Our Videos

Analysis, methods and field experience on AI in industry. Every episode comes with its full transcript and chapters.

Video thumbnail: Why does AI fail? The Pearl Street lesson — TARANIS AI

Episode 1September 2, 2026

AI won't change your company. Method will.

Why doesn't AI deliver in industry? The answer may lie in a story from 140 years ago. 1882, New York. Edison switches on the first power plant. Factories don't change. 1900: the dynamo replaces the steam engine. Factories still don't change. Then something changed everything — and it wasn't the machine. AI follows exactly the same path. Only 6% of organizations capture real value from AI (McKinsey, State of AI 2026 — global survey of 97 countries).

View the episode page →

Video thumbnail: Green Claims: the audit starting September 27 — TARANIS AI

7:13

Episode 2September 18, 2026

Green Claims: the audit starting September 27

Still printing 'eco-friendly' on your packaging without proof? Starting September 27, an environmental claim without traced evidence becomes prohibited. This video gives the four-step method — inventory, qualify, prove, trace — to go from missing proof to traced proof. On the program: the Tuesday-morning scene, the logic of a withdrawn text, the European Commission figures (53% doubtful claims, 40% with no evidence), the penalties (€100,000 administrative fine, up to 10% of revenue under criminal law), then the four-step grid to launch on Monday. Chapters: 00:00 You're at your desk on a Tuesday morning when your head of m 00:43 Inventory, qualify, prove, trace: four concrete verbs you ca 01:27 Hold on to this logic: a withdrawn text doesn't mean a dropp 02:09 Hold on to your seat: fifty-three percent of them turned out 02:51 One hundred thousand euros for a legal entity, for a single 03:24 Except your consumer-facing customers will have to prove wha 03:57 Step two, qualify: for each claim, decide between vague and 04:37 Keep the compass: an environmental claim without traced evid Sources: European Commission, 2020 study of 150 environmental claims (53% vague, misleading or unfounded; 40% backed by no evidence); French Environmental Code, art. L229-69 (€100,000 administrative fine for a legal entity, non-compliant carbon-neutrality claim); French Consumer Code, art. L132-2 (misleading commercial practice: criminal fine up to 10% of revenue, raised to 80% of advertising spend for environmental claims). TARANIS AI — diagnosis, method, experience: https://taranis-ai.com If this analysis resonates with you, let's dig into it at TARANIS AI: diagnosis, method, experience — everything is there to support you. You will find all the links in the comments of this video. Have you ever hunted for the proof behind a claim you have printed for years? Share your experience in the comments, I read everything and I answer every question.

View the episode page →

Video thumbnail: The report that writes itself: the 4-step method — TARANIS AI

8:04

Episode 3September 15, 2026

The report that writes itself: the 4-step method

Friday evening, past six. You are still at the office, and your monthly report still will not add up. This scene, seen in dozens of plants, has a way out: a report that almost writes itself — with no ERP change, no hiring. The condition: start from the method, not from the tool. On the program: the 2019 ritual, the agent principle (search, cross-check, draft), the figures from a machinery company (320 h → 45 h a month, deviation detection from 14 days to under 24 h), the two objections that sting, and the 4-step grid to launch on Monday. Chapters: 00:00 Reporting Friday night 00:58 The promise: method before tool 01:29 2019: three truths, zero common language 02:18 The report that writes itself 02:58 320 hours → 45 hours: the numbers 04:22 Your objections: messy data, AI deciding 05:17 The 4-step grid 06:30 Monday morning: from scribe to pilot Sources: Taranis AI field case (official LinkedIn post, 09/06/2026) — illustrative figures; Taranis AI method (diagnosis, 4 steps, traceability). TARANIS AI — diagnosis, method, experience: https://taranis-ai.com If this analysis resonates with you, let's dig into it at TARANIS AI: diagnosis, method, experience — everything is there to support you. You will find all the links in the comments of this video. Have you ever watched your team lose a Friday night to a report? Share your experience in the comments, I read everything and I answer every question.

View the episode page →