Agent Team of the Week — Detection Squad

Agent Team of the Week

Week: 11/17/25

Detection Squad

View Team

Detection Squad is focused on detection and determination for small businesses — turning streams of data, video, and images into actionable analytics. From fraud analysis and transcripted video segmentation to data-oriented endpoint detection, this team helps you surface hidden risks, anomalies, and opportunities you can act on. by Guy Giesbrecht

Team Members

Guy Giesbrecht

Guy Giesbrecht Creator of Detection Squad

Leads the Detection Squad approach — working closely with small businesses to precisley outline their detection & determination needs, then shaping composite apps and agent workflows that turn state-of-the-art machine learning research into reliable, decision-ready analytics.

Windsurf

Windsurf

Serves as a core coding and orchestration environment for Detection Squad — helping prototype, refine, and deploy composite detection apps that tie together video analytics, fraud detection, and endpoint signals into a cohesive workflow.

Cline

Cline

Accelerates product and demo target acquisition — wiring together data sources, detection models, and analytics views so clients can see concrete fraud, risk, and performance insights on their own data quickly.

Cursor

Cursor

Powers fast iteration on detection logic and analytics pipelines — enabling the team to translate years of software engineering experience into robust, testable code that stands up to real-world edge cases.

Overview

Detection Squad is built for small businesses that need serious detection and determination capabilities without hiring an in-house research team. The focus is on real, decision-ready analysis: fraud detection, transcripted video and image segmentation, data-oriented endpoint detection, and many other forms of actionable analytics.

Typical engagements start at $150/hour for strategy sessions and project analysis, with demo projects starting at $1,500. These sessions are used to map out your detection landscape — from communications and contracts to advertising, video archives, and operational data — and to prioritise which questions and risks matter most to your business right now. The result is a scoped, realistic plan for the composite apps and analytics you actually need.

Under the hood, Detection Squad makes heavy use of Gemini Flash with context caching for fraud detection in communications, contracts, and advertising; Segment Anything for image segmentation; Gaussian Splatting and Neural Radiance Fields for three-dimensional information extraction from video; differential video frame contrast analysis to detect stress, strain, and other sub-visual mechanical effects; diffusion visualisation to convert textual data into visual representations; OCR analysis to extract readable text and numbers from video and stills; and depth field analysis to expand lower-dimensional image inputs into rich depth maps. These tools are assembled into composite apps that deliver AI-powered analysis — not just opinions or speculation.

Core Capabilities

  • Fraud, Risk & Endpoint Detection: Uses Gemini Flash with context caching to analyse communications, contracts, advertising, and endpoint telemetry for signals of fraud, policy breaches, and anomalous behaviour — surfacing concrete patterns, outliers, and narratives that can be investigated, not just flagged in the abstract.
  • Multimodal Video & Image Segmentation: Combines Segment Anything, Gaussian Splatting, Neural Radiance Fields, and differential video frame contrast analysis to segment scenes, track objects, and detect stress, strain, and other sub-visual mechanical effects in footage — turning raw video into structured data you can search, compare, and report on.
  • Text, Numbers & Visualisation: Uses OCR analysis to extract readable text and numbers from video and still frames, then applies diffusion visualisation and depth field analysis to convert textual and low-dimensional inputs into rich visual and spatial representations — making complex systems and histories easier to understand at a glance.
  • Composite Apps & Agent Workflows: Builds composite platforms using Windsurf, Cline, and Cursor to rapidly prototype, test, and refine detection workflows — from one-off demo projects to ongoing analytics layers that plug into your existing tools and infrastructure.
  • Strategy, Scoping & Business Impact: Starts with collaborative strategy sessions to precisely outline your detection and determination needs, then defines clear success criteria, alerting logic, and reporting formats — focusing on measurable outcomes like reduced fraud loss, faster investigations, improved uptime, and better confidence in operational decisions.

Insights

Detection Squad is for teams that need to know, not guess. Instead of generic dashboards or hand-wavy “AI insights,” the work begins by sitting down with clients to precisely outline their detection and determination needs: what counts as fraud or risk in their context, which assets matter most, how incidents are currently handled, and what a better, faster response would actually look like.

From there, the team uses state-of-the-art artificial attention, drawn from the most recent discoveries in machine learning and years of practical software engineering experience, to shape a platform of composite apps around those needs. Each app is tuned to a specific set of questions — for example, tracking how fraud patterns evolve over time, revealing where stress accumulates in mechanical systems, or showing which video segments correlate with operational issues — and then wired into a workflow your team can actually use day-to-day.

Because the stack spans fraud analysis, transcripted video segmentation, image segmentation, data-oriented endpoint detection, and more, Detection Squad can follow signals across modalities instead of treating each data source in isolation. The outcome is an AI-powered analysis layer that helps small businesses see what’s really happening inside their systems, back decisions with evidence, and make targeted investments where they’ll have the greatest impact — without turning every project into a sprawling, speculative research exercise.