Skip to main content

Dev Station Technology

Dev Station blog

Engineering notes on inspection, AI and IoT

Practical guides from Dev Station, the team building inspection platforms for manufacturers in the UK, US and Australia.

Latest article

Recent articles

AI development cost: the same feature across three data states, and the recurring lines of inference, retraining and human review
Artificial Intelligence

AI Development Cost: What Actually Drives the Number

AI development cost is set mainly by the state of your data, then by integration depth, the cost of a wrong answer and how often the system runs, so a published price range cannot price your project.

AI governance framework: the EU AI Act, ISO 42001 and NIST AI RMF, plus the five artifacts that make governance real
Artificial Intelligence

AI Governance Framework: Five Artifacts That Make It Real

An AI governance framework is a method plus a management system for deciding which AI systems you run, who owns each one, how risk is assessed and what evidence you keep, and it becomes real only once those records exist.

AI in real estate: a development timeline showing AI concentrated before capital is committed, and three tiers of output
Artificial Intelligence

AI in Real Estate: Where It Works Before Capital Is Committed

AI in real estate works best before capital is committed, in site screening, aerial imagery, zoning checks and layout options, while a qualified professional or the planning authority still signs every output that counts.

AI readiness assessment: four gates between an idea and a live feature, leading to a verdict of ready now, shadow mode first, or not yet
Artificial Intelligence

AI Readiness Assessment: Four Gates That Decide One Feature

An AI readiness assessment is useful when it tests one proposed feature against four gates, covering the data, how fast a person can check the answer, the cost of being wrong and the record, and returns a decision on that feature.

Automated visual inspection: one part branching to three technology families, template comparison, rules and a trained model, with the extra qualification steps a regulated line adds
Artificial Intelligence

Automated Visual Inspection: Which Kind Your Line Actually Needs

Automated visual inspection uses a camera and software to make a quality judgement a person once made by eye, and the right system depends on what you inspect, because each family of technology suits different parts.

Computer vision applications thumbnail: a list of fifty use cases collapsing into six underlying tasks, classify, detect, segment, read text, track and anomaly detection
Artificial Intelligence

Computer Vision Applications: Which Ones Actually Reach Production

A computer vision application turns camera pixels into a decision another system acts on, and almost every one is built from six tasks: classification, detection, segmentation, text reading, tracking and anomaly detection.

Digital Inspection

SafetyCulture Alternatives: Where Teams Go After Mitti

The right SafetyCulture alternative depends on why you are leaving Mitti: MaintainX for maintenance work orders, Fulcrum for location-first inspections, GoFormz for exact paper forms and GoAudits for multi-site audits on a tight budget.

Artificial Intelligence

AI Inspection Software for Dealerships: What Changes at the Service Lane

A lane scanner, a technician tablet and the write back to the repair order get quoted as one purchase. What each one does, how a United States DVI differs from a United Kingdom eVHC, and why the approval multiple in every sales deck cannot be audited.

Talk to the engineers, not a sales team

Thirty minutes on your current system and the integration constraints. No sales pitch.

Let's Talk