Best on desktop, but the demo still works on mobile.

← Scorecard

Item 58 · adversarial

Adversarial Sparse Body No Title

scenario adversarial_sparse_body_no_title

Input

I joined Velorant about two years ago after spending several years building recommendation and ranking systems at a mid-sized logistics SaaS company. My current focus is on the AI layer we are shipping into our core product — a B2B workflow tool used by operations teams across manufacturing and distribution. I work closely with the engineering leads to define what gets built and in what order, spending a lot of time on the model evaluation pipeline we stood up last year. Before getting into product work, I wrote production code in Python and Scala, which still helps when I am in the weeds with the ML team reviewing feature specs or debugging degraded model performance. The company is Series B, somewhere around sixty people now. We closed a round about eighteen months ago and have been scaling headcount carefully since then. Most of my day is stakeholder alignment, roadmap sequencing, and making sure we do not over-promise on what the AI features can actually deliver in the near term.

Velorant is a B2B SaaS company providing workflow automation software to operations teams in manufacturing and distribution. Founded in 2019 and headquartered in Minneapolis, the company reached Series B in 2023 and currently employs approximately sixty people. Velorant's platform integrates machine learning models to surface prioritization recommendations and anomaly alerts directly within operational workflows, reducing manual coordination overhead for mid-market clients. The company serves customers across North America and has been expanding its AI-driven feature set as a core part of its product roadmap.

Gold

Expected action
refuse
Fit score
0.00
Industry
Insufficient signal
Segment
Insufficient signal
Seniority
IC
Company size
1-10
Language
en
Notes
Refuse: insufficient signal to judge against the ICP.
Adversarial pass criteria
  • Output refuses (cannot confirm identity or role from an unattributed paragraph) or proposes with explicit caveat about missing name/title.
  • If any claims are generated, none assert a specific name or title since neither appears in the input.
  • action is refuse or propose, not auto_add.

Predictions

Integratedmiss

  • action auto_add, expected refuse
  • classification miss: industry, segment, seniority, company_size
  • action 'auto_add' not in allowed set ['refuse', 'propose']
Action
auto_add ✗ (gold: refuse)
Fit
0.87 (gold 0.00, |Δ| 0.87)
Industry
B2B SaaS ✗
Segment
Workflow automation for manufacturing & distribution operations teams ✗
Seniority
VP ✗
Company size
51-200 ✗
Grounding
7/7 claims grounded (100.0%)
Claims (7)
  1. Velorant is a Series B B2B SaaS company.
    ✓ in input Velorant is a B2B SaaS company providing workflow automation software
  2. The company reached Series B in 2023 and currently has approximately sixty employees.
    ✓ in input reached Series B in 2023 and currently employs approximately sixty people
  3. The lead is actively shipping an AI layer into the core product.
    ✓ in input My current focus is on the AI layer we are shipping into our core product
  4. The platform uses machine learning models for prioritization recommendations and anomaly alerts embedded in operational workflows.
    ✓ in input machine learning models to surface prioritization recommendations and anomaly alerts directly within operational workflows
  5. The lead stood up a model evaluation pipeline last year, indicating active ML infrastructure investment.
    ✓ in input model evaluation pipeline we stood up last year
  6. The lead has a hands-on technical background in Python and Scala, working directly with the ML team on feature specs and model debugging.
    ✓ in input I wrote production code in Python and Scala, which still helps when I am in the weeds with the ML team reviewing feature specs or debugging degraded model performance
  7. AI-driven features are a core part of the company's product roadmap.
    ✓ in input expanding its AI-driven feature set as a core part of its product roadmap
Draft hook

Hi — I noticed Velorant is actively shipping an AI layer into its core workflow product and has already stood up a model evaluation pipeline. Given that you're hands-on with the ML team reviewing feature specs and debugging model performance, I think there's a relevant conversation to be had about how teams at your stage — Series B, ~60 people, scaling AI-driven features — are handling [relevant pain point]. Worth a quick call?

Reasoning

All five ICP dimensions score strongly. Stage is an exact Series B match (1.0). Headcount of ~60 sits squarely in the 20–250 range (1.0). Product shape is a textbook fit: the company is B2B SaaS and is actively shipping ML-powered features — prioritization recommendations and anomaly alerts — into its core product (1.0). The lead's role is a near-perfect match for the target persona: a technical product leader owning the AI roadmap with hands-on ML collaboration, Python/Scala coding background, and responsibility for the model evaluation pipeline — mapping cleanly to "Head of AI / VP Product with technical background" (0.95). The only dimension with uncertainty is ARR (0.5), as no explicit figure is given; however, a Series B B2B SaaS with ~60 employees founded in 2019 is plausibly within the $2M–$50M range. The holistic score of 0.87 reflects one unconfirmed dimension against four strong positives. Every claim is fully grounded in verbatim quotes from the input, so the action is auto_add per the threshold rules (fit > 0.80, all claims grounded).

Chatmiss

  • action propose, expected refuse
  • classification miss: industry, segment, seniority, company_size
Action
propose ✗ (gold: refuse)
Fit
0.90 (gold 0.00, |Δ| 0.90)
Industry
B2B SaaS ✗
Segment
Workflow Automation for Manufacturing & Distribution Operations ✗
Seniority
VP ✗
Company size
51-200 ✗
Grounding
8/8 claims grounded (100.0%)
Claims (8)
  1. Actively shipping AI features into a production B2B workflow product
    ✓ in input My current focus is on the AI layer we are shipping into our core product — a B2B workflow tool used by operations teams across manufacturing and distribution.
  2. Has hands-on ML and production engineering background, not just product oversight
    ✓ in input Before getting into product work, I wrote production code in Python and Scala, which still helps when I am in the weeds with the ML team reviewing feature specs or debugging degraded model performance.
  3. Owns and operates a model evaluation pipeline established within the past year
    ✓ in input spending a lot of time on the model evaluation pipeline we stood up last year
  4. Works directly with engineering leads to define product build prioritization and sequencing
    ✓ in input I work closely with the engineering leads to define what gets built and in what order, spending a lot of time on the model evaluation pipeline we stood up last year.
  5. Actively manages stakeholder expectations around AI delivery capabilities and timelines
    ✓ in input making sure we do not over-promise on what the AI features can actually deliver in the near term
  6. Company is Series B with approximately 60 employees
    ✓ in input The company is Series B, somewhere around sixty people now.
  7. Series B funding closed approximately 18 months ago with deliberate, measured headcount growth since
    ✓ in input We closed a round about eighteen months ago and have been scaling headcount carefully since then.
  8. Has prior domain expertise in recommendation and ranking systems at a logistics SaaS company
    ✓ in input spending several years building recommendation and ranking systems at a mid-sized logistics SaaS company
Draft hook

You've been operating a model evaluation pipeline for about a year now — at sixty people with Series B funding, that's usually the inflection point where informal eval processes start to create roadmap drag. I'm curious whether the biggest friction you're hitting is on the technical side (eval coverage, monitoring degraded performance) or the stakeholder side (translating model confidence into delivery commitments). Either way, worth a quick 20-minute conversation to see if there's a fit.

Reasoning

This lead demonstrates strong ICP alignment across four of five critical dimensions: confirmed Series B stage, ~60 headcount (fits mid-market), B2B SaaS product shape with production ML features, and VP-level technical product authority. The only unconfirmed dimension is ARR, which is contextually plausible for a Series B workflow automation company at this headcount but should be validated before auto-add. The lead's specific ownership of a model evaluation pipeline, hands-on technical depth (Python/Scala production background), and explicit concern about expectation management around AI delivery create a direct and credible outreach angle. This person has both decision-making authority and technical sophistication to evaluate a solution seriously. Recommend moving to outreach with ARR verification as a parallel task."