Skip to content →

Our AI Workflow and Content Approach

Last updated August 14, 2026. Reviewed monthly and whenever policy changes.

This page describes how we develop content and how AI fits into our workflow. Each section is a short list so you can see how we work at a glance.

Websites and platforms

  • WordPress with accessibility-focused themes.
  • Content is developed through a combination of voice-to-transcript, documents, and human editing.
  • We use AI in some cases to edit for readability, and in most cases to check consistency across sites.
  • Every published piece is reviewed by a person before it goes live.

What is human-written and what is AI-generated

  • Human-written or human-first, then AI-checked. Website copy, messaging templates, learning circle challenges, personas, blog and thought pieces.
  • AI-generated with human review. Compliance and regulatory language: Twilio SMS 10DLC requirements, WhatsApp Business Platform templates, email deliverability protocols. This content is prescriptive by design and the wording is largely dictated by the platforms it must satisfy. Our review checks accuracy, completeness, and fit to our business, not tone.
  • All content, whether editorial or regulatory, is reviewed by a person for readability and for our intent to connect at a human level.

Our messaging templates

  • 2 Degree Shift has been developing in career technical education, adult learners, and work-based learning for over 20 years.
  • We offer 12 plus messaging templates. The templates are developed by people.
  • Our template content is informed by podcasts, white papers, news, research, and newsletters we listen to and read.
  • We use AI as a research and validation mechanism to check whether our messaging is current with model versions and relevant to industry practice.

Learning Circle Challenges

  • Learning circle challenges are drafted by people. We outline the project first.
  • We then feed the draft into AI and ask whether there are add-ons beyond our sector knowledge that would serve circle members.
  • When we receive add-on ideas or validation, we as people reverse-engineer the challenge into work-based learning. We draft the steps and the messaging that informs the challenge.
  • We then use AI again to check for errors, omissions, and messaging structure that meets SMS and WhatsApp formatting requirements.
  • In the context of learning circle challenges, AI is a check and balance, not a creator.

Personas

  • Personas for reduced token cost customers are formed through conversations with people who work in like industries and through a common UX and UI persona approach.
  • We are applied researchers. We believe the quality of a product is always in the human connection.
  • We use AI to check for current trends, companies, people, or priorities we might have missed. We validate personas with AI or receive recommendations for expansion or improvement.
  • When we personalize a persona for a company, that persona is based on the roles, priorities, and strategies of that company. We gather that from conversations with the company, materials they have posted, and news related to them.

Personalized messaging for companies

  • We take the personas and the 12 templates and ask AI for topic recommendations that may sit outside our direct experience.
  • We combine our own ideas and the AI recommendations to draft personalized messaging.
  • As we draft, we upload our progress into AI for additional recommendations and validation.
  • Before we send, we check with AI that the messaging is still in compliance with Twilio for SMS and Meta WhatsApp for WhatsApp messaging.

Email

  • We use Missive so that all email lives in one central location.
  • People respond to all email messages.

Help desk tickets

  • We outline the response first as people.
  • We upload the outline to AI to validate our direction for the response.
  • We review the AI response for fit to the specific question.
  • For higher-complexity tickets, we may validate across a couple of different AI models. We do this because we do not want to build assumption bias, either from one model or from one staff member.
  • Our purpose is to use AI as a peer review alongside our staff, not as the responder.

Language translation

We rank all messages by translation complexity into three tiers.

  • Tier 1: Simple message. AI can translate, and a person reviews.
  • Tier 2: Message needs editing. A human translator does the work with AI as a check or validation.
  • Tier 3: Cultural context is involved. Human translation only.

How this workflow relates to the EU AI Act

The European Union’s Artificial Intelligence Act includes transparency obligations for AI-generated content under Article 50. These obligations apply from August 2, 2026 (European Commission, Guidelines on transparency obligations).

Two provisions of Article 50 are relevant to how we work.

Article 50(2) requires that providers of generative AI systems mark their outputs in a machine-readable format so the content is detectable as artificially generated. This obligation falls on the AI model providers, not on organizations that use those models to draft content.

Article 50(4) requires that deployers who publish AI-generated or AI-manipulated text “with the purpose of informing the public on matters of public interest” disclose that the text is artificially generated. The Act provides an exception: this obligation “shall not apply where the AI-generated content has undergone a process of human review or editorial control and where a natural or legal person holds editorial responsibility for the publication of the content” (EU AI Act, Article 50).

The workflow described on this page documents that human review and editorial responsibility. Every piece of content published under the 2 Degree Shift name or under any of our product names, whether editorial or regulatory in origin, is reviewed by a person before publication. A named person on our team holds editorial responsibility for what is published.

For content that is AI-generated by design, the compliance and regulatory language described above, the human review checks that the required regulatory elements are present, that the language is accurate to our business, and that citations are correct. This review is substantive. It is not a cursory approval.

We revisit this page monthly and whenever policy or product scope changes, so that our stated workflow continues to match how we actually work.

Our position on AI as a tool

  • Our business model has always been people first.
  • We welcome AI as a tool that supports regulatory compliance and extends our understanding beyond our immediate cultural assumptions.
  • We see AI as an invaluable tool for industry, and one that is operationalized with people, not around them.

Questions about this workflow, or about how a specific product uses AI, can be emailed to support@2DegreeShift.com