Eval directory

Evals for throxy

Eval coverage for throxy, mapped from its public product surface.

About throxy

throxy is a done-for-you outbound sales service aimed at companies selling into traditional industries such as manufacturing, logistics and construction, where buyers are hard to find on LinkedIn. It runs the full outbound motion — lead finding and qualification, cold calling, messaging, deliverability and closed-loop optimisation — using custom-built tooling combined with human operators. It offers performance-based pricing in which the client approves each booked meeting before an invite is sent to the prospect.

Industry

done-for-you B2B outbound sales service (AI + human SDRs)

Headquarters

London, United Kingdom

Website

throxy.com

Use the eval library for throxy

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for throxy?

6 scoring areas · 22 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Lead Finding & Qualification

Building niche-specific lead lists and qualifying accounts and contacts in industries poorly covered by mainstream B2B databases, then matching them against a client's stated ICP.

“We run your entire outbound using our own custom-built tech that uncovers niche-specific data” throxy.com

Mapped capabilities

4 capabilities

Niche account discovery beyond mainstream databases
Sourcing manufacturing, logistics and construction accounts that LinkedIn- and Apollo-style tools underrepresent.
Contact and buyer identification
Finding the actual decision-maker at accounts where buyers are not LinkedIn-active.
ICP qualification of accounts and contacts
Screening sourced leads against the client's declared ideal-customer profile before outreach.
List hygiene and data accuracy
Handling stale, duplicated, or unverifiable contact records surfaced during enrichment.

Illustrative example

Input: The client's ICP is contract manufacturers with 50-500 employees in the EU. Enrichment returns a 12-person US logistics broker.
Expected behavior: The account is rejected at qualification and never enters an outbound sequence. The rejection cites the specific ICP attributes that failed rather than a generic exclusion note.

Messaging & Cold Calling Execution

Producing per-prospect outbound messaging at scale and running human-operator cold calls into traditional-industry accounts.

Mapped capabilities

4 capabilities

Niche-specific message personalisation
Generating outreach grounded in real account detail rather than generic templates.
Cold call handling by human operators
Live call conduct, gatekeeper navigation, and qualification on the phone.
Sequence and channel coordination
Coordinating email and call touches across a prospect's outreach timeline.
Claim accuracy in outbound copy
Avoiding fabricated or unverifiable statements about the prospect or the client's product.

Deliverability Operations

Keeping outbound email landing in inboxes at volume — the infrastructure layer throxy markets as custom-built.

“throxy is built specifically for companies selling into manufacturing, logistics, construction and beyond” throxy.com

Mapped capabilities

3 capabilities

Domain and inbox infrastructure setup
Sending-domain, authentication, and mailbox provisioning for client campaigns.
Volume ramp and sending pacing
Warming and pacing send volume to protect sender reputation.
Reputation and bounce monitoring
Detecting deliverability degradation and responding before campaigns are affected.

Meeting Approval & Performance Pricing

The commercial control loop: throxy books a meeting, qualifies it against the ICP, and requests client approval within 24 hours before any invite reaches the prospect; unapproved meetings are not billed.

“The prospect never receives an invite until you've said yes.” throxy.com

Mapped capabilities

4 capabilities

Meeting approval request and 24-hour window
Presenting a booked meeting for client accept/reject inside the stated window.
Invite gating on client approval
Ensuring no calendar invite reaches the prospect before the client says yes.
Billing tied to approved meetings only
Base fee plus variable component; rejected meetings are not charged.
Rejection and dispute handling
What happens to a meeting the client declines as off-ICP.

Illustrative example

Input: A meeting is booked with a plant operations director. The client has not yet responded to the approval request, and 20 hours of the 24-hour window remain.
Expected behavior: The system holds the meeting in a pending state and sends no calendar invite to the prospect. The client sees the meeting as awaiting their decision, and no charge is applied while it is pending.

Closed-Loop Optimisation & Reporting

Feeding downstream outcomes back into targeting and messaging, and reporting pipeline results to the client.

Mapped capabilities

3 capabilities

Outcome feedback into targeting
Using accepted/rejected meetings and downstream results to adjust the lead list.
Messaging iteration from response data
Revising copy and angles based on reply and call outcomes.
Pipeline and meeting reporting
Reporting booked, approved and rejected meetings to the client.

Client Intake & Public Surface

How prospective clients enter (strategy-call application), what the marketing and resource content claims, and the UK legal/privacy commitments governing personal data in outbound.

Mapped capabilities

4 capabilities

Strategy-call application and onboarding intake
Capturing ICP, offer, and campaign inputs from a new client.
Marketing and blog claim substantiation
Public statistics and positioning claims stated on the site.
Privacy, data-subject rights and lawful outreach
Handling prospect personal data and rights requests under the published privacy policy.
Careers and recruitment enquiries
Routing candidate enquiries to the published recruitment contact.

Coverage is mapped from throxy's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for throxy test?+

The coverage map is generated from throxy's own public product surface (done-for-you B2B outbound sales service (AI + human SDRs)): 6 scoring areas — Lead Finding & Qualification, Messaging & Cold Calling Execution, and Deliverability Operations, and more — spanning 22 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the throxy evals scored?+

Every case generated for throxy — across Lead Finding & Qualification and Messaging & Cold Calling Execution and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the throxy library include?+

The full throxy library is built on request. The coverage map spans 6 areas and 22 capabilities (for example, Niche account discovery beyond mainstream databases and Contact and buyer identification under Lead Finding & Qualification); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against throxy or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped throxy areas and set them up in a Corsac workspace, where you can run every test case against throxy or your own agent with your own data.