Smalldata AI Control Plane

Inline control for every prompt, plugin and agent.

One gateway between your company and every AI model. It authorises, redacts, reroutes or blocks each request in real time, and writes a signed record of every decision. No endpoint agents, no code changes, under 50 ms.

Live decision feed (illustrative)
TimeUser / agentDestinationActionReasonSigned
09:41:07 marketing / j.rossi chatgpt.com Rerouted Consumer tier → enterprise tenant ✓Signed
09:41:12 pricing-agent api.external-llm.example Blocked Tool scope exceeds read-only policy ✓Signed
09:41:18 support / m.bianchi helpdesk-plugin.example Blocked Unvetted SaaS plugin ✓Signed
09:41:25 hr / l.verdi chatgpt.com Redacted Codice fiscale, 2 emails removed ✓Signed
09:41:31 dev / a.neri api.anthropic.com Redacted API key pattern removed ✓Signed
09:41:40 ops / c.ferri copilot.microsoft.com Allowed Sanctioned tenant ✓Signed
09:41:46 strategy / s.conti perplexity.ai Rerouted Unreleased spec detected ✓Signed
09:41:52 finance / d.greco chatgpt.com Redacted IBAN removed ✓Signed
09:42:03 dev / r.marino api.mistral.ai Allowed Approved model, quota ok ✓Signed
09:42:09 sales / f.costa chatgpt.com Rerouted Consumer tier → enterprise tenant ✓Signed
Illustrative requests, not real traffic.

AI is outpacing your controls.

Visibility

You can't govern what you can't see.

Tools, plugins and agents arrive before security knows. In our sample, 5 of 6 AI apps were unsanctioned.

Data

Data leaves in prompts, not files.

PII, contracts and source code get pasted into consumer AI. DLP was built for files, not prompts.

Agents

Agents are the new attack surface.

Broad database and tool access means one prompt injection can trigger a full table scan.

Blocking drives it underground. Governing it inline doesn't.

See. Control. Prove.

See

Every AI tool, model, plugin and agent, on one map.

Passive inline discovery of web and API AI traffic. Sanctioned versus unsanctioned, users and departments, token volume, and risk severity per app.

Marketing HR Engineering Support Pricing agent Gateway Consumer chatbot Model API Enterprise tenant SaaS plugin Self-hosted model

Sanctioned Unsanctioned

Every discovered AI destination flows through one inline point.

Control

Decide, inline, what reaches a model.

Redact PII and secrets before they leave. Block or redirect unapproved tools to sanctioned tenants. Sandbox tools and MCP for agents. Set department quotas and model routing.

Policy

IF payload contains IBAN | codice fiscale | customer email → REDACT
IF destination = consumer ChatGPT → REROUTE to enterprise tenant

Prove

An evidence-grade record of every decision.

A cryptographically signed, tamper-evident log of prompt origin, sanitised payload, tool execution and token cost, mapped to GDPR and the EU AI Act.

Audit record

timestamp
2026-03-12T09:41:25Z
identity
hr / l.verdi
destination
chatgpt.com
payload_hash
sha256:9f2c…e41a
action
REDACT (2 fields)
token_cost
1,204 in · 318 out
signature
✓ verified

What we'll show you in 30 minutes

01

Shadow AI discovery

every tool, model and plugin, with users, departments and token volume.

02

Data minimisation

PII, financials and secrets redacted inline before they leave.

03

Operational control

block or redirect unapproved tools and SaaS plugins at the perimeter.

04

Agent & MCP governance

per-tool policies, schema-level boundaries, and signed traces across every hop.

05

Cost & token governance

per-department attribution, quotas, model routing and failover.

06

Evidence-grade audit

an immutable, searchable record of every decision, ready for auditors.

Test a control, such as “block all non-EU-hosted models”, against real traffic before you enforce it.

Deploys in days. Invisible to users.

Users · Apps · Agents
Smalldata AI Control Plane gateway identify · inspect · decide · sign
  • OpenAI
  • Anthropic
  • Mistral
  • Google
  • Azure
  • Self-hosted
  • No endpoint agents
  • No code changes
  • <50 ms overhead
  • Provider-neutral
  • EU data residency
  • Italian-language team

What an assessment uncovers

142,500AI requests analysed
315Active AI users and endpoints
18AI destinations discovered
42PII payloads sent to public models
€14,400Per year in shadow subscriptions

From a 14-day sample assessment (illustrative)

See what an assessment uncovers →

Who it's for

Inline visibility and enforcement

See and control AI traffic with no endpoint agents and no code changes. Under 50 ms, fail-closed.

Book a 30-min demo

A short form. We reply within two working days to agree a time.

Preferred language

A live walk-through on sample data. Nothing to install.

Frequently asked questions

What exactly do you need to install?

A network-level gateway in passive mode. No endpoint agents and no application changes. We'll confirm the exact routing (proxy, DNS or egress) with your network team on the scoping call.

Where is our data processed and stored?

In the EU. Payloads are obfuscated for PII by default in the workbench. A DPA is signed before kick-off.

Does the EU AI Act apply to us?

If you deploy AI systems used in the EU, very likely. The report maps your inventory to risk tiers.

What happens after the assessment?

You can switch the same gateway to active protection, with critical leak prevention in 24 hours and full posture in 14 days. Or you can stop there: the report is yours.

Which AI providers do you support?

All major model providers and self-hosted models; the gateway is provider-neutral.

Does it work with Microsoft Copilot or enterprise tenants we already approved?

Yes. Sanctioned tools are recognised and allowed under policy, and their usage is logged.

What about latency?

Under 50 ms of gateway overhead, with fail-closed defaults.

See every AI request. Control what leaves. Prove it to any regulator.