labs@atheron:~/industries/security-services/ai$ train agent --grounded

Incident reports

First-pass reports from a guard's notes, checked by a supervisor.

Supervisors spend hours rewriting guards' notes into reports for each client. An assistant that drafts the report in the client's format, from the guard's own notes, gives that time back while the supervisor still decides what is sent.

Start a project See the example projects

// what matters here

What is different in this industry.

Drafts, never sends

Every report is reviewed and sent by a supervisor. The assistant only drafts.

From the guard's own notes

The first pass is built from what the guard recorded, and says so where a detail is missing rather than inventing it.

Private by design

An open-source model in your own cloud account keeps incident details out of anyone else's systems.

Tested before each change

A set of real past incidents is run before any change goes live.

// example projects

Priced examples for this service.

Each one is hypothetical, labelled as such, and priced live from our rate card. Open any of them in the estimator and make it yours.

Example projectWeb appAIOpen-source model

Incident report assistant

Hypothetical. Not a client, not a result.

// The problem

Guards write incident notes in a hurry, supervisors spend hours turning them into client reports, and the reports differ from guard to guard.

// What we would build

An assistant that turns a guard's notes and photos into a first-pass report in the client's format for the supervisor to check and send, on an open-source model in your own cloud account so incident details stay private. It drafts; the supervisor decides what is sent.

// What is in it

First-pass reports from a guard's notes and photos, in each client's format
Every first pass reviewed and sent by a supervisor
Past reports and post orders used as reference
An open-source model served in your own cloud account
A set of real past incidents run before every change

// Stack

Llama or Mistral (open-source)
vLLM
PostgreSQL with pgvector
Next.js
DigitalOcean GPU

// estimate

Build: ≈ US$84,900 to US$130,000, delivered within 20 weeksCAD 120,900 to 184,700
Hosting: ≈ US$4,360 a monthCAD 6,210 a month
Support: ≈ US$1,000 a monthCAD 1,425 a month

Prices in your currency are estimates from today's Bank of Canada rate. All invoicing is in CAD or USD.

AI route: Open-source model
Model running cost: ≈ US$4,190 a monthCAD 5,960 a month

Timeline by milestone

Discovery: 2.9 to 3.4 weeks
Specification and evaluation plan: 0.4 weeks
Design approved: 2.4 to 3.4 weeks
Core features: 1.4 to 1.9 weeks
Full build: 0.9 to 1.4 weeks
Working pilot: 2.9 to 5.4 weeks
Testing and fixes: 0.9 to 1.4 weeks
Launch: 0.9 to 1.4 weeks
Production: 0.9 to 1.4 weeks

How it is paid

Deposit 20%: CAD 24,120 to 36,880
Discovery 2%: CAD 2,412 to 3,688
Specification and evaluation plan 10.3%: CAD 12,421.80 to 18,993.20
Design approved 2%: CAD 2,412 to 3,688
Core features 6%: CAD 7,236 to 11,064
Full build 4%: CAD 4,824 to 7,376
Working pilot 31.1%: CAD 37,506.60 to 57,348.40
Testing and fixes 2%: CAD 2,412 to 3,688
Launch 2%: CAD 2,412 to 3,688
Production 10.6%: CAD 12,783.60 to 19,546.40
Holdback, 30 days after launch (10%): CAD 12,060 to 18,440
GPU time for training and testing, at cost: CAD 300. Billed up front, at cost, outside the milestones

Start from this example

// questions

Questions we get about this.

Will the assistant add details the guard did not write?

It is built not to. Where a detail is missing it marks the gap for the supervisor, and every first pass is checked before it is sent.

Can it write in each client's own report format?

Yes. Each client's format is set up once, and drafts follow it.

Why not use a frontier model through an API?

You could, at a lower start cost. Most security firms prefer incident details to stay in their own cloud account, so we recommend the open-source route.

Back to the industry: Security services About the service: AI agents and models

// next

Not quite your project?

Tell us what you have in mind. We will come back to you with a range and the questions that would narrow it. Or book a call and talk it through.

Start a project Book a call