New Research · Q2 2026 10,765 agents · 4 yrs production data AI live

98% of your top performers would never have been interviewed.

And that is just the hiring data.

NODES tested every screening filter a Fortune 500 insurance carrier runs (keywords, industry experience, assessments) against 10,765 hires and four years of actual production data. The filters do not predict. NODES built the infrastructure that does.

The same brain reads every other system where signal is hiding. The HR ticket your team is answering for the fourteenth time. The deal slipping a week at a time. The new hire dipping at the same point in week four. Same problem. Same data. Different decision.

Dataset
10,765 agents
Lift, p=0.006
2.47× top performer
Status
VPC · live
fig 01 · decision trace · live artifact
customer-owned · reproducible
decision_trace · v0472.json
req: rv-2412-08831
MR
M. Rojas, Producer, Commercial Lines
req · austin-tx · screened 2026-04-14
87
fit score
Evidence chain
01 ATS filter
Industry experience ≥ 3 yrs retail banking · adjacent
would reject
02 Keyword
"licensed producer" · "P&C" not in resume
would reject
03 Assessment
Conscientiousness 82 · Coachability 79 dims: 28
above threshold
04 AI interview
Structured · 4 of 6 signals present 12 min
pattern match
05 Production
Predicted 18-mo output cohort p74 · $54/day lift
top-quartile
model: carrier-hire-2025.11 · calibrated on 4,214 hires
open trace →
Trusted inside regulated enterprises Deployed in customer-owned environments
The screening gap

Keyword filters don't predict production. Ours do.

NODES ran 8,181 candidate keywords against four years of post-hire production at a Fortune 500 carrier. After Bonferroni correction, none predicted sustained performance. Thirty were anti-predictive, correlated with lower output.

The standard ATS funnel at this customer eliminated 98% of their eventual top performers cumulatively. The industry-experience filter alone eliminated 80% of them.

80%
of top performers were filtered out by the "relevant industry experience" requirement alone.
$17.7M in annual production at a 2,000-hire volume that this single filter would have excluded from the pipeline.

Source: 4-year retrospective on 10,765 agents at a Fortune 500 carrier. Top performers defined as sustained-production cohort (p75+ over 18 months).
Keywords tested: 8,181
0 predict production (Bonferroni-corrected)
30 anti-predictive
AUC 0.512 → 0.618
Fig 02 · Production by Fit Score decile
5.37× production gap between top-scored and bottom-scored segments.
Avg. daily production ($) by Nodes Fit Score decile, 18-month sustained cohort. n=10,765. p=0.0007.
Bottom decile Top decile
decile 010203040506070809decile 10
Core IP · Decision Traces

Every hire, from filter to field, as queryable evidence.

A Decision Trace connects what the ATS screened on, what the assessment measured, what the interview surfaced, and what actually happened in production, for every candidate, indefinitely.

Institutional knowledge becomes auditable. When a hiring manager leaves, their judgment doesn't leave with them.

  • 01 · Ingest
    Every candidate event from your ATS, HRIS, assessments, and CRM zero data egress
  • 02 · Attribute
    Every decision (screen, advance, reject, hire) linked to its originating signal reversible
  • 03 · Measure
    Each signal scored against real production outcomes, not interviewer gut feel compound over time
  • 04 · Replay
    Query "show me every hire where the industry filter would have rejected a top performer" auditable
candidate #8871 of 10,765
Signal captured
What it measured
Production outcome
ATS keyword: "producer"
src: workday · stage: req-level
Resume string match
present: no · confidence 1.00
excluded
Would-have-been top quartile
counterfactual · monte carlo · n=312
Industry tag: "insurance ≥ 3y"
src: recruiter_notes · stage: screen
Prior industry tenure
value: 0 yrs · flag: adjacent
excluded
Top-performer pattern match
dims: 28/28 · score 87
overrode filter
Behavioral assessment
src: assessio · stage: mid-funnel
Conscientiousness, coachability
c: 82 · co: 79
advanced
Correlates with 18-mo output
r: 0.31 · p: 0.004
AI interview (structured)
src: nodes.interview · 12 min
4/6 top-performer signals present
transcript · sig hash 0a91f…
advanced
Hired · promoted at month 11
output: $54.35/day over cohort
validated
trace_id · dc_8871_2026-04-20
signed · reproducible · customer-owned
Platform

One brain. Every decision.

Same model, same data, same proof primitive. Different decision. Three pillars, sixteen use cases.

Hire & Develop

From the first applicant to the first promotion. The brain reads every signal in the pipeline, scores it against real production data, and tells your team who to call next.

01

Recruitment & Screening

Problem: 1,000+ applicants per role, screened by keyword filters that reject 80 percent of future top performers.

Solution: The brain reads every applicant against 28 behavioral dimensions calibrated on your own production data. A 0-100 Fit Score writes back to the ATS as a native field.

Proof: 80 percent of top performers at a Fortune 500 carrier were filtered out by a single "relevant industry experience" requirement. The Fit Score caught them.

Read the use case
02

AI Interviews

Problem: Structured interviews take 45 minutes per candidate. Hiring managers skip them or conduct them inconsistently.

Solution: A 12-minute async interview surfaces the same signals, scored against production outcomes instead of interviewer gut feel. Full transcript and audit trail included.

Proof: Four of six top-performer signals detected in a candidate the ATS would have rejected. That candidate was promoted at month 11.

Read the use case
03

Persona-based Sourcing

Problem: Sourcing teams search LinkedIn with the same keywords that fail in the ATS. The passive talent pool looks identical to the active one.

Solution: The brain identifies passive candidates across LinkedIn, GitHub, and public portfolios using patterns learned from your own validated top performers, not inferred boilerplate.

Proof: Sourced candidates matched the top-performer pattern at 2.1x the rate of inbound applicants at pilot volume.

Read the use case
04

Ramp Acceleration

Problem: New hires take 6-9 months to reach full productivity. Managers rely on gut feel to decide who needs coaching.

Solution: The brain reads onboarding signals (training completion, early production, manager check-ins) and flags exactly where each new hire is diverging from the top-performer ramp curve.

Proof: Role-calibrated ramp signals identified at-risk hires 6 weeks earlier than manager escalation at pilot scale.

Read the use case
05

Retention & Attrition

Problem: Turnover costs 50-200 percent of annual salary. By the time HR sees an exit interview, the decision was made months ago.

Solution: The brain reads production trends, engagement signals, and cohort patterns to surface retention risk before the employee starts job-searching.

Proof: Cohort-level attrition forecast within 3 percentage points of actual over an 18-month observation window (n=10,765).

Read the use case
06

Internal Mobility & Succession

Problem: High performers leave because they do not see a path. Succession plans live in spreadsheets updated once a year.

Solution: The brain scores every employee against every open or projected role using the same Fit Score. Cross-role pattern matching replaces manager nominations.

Proof: Internal mobility candidates matched the top-performer pattern for the target role at rates comparable to external hires scored by the same model.

Read the use case

Operate & Run

Every operational system generates signal. The brain reads it, drafts the next action, and lets your team approve. HR tickets, compliance reviews, compensation benchmarks: same model, same proof trail.

07

HR Service & Ticket Triage

Problem: Your HR team answers the same fourteen questions every week. Each answer takes 8-12 minutes and a policy lookup.

Solution: The brain reads every inbound ticket, drafts a policy-grounded response, and routes edge cases to the right specialist. The human approves; the system learns.

Proof: Ticket resolution time reduced by 60 percent in pilot, with zero policy-violation escalations on auto-drafted responses.

Read the use case
08

Compliance & Audit Prep

Problem: Audit prep takes weeks of pulling records across five systems. Evidence gaps surface during the audit, not before.

Solution: Decision Traces provide a continuous, queryable audit trail. Every hiring decision, every score, every override is linked to its evidence chain and signed.

Proof: Legal approval in 17 days at a carrier where six prior AI vendors were rejected over eighteen months.

Read the use case
09

Compensation & Pay Equity

Problem: Compensation bands are set by title and tenure. Actual production varies 5x within the same band.

Solution: The brain maps pay to production at the individual level, surfaces pay-equity gaps, and recommends adjustments grounded in output data rather than market surveys.

Proof: 5.37x production gap between top-scored and bottom-scored deciles within the same role and location (n=10,765).

Read the use case
10

Manager Intelligence

Problem: Some managers consistently develop top performers. Others consistently lose them. Nobody measures why.

Solution: The brain traces team-level production, ramp curves, and retention rates back to manager actions. The result is a coaching signal, not a leaderboard.

Proof: Manager-level production traces revealed a 2.1x output variance attributable to management patterns, controlling for hire quality.

Read the use case
11

Territory & Placement

Problem: New hires are placed by zip code or manager preference. The wrong territory can turn a top performer into an early attrition.

Solution: The brain matches candidate patterns to geo-calibrated production data, recommending the placement most likely to produce sustained output.

Proof: Geo-calibrated placement recommendations correlated with $54/day production lift over cohort average at the pilot carrier.

Read the use case
12

Workforce Planning

Problem: Headcount planning runs on last year's numbers plus a growth target. Nobody models what the incoming cohort will actually produce.

Solution: The brain projects cohort-level output by combining pipeline quality scores, historical ramp curves, and attrition forecasts into a single planning model.

Proof: Cohort output forecast within 8 percent of actual over a 12-month window at 2,000-hire volume.

Read the use case

Sell & Grow

The brain that reads people reads deals. Pipeline velocity, renewal risk, referral signal: the same pattern-matching engine applied to revenue systems.

13

Pipeline & Deal Intelligence

Problem: Deals slip a week at a time. By the time the forecast misses, the quarter is over.

Solution: The brain reads CRM activity, email cadence, and meeting patterns to score deal health in real time. Slipping deals surface before the rep flags them.

Proof: Deal-health scoring identified at-risk opportunities 3 weeks earlier than rep self-reporting in pilot CRM integration.

Read the use case
14

Renewal & Retention Risk

Problem: Customer churn is measured after it happens. Renewal conversations start too late and rely on relationship intuition.

Solution: The brain reads usage patterns, support ticket sentiment, and engagement signals to score renewal risk months before the contract date.

Proof: Renewal risk scores flagged 70 percent of eventual churns at least 90 days before contract expiration in pilot data.

Read the use case
15

Referral Signal

Problem: Referral programs pay bounties but do not target. The best referrers are not identified; the best candidates are not described.

Solution: The brain identifies which employees are most likely to refer a top performer, based on network overlap with the validated top-performer pattern.

Proof: Referred candidates from targeted referrers matched the top-performer pattern at 1.8x the rate of untargeted referral programs.

Read the use case
16

Career Pathing

Problem: Career development conversations happen once a year. Employees leave because they cannot see what is next.

Solution: The brain maps each employee's current signal profile against every role in the organization, surfacing the highest-fit next moves and the gaps to close.

Proof: Next-role fit scores correlated with actual promotion outcomes at r=0.34 (p<0.001) across the observation cohort.

Read the use case

The math is the credibility. The story is bigger.

The same brain that read these 10,765 hires reads every system in your company, drafts the next action, and lets your team approve. From the day Alex applies to the day Alex retires. And every system in between.

Read the full story
Deployment

Inside your VPC. Zero data egress. Legal in weeks.

Single-tenant deployment in customer-owned infrastructure. Fine-tuned, open-source models. No third-party AI in the chain. SOC 2 Type II. Reviewable by your security team the same way Snowflake is.

Legal approval
17days
After six prior vendors were rejected over eighteen months at the same carrier.
Contract → production
34days
VPC-deployed, integrated with the customer's ATS and HRIS, scoring live pipeline.
Third-party AI in the chain
0
Open-source foundation models, fine-tuned inside the customer boundary. Weights owned by the customer.
Compliance posture
SOC 2 Type II
Single-tenant. No shared inference. No shared training. Customer-owned fine-tuned models.
deployment.topology operational
Customer VPC · single-tenant
ATSworkday · live
HRISukg · hourly sync
CRMsalesforce · prod
Nodes modelfine-tuned · customer-owned
Trace storepostgres · encrypted
Inferencein-cluster · gpu
egress: 0 bytes · no third-party AI
External surface
Admin console only · SSO · SCIM · audit log stream
30 minutes · Your data · No pitch deck

See what the last four years of your hires actually predicted.

NODES will run a backtest against a sample of your production data, in your environment. You will see, numerically, which of your filters worked, which did not, and what a calibrated Fit Score would have done differently.

01
30-minute call, your research & security team invited
02
NODES backtests against a de-identified sample of your data
03
You see, numerically, what your filters missed
04
If the math works, NODES scopes a VPC pilot (median: 34 days to prod)