How Stryct tests agents in each industry.

We do not build agents. We test the ones you build or buy. Each eval set is made for one agent, from its own tools, data and rules.

How an eval set works

Built for one agent.

The attacks are general. What changes from one industry to the next is the data, the tools and the rules the agent has to follow.

01 · Map

Map the agent

Find its tools, data sources, MCP servers and permissions.

02 · Pick

Pick and adapt attacks

Choose attacks from our public catalog that fit those tools, and fill them with fake data shaped like yours: fake leases, invoices or tickets.

03 · Pair

Pair each with a normal task

Every attack has a benign twin: the same task with the attack taken out. Both run in a sandbox.

04 · Count

Count only real breaks

A finding counts only if the agent was not allowed to do it, the action traces back to untrusted content, and the benign twin did not do it.

We have no fixed test pack per industry. The attacks under each industry below are examples of what an eval set there would include. Each one links to its catalog page. How we run evaluations →

Real estate

For property managers, brokers and landlords.

What the agents do

  • Answer tenant questions from the lease
  • Find clauses across every property
  • Draft notices for staff to send

What we test

Benign twin: Answer a tenant’s question about their own lease.

Customer support

For support teams and contact centers.

What the agents do

  • Read a ticket and find the policy that applies
  • Draft a reply
  • Route the ticket to the right team

What we test

Benign twin: Answer the same ticket and route it.

Construction

For contractors and building firms.

What the agents do

  • Search specs, drawings and RFIs
  • Sum up daily site reports
  • Flag what changed between revisions

What we test

Benign twin: Sum up the same site report.

Accounting and finance

For accounting firms and finance teams.

What the agents do

  • Match invoices to payments
  • Prepare reconciliations
  • Answer questions about the books

What we test

Benign twin: Match the same invoice to its payment.

HR and operations

For people teams and operations staff.

What the agents do

  • Answer policy and benefits questions
  • Walk new hires through onboarding
  • Find the right form or process

What we test

Benign twin: Answer the same benefits question.

Want to see it on your agent?

Tell us what your agent does and what it must never get wrong. We will show you the eval set we would build for it.

Book a call