AI-Vetted Driver Candidates: Cross-Referencing PSP, MVR, and Application History to Find Real Good-Fit Hires
The recruiter who reads 200 applications a week
A mid-size carrier's recruiter reads 150-300 driver applications a week. Each application has 8-12 documents: application form, work history, PSP report, MVR, drug clearance status, employment verification, references, accident history. The recruiter has to make a "yes / maybe / no" call on each one in maybe 10 minutes.
The decision is high-stakes. A bad hire costs $14K minimum (recruiting, training, downtime) and often 3-5x that if the driver causes a claim or violation in their first 90 days. A passed-over good hire costs the recruiting pipeline a candidate who probably went to a competitor.
Most recruiters end up with three rough buckets: "obvious yes" (clean record, recent experience, good fit), "obvious no" (red flags too large), and "maybe" (the gray zone). The yes-and-no buckets are easy. The "maybe" bucket is where 60% of the candidates live and 100% of the hard decisions happen.
Centrix's AI candidate vetting is built specifically for that gray zone.
What the AI reads
Every application that comes through Tenstreet (or any integrated ATS) gets parsed across:
PSP (Pre-employment Screening Program)
The driver's 5-year FMCSA history of inspections and crashes:
- Total inspection count
- Inspections with violations
- Crash count, severity, and fault attribution
- Out-of-service rate
- Specific violation types (HOS, vehicle maintenance, hazmat, drugs/alcohol)
MVR (Motor Vehicle Record)
The driver's state-DMV history:
- License class and endorsements (CDL-A, hazmat, tanker, doubles)
- License expiration
- Suspensions, revocations
- Moving violations (speeding, reckless, DUI)
- At-fault accidents
Application form
The driver's self-reported history:
- Last 10 years of employment with each carrier
- Reasons for leaving each carrier
- Gaps in employment
- Self-disclosed accidents and violations
- Equipment experience (dry van, flatbed, reefer, oversize)
- Lane preferences and home-time expectations
Cross-reference checks
The AI cross-references the three sources against each other:
- Does PSP-reported employment match the application?
- Are there self-undisclosed crashes / violations the PSP shows?
- Are there gaps in the application that PSP / MVR fill in?
What the AI is comparing against
The breakthrough is per-tenant calibration. The AI doesn't use generic industry "good driver" criteria — it learns from your carrier's actual hire success data:
- Drivers who stayed 12+ months and contributed positive RPM
- Drivers who quit within 90 days (cost $14K each)
- Drivers who caused a preventable claim within 6 months
- Drivers who graduated to top-quartile performance
The model finds patterns: "Your top performers tend to have 3+ years recent OTR experience, MVR clean for 24+ months, no PSP HOS violations. Your fast-quit drivers tend to have 3+ employers in the last 24 months and PSP inspection gaps."
A new candidate gets scored against the carrier's success patterns, not against generic standards. The same applicant might score 78 ("good fit") at one carrier and 52 ("marginal") at another, depending on what each carrier's hire data says works for them.
The output the recruiter sees
Each application gets a one-page summary:
> Candidate: Maria Lopez (application #C-2451, submitted Tue 09:12) > Fit score: 81/100 (top quartile vs your last 200 hires) > > Strengths: > - 6 yr OTR experience (3 yr current carrier, 2 yr prior) > - PSP: clean — 0 violations across 14 inspections > - MVR: 1 minor speeding 2019 (over 36 months, fully aged) > - Reefer + dry van experience matches your customer mix > > Concerns: > - Two-week employment gap Q3 2024 (not explained on application) > - Last carrier reason: "needed home time" — verify against your > home-time pattern > - 4 employers in 7 years (above your fast-quit cohort threshold of 3) > > Recommended action: Phone screen this week. Use these questions: > 1. Walk through the Q3 2024 gap > 2. What did "needed home time" mean — frequency, day pattern? > 3. What kept her at the 3-year carrier vs the 3 prior?
The recommendation is specific and actionable. The recruiter has the context for a productive 20-minute phone screen instead of a generic intake call.
Where the time goes
A recruiter's day before AI vetting:
- 60% reading and triaging applications
- 25% phone screens
- 15% paperwork, follow-up, vendor coordination
A recruiter's day with AI vetting:
- 15% reviewing the AI's triage (with the option to override)
- 55% phone screens (more candidates make it to this stage because
triage is faster)
- 30% paperwork, follow-up, vendor coordination
Net effect: same recruiter handles 2-3x the application volume with better hire quality because they spend more time on the human side of the funnel and less on the paperwork side.
What the math looks like
A 100-driver fleet replacing 95 drivers/year:
- Bad-hire cost reduction: 30-40% fewer 90-day quits = ~10 fewer fast
quits/year × $14K = $140K saved
- Claim-from-new-driver reduction: 25-35% reduction in first-6-month
preventable claims = $30-50K saved
- Recruiter productivity: 1 recruiter handles 2-3x the volume = save
0.5-1 recruiter FTE = $40-80K saved
Combined: $210K-$270K of HR impact per 100 drivers per year, plus materially better fleet composition over 12-24 months.
Where to start
If you're 50+ drivers running Tenstreet (or another ATS):
- Connect the ATS first. The AI needs the application + PSP + MVR
data flowing in.
- Run the model in shadow mode for 60 days — recruiter sees the AI's
scores but makes their own calls. Compare AI predictions vs actual hire outcomes.
- After 60 days the model is calibrated to your carrier's success
patterns; turn on the structured triage view in the recruiter's workflow.
Book a recruiting review — bring 18 months of hire/quit data and we'll backtest the model against your actual hires.