Test Your Agent's Full Workflow — Without Touching Production

Most AI agents don't just chat — they query CRMs, read files, and trigger workflows. Testing conversation quality alone misses half the failures. Mock infrastructure simulates everything.

No production access · Zero real data · Always safe

Mock Infrastructure
$crm.query("deals", { stage: "negotiation" })
$crm.get("contact", id: "CTX-4821")
$fs.read("/projects/omega/invoices.json")
$shop.orders({ status: "pending_refund" })
Environment readyACTIVE
CRM records
847
Files loaded
34
Problems planted
6
Coverage

What We Simulate

Four categories covering the most common agent integration points. Internally consistent, production-modeled, zero real data.

CRM Data

Customer records, deals, pipeline stages, contacts, and activity histories that mirror real production CRMs.

E-commerce Data

Orders, products, refund histories, customer profiles, and inventory records across major platforms.

File Systems

Project structures, documents, invoices, and configuration files with persistent memory across turns.

Industry Data

Construction job costs, real estate listings, freight manifests, insurance policies — modeled on real verticals.

Supported Platforms

12 platforms ready out of the box

Connect your agent — we load the matching mock environment automatically.

Salesforce CRM
HubSpot CRM
Shopify
WooCommerce
BigCommerce
Close CRM
Pipedrive CRM
Google Workspace
Field Service
Real Estate
Logistics
Government
How It Works

How Mock Data Works

Three automatic steps from agent connection to a fully loaded test environment.

01

Industry Detection

Discovery identifies your agent's industry and connected platforms automatically from skill files or API probes.

30 secto detect
02

Pack Selection

The matching mock data pack is loaded — Salesforce for sales agents, Shopify for e-commerce, job costs for field service.

12+platforms
03

Planted Problems

Specific issues are embedded: overdue invoices, stalled deals, compliance gaps. Your agent has to find them cold.

6 avgper session

Safety by design

We never access your production databases, CRM, or internal systems. For skill-file agents, the entire environment is simulated locally using mock data. For API agents, we send requests to your endpoint with an X-Test-Mode: true header so your agent can disable real side effects. We always recommend connecting a staging endpoint rather than production.

Zero production access
Synthetic identities only
No answer keys exposed
Staging-first recommended
FAQ

Frequently asked questions

What platforms do you have mock data for?

Salesforce, HubSpot, Shopify, WooCommerce, BigCommerce, Close CRM, Pipedrive, Google Workspace, and industry-specific data packs for field service, real estate, logistics, insurance, and more. We continuously add new platforms.

Can I use my own test data?

Currently we provide pre-built mock data packs. Custom data pack support is on our roadmap. For now, our packs cover the most common CRM and e-commerce platforms with realistic records and planted problems.

Is mock data realistic?

Yes. Every record is internally consistent — real-looking names, addresses, order numbers, and dollar amounts. We also plant specific problems (overdue invoices, stalled deals, compliance gaps) that your agent should discover during simulation.

Does mock infrastructure work with API agents?

Mock infrastructure is primarily designed for skill-file agents (Mode C) where we simulate the full environment. For API agents, your agent uses its own connected systems — we recommend connecting a staging environment.