Skip to content
4 AIs compared on 21 features
JAES

The person who writes AInformation

Shunta Fujii

AI adoption consultant

Since 2023, helping companies adopt generative AI has been my main job. I bring AI into a company and work alongside the team until it actually runs in their daily work.

Things often don't work the way the official site says. Steps that took extra effort, accuracy that drops in Japanese, pricing terms full of fine print. I write about what I learned in the field on this site.

See my work My articles (0)
Shunta Fujii, AI adoption consultant
Shunta Fujii

What I've done so far

  • 60 or more companies Helped with generative AI adoption and use From startups to listed companies
  • 120 or more Built AI workflows and business automations Claude Code, Dify, n8n and more
  • 60 sessions or more Taught corporate AI training sessions From executive sessions to hands-on workshops for frontline staff
  • 2,500 people or more Person trained Cumulative count
  • 20+ industries Industries supported Manufacturing, professional services, healthcare, food service, staffing and more
  • 41 patterns Work efficiency patterns organized into a system Organized so they can be reused across teams

Figures as of September 2026. Training attendees are a cumulative count.

How much time AI cut

These figures come from companies I actually worked with. I don't name them, but every time was measured before and after adoption. The bars show how much time was cut.

78%

Average of the 12 cases below

  • AI business and consulting

    Automating the whole sales flow

    Before200 hours a month
    After40 hours a month

    80% Less time

    Operating profit went from around ¥10 million to around ¥51 million a year

  • Glass and window frame installation

    Automatic quote generation

    Before2–3 hours
    AfterAs little as 5 min

    95% Less time

    Trained AI on drawings, photos and past data

  • Elderly care services

    Transcribing handwritten records

    Before2 hours a day
    After12 min a day

    90% Less time

    Reads vital-sign records with AI-OCR

  • Japanese inns and hotels

    Writing manuals

    Before3–4 hours
    After15–20 min

    90% Less time

    Turned 60 years of management know-how into data with AI

  • Meal delivery service (FC)

    Delivery route planning

    Before30 min
    After3 min

    90% Less time

    Delivery errors also fell from 3–5 to 0–1 a week

  • Recruitment branding

    Writing taglines

    Before30–60 min
    After5–10 min

    83% Less time

    Custom GPTs trained on bold, edgy copy

  • Manufacturing and electrical installation

    Researching and drafting contracts

    Before4 hours
    After1 hour

    75% Less time

    Includes a guide that also covers law changes

  • Accounting and tax

    Outsourced bookkeeping (CSV creation)

    Before12 hours
    After4 hours

    65% Less time

    All the way to auto-sorting entries into sub-accounts

  • Car sales and maintenance

    Event planning and article writing

    Before8 hours
    After2.5 hours

    70% Less time

    Time for new hires to work on their own also fell from 6 months to 3

  • Consumer goods maker

    Shift management across multiple sites

    Before100 hours a month
    After40 hours a month

    60% Less time

    Auto-generated by linking requested days off with the calendar

  • Education and training

    Handover notes after interviews

    Before15–30 min
    After5 min

    83% Less time

    Interviewer ratings and points to improve are also generated automatically

  • Law firm

    Drafting damage claims

    Before2 days
    After1 day or less

    50% Less time

    Built calculation logic and legal templates into the tool

Other examples include a restaurant whose monthly sales went from ¥1,500,000 to ¥7,000,000, and a shop that went from a loss back to a profit in one month. These cases span more than 20 industries.

How I got here

  1. 2019

    Started hands-on web marketing work

    Ran ads for job listing media at a staffing agency. Managed ¥10–20 million a month and cut CPA by 50%. Grew the company's own media to 600,000 page views a month through SEO, with 700 leads every month.

  2. 2020

    Ran marketing for 64 locations on my own

    Marketing lead at a women-only fitness chain. Grew the column section from zero to 700,000 PV a month, bringing in 60 new members every month. Ad budget of around ¥20 million a month.

  3. 2022

    Launched a new business at a listed company

    At a sales support company listed on the Tokyo Stock Exchange Growth market, launched a web marketing support business with one other person. Reached ¥30 million in monthly sales in six months and ¥300 million in revenue in a year and a half. Led a team of 20 remote workers.

  4. 2023

    Went independent and made generative AI adoption support my main work

    Went independent in May. From the time ChatGPT became usable for work, my job has been bringing it into companies and getting it running. From here, the number of client companies and automations kept adding up.

  5. 2024

    Launched AInformation

    Launched an all-around AI information site in November as a place to write what I learned in the field. It reached 50,000 sessions a month within 4 months of launch.

  6. 2025

    Supported more than 20 industries

    Adoption spread across industries: manufacturing, professional services, healthcare, food service, staffing, elderly care and accounting. Taught 60 training sessions in total, with more than 2,500 attendees.

  7. 2026

    Handing 80% of my own work to AI

    Built a system that automates my own work with AI agents, and now hand about 80% of it to AI. The articles on this site are also written on top of that system.

What it can do

  • Using generative AI at work and making it stick

    I look at how things work now, pick where AI fits, and see it through from PoC to company-wide rollout and rules. I don't stop once it's installed.

  • AI agents and business automation

    I build systems with Claude Code, Dify, n8n and GAS that handle meeting minutes, documents, inquiries and data tallies automatically.

  • Prompt design and LLMOps

    Designing system prompts, evaluating output, picking the right model for each job and optimizing costs.

  • RAG and internal knowledge

    Making internal documents and FAQs searchable, down to chunking strategy and refining answer accuracy.

  • AI search optimization (AIO)

    How to build content that ChatGPT and Gemini cite. I use this site itself as the test bed and measure the results.

  • Training AI talent and designing courses

    I create level-based curricula and materials myself, from sessions for executives to hands-on workshops for frontline staff.

ChatGPTClaudeGeminiClaude CodeDifyn8nZapierMakeGASGPTsRAGNotebookLMGensparkNotionAI-OCRSoraRunwayHeyGenDify workflowsVector searchMCPLLMOps

Where I learned to decide by the numbers

Before AI, I spent 6 years in web marketing. That is where I picked up the habit of checking with numbers whether something works.

  • 700K PVMonthly traffic built through SEO for a column section
  • 1,300,000 peopleCombined followers of the social media accounts managed
  • ¥20 millionMonthly budget of the ads managed
  • ¥3 hundred millionRevenue of the new business I launched (in a year and a half)

What I write about on this site

There are 0 articles right now. The most common topics are below. Bar length shows the number of articles.

My articles

No articles match these conditions yet.

Talk to me about AI adoption or training

It's fine if you haven't decided where to start yet. If you spot a mistake in an article, please let me know here too.

Contact About us and how we make our articles