agent: risk rating, owner, active policies, blocked attempts.
Agent identity Tool permissions Risk rating Incidents
Agent registry PRIVATE PREVIEW
trading-analystHigh4 of 11 tools
claims-triageHigh2 of 11 tools
support-copilotLimited6 of 11 tools
kyc-reviewerLimited3 of 11 tools
Last blocked actionplace_order
Owner, risk and permitted tools travel with the agent, not with the codebase.
Redaction, EU standard in path
PROMPT AS SENT BY THE APPLICATION
Refund Anna Kowalska, IBAN PL61109010140000071219812874, card ending 4471.
PROMPT AS IT LEAVES YOUR NETWORK
Refund {PERSON_18b}, IBAN {IBAN_3d2}, card {PAN_a09}.
Entities detected3 redacted, 0 passed
Re-identificationOn response, inside your boundary
If detection cannot runFail closed
02 Data

Protect sensitive data

Personal data, credentials and confidential records reach models through prompts, retrieved context and tool responses. Minimisation has to happen in the request path, not in a policy document.

Where it breaks today
Redaction is implemented per application, so every team gets it slightly wrong.
Retrieval and tool responses pull data back in after the prompt was cleaned.
A DPIA claims minimisation that nobody can demonstrate per request.
What obstruo does
Detects and tokenizes personal data and secrets before the request leaves your network.
Applies the same standard on the way back, including tool and retrieval responses.
Records what was redacted per request, so GDPR Article 32 has an artefact behind it.
Redaction Secret detection Per-country standards Fail closed
03 Models

Control models and providers

Approval decisions are made per model and per region, then quietly bypassed the next time a team hardcodes a model name. Logical model names put the decision back in policy.

Where it breaks today
Model names are hardcoded, so an approved-model list is advice rather than a control.
A provider deprecates a version and every project has to be patched.
Nobody can prove that EU traffic stayed on EU endpoints last quarter.
What obstruo does
Applications request a logical model name, and policy resolves which provider answers.
Unapproved models and out-of-region endpoints are refused in path.
Swap a provider, add a fallback or pin a jurisdiction without touching application code.
Approved models Routing Residency Fallbacks
Model catalog
obstruo-chat-euAPPROVED
resolves to eu-central, fallback eu-west
obstruo-reasoningAPPROVED
high-risk projects require approval record
obstruo-cheap-batchNON-EU, LIMITED
allowed for anonymised workloads only
provider-x-previewBLOCKED
no DPA on file, 41 attempts refused this week
Applications never name a provider. Policy decides, and the decision is recorded.
Tool permissions, trading-analyst 4 of 11 allowed
market_data.readAllowed
portfolio.readAllowed
research.searchAllowed
crm.contact.readAllowed, redacted
place_orderDenied
payments.transferDenied
unregistered MCP serverRefused
Reads allowed, writes denied, every attempt recorded with its arguments schema.
04 Tools

Govern MCP tools

MCP turned every internal system into something a model can call. The question stops being what the model says and becomes what it is permitted to do.

Where it breaks today
Teams register MCP servers themselves, and security learns about it later.
A single server exposes both harmless reads and irreversible writes.
Tool responses carry personal data straight back into the model context.
What obstruo does
Only registered MCP servers are reachable, and only per agent.
Permissions are set per tool and per action, so reads and writes are separated.
Tool responses pass the same redaction standard as prompts before re-entering context.
MCP registry Action permissions Response redaction
05 Policy

Enforce organisation-wide AI policies

An AI policy that lives in a wiki gets read once. The same policy, expressed as locked, recommended and optional controls, applies itself to every project on creation.

Where it breaks today
Each team interprets the policy differently, and reviews find it out afterwards.
New projects start ungoverned until someone remembers to onboard them.
Exceptions are granted in email threads and never expire.
What obstruo does
Organisation policies are inherited by every project, with locked controls that cannot be edited locally.
Exceptions are requests with an owner, a scope and an expiry date, recorded as evidence.
Policy versions are tracked, so every past decision points at the rules in force at the time.
Inheritance Locked controls Exceptions Versioning
Inheritance, policy v14
ORGANISATION
EU personal data redactionLOCKED
Approved models onlyLOCKED
High-risk agent baselineRECOMMENDED
inherited by 23 projects
PROJECT, CLAIMS AUTOMATION
3 inherited controlsnot editable
Claims retention rulePROJECT
Exception, batch modelexpires 14 Sep
Vendor and processor register from live traffic
EU model providerDPA on file
eu-central · personal data · 6 projects · last call 4 min ago
US model providerDPA on file, SCCs
us-east · anonymised only · 3 projects · last call 22 min ago
Embedding providerReview due 30 Aug
eu-west · confidential · 11 projects · last call 1 min ago
provider-x-previewUnapproved
blocked in path · 41 attempts · requested by 2 projects
Every row is derived from requests that actually happened, with the projects behind it.
06 Vendors

Build a live vendor and processor register

Your Article 30 records and vendor inventory are usually a spreadsheet describing last year. The control plane already knows which processors receive data today.

Where it breaks today
The register is assembled by asking teams, once a year, from memory.
A subprocessor is added through a new SDK, and procurement never hears about it.
Nobody can say which data categories each provider actually received.
What obstruo does
Builds the register from real traffic: which provider, which region, which data classification, which projects.
Holds DPA status, transfer mechanism and review dates against each vendor.
Refuses providers that are not approved, and records who tried to reach them.
Processor records DPA status Residency Review dates
07 Evidence

Prepare evidence for audits and reviews

Auditors and enterprise buyers ask the same question: show me that the control was active on this date, for this system. Answering it should take an export, not a project.

Where it breaks today
Evidence is reconstructed from logs, tickets and screenshots weeks after the fact.
Security questionnaires block deals for weeks because the answers are manual.
Controls are described in prose, and mapped to frameworks by hand each cycle.
What obstruo does
Keeps the policy version, decision, routing, redaction and tool call as one linked record.
Maps controls to EU AI Act, GDPR, ISO 27001 and SOC 2 references once, then reuses them.
Exports an evidence pack per project, agent, framework or date range.
Audit log Framework mapping Evidence export Approvals
Evidence pack, Q2 2026 COMPLETE
Governed requests4,182,551
Redaction events318,904
Blocked actions1,046
Open exceptions3
Incidents closed7 of 9
MAPPED TO
EU AI Act, Article 14 GDPR, Article 32 ISO 27001, A.8.16 SOC 2, CC6.1
Budgets, this month €38,140 of €52,000
Claims automation€14,900 / €18,000
Support copilot€9,240 / €20,000
Trading research€14,000 / €14,000
Ceiling reached, further calls refused until reset or increase
Budgets act on the request. The refusal is recorded with the policy that caused it.
08 Cost

Control AI spend and usage

Runaway spend is a governance failure with an invoice attached. Budgets and rate limits belong next to the other controls, enforced on the same request.

Where it breaks today
Cost is discovered at the end of the month, on a provider invoice.
One retry loop in one agent consumes a quarter of the quarterly budget.
Spend cannot be attributed to a project, agent or business owner.
What obstruo does
Enforces spend ceilings and rate limits per organisation, project and agent.
Attributes every request to its project, agent and owner, so cost has an address.
Routes eligible workloads to cheaper approved models, without weakening the data rules.
Budgets Rate limits Attribution Routing
09 Assistants

Extend governance to coding assistants

The largest source of unmanaged AI traffic in most organisations is not an application. It is an IDE assistant and a browser tab, both reading source and customer data.

Where it breaks today
Assistant traffic bypasses obstruo entirely, so no policy applies to it.
Secrets and customer records get pasted into prompts nobody records.
Blocking the tools outright pushes engineers onto personal accounts.
What obstruo does
Any assistant that accepts a custom base URL can be pointed at obstruo today, and inherits the same policy.
Secret detection and redaction apply to pasted context, not only to application prompts.
Dedicated adapters for coding assistants and browser AI are on the roadmap, not shipped. We will say so plainly in a security review.
Base URL onboarding Secret detection Adapters, planned
Coverage, honestly labelled
Assistants with configurable base URLLIVE
Secret and PII detection on pasted contextLIVE
Coding-assistant adaptersPLANNED
Browser adaptersPLANNED
Exposure map for unmanaged AI useIN DEVELOPMENT
Design partners shape this surface. Ask us where your assistants would sit today.
Working with an advisor on this?

Consultancies, ISO 42001 implementers and AI Act advisors connect their assessments to obstruo enforcement, so their recommendations end up as active controls instead of a report.

See the partner programme

Your AI. Your data. Your control.

Bring the use case that is blocking your security review. We will show the controls and the evidence it produces.

See obstruo in action See pricing