OpsCom / EvoTrain

The threat never runs the same problem twice. Why does your training?

AI-generated scenarios for every officer, at every level. Fought for real. Measured through everything. Evaluated with a path to improve.

GenerateFightCaptureScoreAdapt
NVIDIA Inception Program member badgeMember

The problem

American training certifies memory and calls it readiness.

Scenarios repeat until officers train memory instead of decisions.

The scenario library is finite. Instructors know it. Officers know it. By the third exposure, the room is teaching pattern recall, not judgment.

The record at the end is a sign-in sheet.

Hours are counted. Decisions are not. When an incident is later questioned, there is no per-officer performance record to defend the training, and no evidence trail to defend the officer.

The loop

One loop. Every officer. Every evolution.

Generate a new problem, run it live, capture what happens, score it against evidence, and let the next problem attack the gap.

01

Generate

A new problem, built for that officer, at their level, informed by the gap their prior record showed. Grounded in real threat behavior and your own venue. Never reused.

02

Fight

Your instructors run it live, in your facility, or inside the VR and simulators you already own. Every packet is instructor reviewed and signed off before it runs.

03

Capture

Body-worn and facility video, radio and comms audio, timing, movement, simulator telemetry. Sensor agnostic. Nothing to install.

04

Score

Multimodal scoring against a sealed ground-truth key across decision latency, procedure adherence, communication, person accounting, and force decision. Plus discrepancy detection.

05

Adapt

The officer gets an evaluation with specific recommendations. Command gets a Readiness Score over time. The next problem attacks the gap.

Feedback into the next problem

Loading demos.

Academies

For academies and POST commissions.

Mapped to your curriculum.

Every generated problem and every score ties to an objective your academy already teaches and already reports on.

Cohort before individual.

Class-level performance by objective shows which block of instruction to rebuild, not just which recruit to counsel.

The record follows the recruit.

Objective-level history carries into field training and into the officer's first in-service year, on the same scale.

We prepare the course-approval documentation your state requires.

Capability

What the AI actually does.

The loop is the process. This is the technical capability behind it.

Doctrine-grounded generation

Problems are built from how real threats actually behave, not by randomizing variables in a template. The adversary has intent, and it holds up to a veteran's scrutiny.

Venue aware

Ingests your floor plans, your range, your building. The problem happens where your officers actually work, not in a generic room.

Per-officer targeting

Reads that officer's own measured record and generates against the specific weakness it found. Every officer in the same class can face a different problem.

Multimodal scoring

Aligns video, radio traffic, timing, and movement against a controlled ground-truth key, so the evaluation is built from evidence rather than opinion.

Discrepancy detection

Finds the divergence between what was planned, what was reported, and what actually happened. That gap is where training value and legal exposure both live.

Every output is human approved and explainable. No autonomous action. No weapons integration. Decision support, never decision replacement.

Every officer

Every officer. Every level.

Generated for the individual: their level, their assignment, the weaknesses their own record shows. No library. No reuse. No script to leak between cohorts.

Basic

Academy scenario blocks mapped to your curriculum: the recruit's first measured decisions.

Intermediate

Patrol and in-service problems for mandated hours that never repeat.

Advanced

Complex, multi-element problems for veterans and supervisors.

Refresher

Short perishable-skills blocks: firearms decisions, tactics, vehicles.

Specialized

SWAT, hostage response, protection details, corrections, airport and rail: full opposing-force evolutions.

Not five products. One adversary that grows with the officer, from first day to tactical team.

Integration

Connects to what you already own.

VR and simulators

Your systems become our delivery surface, and their telemetry feeds the Score. The menu becomes a generator.

Ranges and facilities

Shoot houses, ranges, and scenario villages: where generated problems deploy live.

Cameras and radios

Body-worn and facility video, radio and comms: already capturing, now scored.

Records and compliance

Evaluations and Scores export to training records, accreditation files, and the systems you report through.

Handled to CJIS-aligned standards. Keep the simulator. Keep the range. We are the adversary and the judgment they never had.

Position

Why nobody else can say this.

CategoryFillsLacks
Judgment simulatorsImmersion and safe repsCanned scenarios, session-only scoring, no live adversary
VR platformsCost-accessible immersionLimited measurement, no per-officer generation
Online training and LMSHours compliance at scaleNo performance data of any kind
Live schools and role playersReal pressure, credible cadreHand-written scenarios, after-action by memory

Nobody generates per-officer scenarios, runs them for real, measures through everything, and closes the loop with an evaluation.

Procurement

How agencies buy this.

Lane 01

Micro-purchase

One measured pilot day or course block. No competitive quotes required: a purchase order or card, documented for price reasonableness. The federal micro-purchase threshold is $15,000 as of October 2025, and grant recipients may self-certify up to $50,000.

Lane 02

Simplified acquisition

A term program across a training year. Price quotations from an adequate number of qualified sources. No sealed bidding below the $350,000 Simplified Acquisition Threshold.

Lane 03

Grant funded

COPS, Byrne JAG, and state training and violent-crime programs. We supply scope, pricing, and budget-narrative language for the application, and invoicing documentation for reimbursement.

We will not split a purchase to dodge a threshold. Where a pilot leads to a program, we scope the follow-on properly from the start.

CJIS-aligned handlingSole source justification narrative availableWe prepare your state's course-approval paperwork

Justification

The questions you will be asked after this meeting.

You do not decide alone. Here is what each body asks, and what this lets you answer.

Who asksWhat they askWhat you can say
Council or commissionWhy fund another vendor?You are already paying for the alternative. Chicago taxpayers spent $259M resolving police misconduct suits in 2025, roughly triple 2024, against an $82M annual set-aside. New York paid $117M in 2025 and over $796M since 2019.
Oversight boardHow do we know training changed anything?Per-officer performance measured over time against a defined proficiency threshold, auditable, and reported as a trend rather than an assurance.
The DA or a courtWill this hold up?Every score is anchored to timestamped evidence and can be reopened against the original recording. Findings are explainable line by line.
Finance or budgetWhat does it displace?No capital purchase and no installation. It targets the paid time and backfill line the agency already carries for every training day.
The unionIs this surveillance of officers?A development record, owned by the agency, produced for training. It is also the strongest defense an officer has ever had when a decision is questioned later.
State and federalDoes this fit policy direction?Executive Order 14288 of April 2025 directs federal agencies to expand access to and improve the quality of training for state and local law enforcement.

Sources: Chicago annual police litigation report, 2026 / Legal Aid Society analysis of New York City data, 2026 / Executive Order 14288, 90 Fed. Reg. 18765

Where this came from

Where this came from.

The method behind EvoTrain came out of United Nations protective operations. Planning time was reduced more than 70 percent, training frequency rose two to four times, across more than 60 operatives working in three high-risk countries.

OpsCom is founded by a former US police officer. Scott Hudson, COO, served with DHS and served in Iraq with the CEO. The CEO's background is Presidential Security Service and United Nations personal security detail operations. The CEO is an operator who ran personal security details.

The offer

Send us one scenario you ran this year.

We return three generated variations, free. No meeting required.

  • Same objectives, same environment, three problems your officers have never seen.
  • Delivered as a packet your instructors can run as written.
  • You keep everything, including the ground-truth key.

Optional.

A few sentences is enough. Format does not matter. If you would rather send a document, submit this and reply to our email with the attachment.

We use your info only to respond to this request.

The threat adapts. Now your training does.

OPSCOM, CORP. Contact hello@opscom.io. Demo content on this page is illustrative and contains no real agency, officer, or incident data.

This page uses cookieless product analytics with no session recording, no cross-site identifiers, and no personally identifiable information in any event payload.

NVIDIA and the NVIDIA Inception Program are trademarks of NVIDIA Corporation.