All speakers
Mr. Anuj Magazine
Workshop Faculty

Mr. Anuj Magazine

Technology Leader, Writer & AI Practitioner

Day 2 · Workshop on AI for Edupreneurs
“It comes into your life before it comes to your work.”
How AI arrives
Post-event

Session recap

Execution sense: putting AI into an ordinary working day

Anuj Magazine took the second half of the workshop and moved it from ideas to hands. His opening argument was simple: general purpose technologies never enter our work first — they enter our lives first. Electricity took forty years to reach the factory floor. The first thing most of us did with the internet was open a mail account, not a corporate site. AI will follow the same route, so the fastest way to get good at it professionally is to start using it personally, without pressure.

From there the room worked, rather than watched. Volunteers came up and spoke to AI by voice instead of typing, and the demonstrations were deliberately mundane: preparing for a teacher interview on the commute, uploading a job description, asking for unusual interview questions one at a time, then sharpening them live until they were genuinely useful. Within minutes the session had produced a set of questions no candidate could rehearse for — including asking a candidate what their principal, a colleague and a student would each warn a new employer about.

Three habits ran through everything he showed. **Prime the model** — tell it what is coming before you give it the task, exactly as you would brief a colleague; responses measurably improve. **Interrupt it** — chatbots are built to be helpful and will over-explain, so cutting them off mid-sentence is a feature, not rudeness. And **structure the prompt**: his CRAFT formula — Context, Role, Action, Format, Tone — turns a vague request into a reliable one. "Act as a teacher hiring review assistant" pulls the model into a persona it has been trained on; "stay close to what the candidate actually said" and "separate evidence from opinion" keep it from flattering you.

“AI will not come to your work first. It comes into your life first — that is how every general purpose technology has ever arrived.”

He closed the loop with recordings. Interviews, parent meetings and staff conversations can be recorded with consent on any phone, and the transcript is where the real value sits: a first-pass interview review, a summary of commitments made, follow-ups nobody would otherwise remember. He demonstrated wearable recorders such as the Indian-built Neo Sapien as a "second brain", while being emphatic that consent comes first, every time.

The through-line was that none of this asks for extra hours. It fits into the commute, the gap before a meeting, the walk between blocks — which is precisely why it works for people who run schools.

Editorial summary compiled by the LDP team — not a verbatim transcript. Spotted an inaccuracy? Let us know.

Watch the session — Anuj Magazine at the AI for Edupreneurs workshop, Day 2 of INSIGNIA 2026.

Key takeaways

  1. 1

    Use CRAFT for every prompt that matters

    Context (the situation), Role (“act as a teacher hiring review assistant”), Action (the single thing you want), Format (what the output should look like), Tone (evidence-based, brief, formal). Vague in, vague out.

  2. 2

    Switch to voice

    Dictate instead of typing. Latency is now low enough for real conversation, which means the commute, the corridor and the ten minutes before a meeting become working time.

  3. 3

    Prime before you ask

    Tell the model what is coming — “I am going to give you a job description, are you ready?” — before handing over the task. Prepared context produces noticeably better answers.

  4. 4

    Interrupt freely

    Models are trained to be agreeable and verbose. Cut in with “shorter” or “that's fine, but I want something else”. Steering mid-answer is faster than re-prompting.

  5. 5

    Record, transcribe, then extract

    With consent, record interviews and meetings on the phone you already carry. The transcript feeds a first-pass review, a list of commitments and the follow-ups you would otherwise lose.

  6. 6

    Ask for evidence, not validation

    Instruct the model to stay close to what was actually said and to separate evidence from opinion. Left alone it will flatter your assumptions.

What he spoke on

  • Why AI reaches personal life before professional life
  • Voice-first AI: live dictation and conversation on a phone
  • The CRAFT prompting formula — Context, Role, Action, Format, Tone
  • Priming and interruptibility as everyday prompting habits
  • Live build: interview preparation for a senior mathematics teacher role
  • Recording with consent, transcripts and the AI-assisted interview review
  • Wearable AI recorders as a “second brain”

Q&A captured

Q. After recording a meeting, do you just hand the recording to ChatGPT?

Either works — many phones and laptops now transcribe on their own. Once you have the transcript you can do a great deal with it: a first-pass review, a summary, or a list of commitments to follow up on.

Q. Is it acceptable to record an interview?

Only with consent, and say so up front. In an interview you hold the leverage, so the courtesy matters even more.

Q. Should I let the AI finish speaking before I respond?

No. Interruptibility is designed in. Be polite with people; with a chatbot, cut in the moment it drifts.

View the pre-event page (archived)— bio, predicted topics and pre-reads as shown before the session.