Fractal Summary： 2024／5／9 I'm tired, so I want to organize the assumptions and verifications of the hypothesis

from Fractal Reader Development and Operation Diary Fractal Summary: 2024/5/9 Feeling a bit sluggish, so organizing hypotheses for validation

• 
  • It seems like the decline in the number of repeat users has stopped (blu3mo)

Feeling a bit sluggish lately

• The main reason seems to be the lack of a clear direction
• Even if I repeat the same things I tweet in Japanese on Twitter, only about 5% of the gathered traffic seems to stick around
• I want to continue the cycle of “creating hypotheses and validating with minimal resources”

Assumptions (Regarding the allocation of resources in my life)

• I don’t intend to invest my development resources heavily in this
  • I don’t particularly aim to become a strong web engineer, so I don’t find the skills or experience gained there very appealing
• I will use my resources for business creation and decision-making (blu3mo)
  • I find great value in the knowledge and insights based on the real experiences gained there
  • There’s also motivation from the enjoyment of new experiences#living like a mist in the world
  • I already feel that I’ve gained a perspective and mindset of someone who creates businesses, which I couldn’t get from research or hobby development (blu3mo)
    • Like in 29 Customer Value and Delivery Method

Withdrawal Conditions

• I don’t have a strong attachment to FractalReader itself
• I choose to continue because it seems like the best option among the cards I currently hold for doing I want to spend 10-20% of my time and prioritize the value for customers to create something once
• There isn’t much reason to continue if it’s not easy to commercialize

Hypotheses

• Achieve MMR$1000 with Fractal Summary without using too many development resources

  • I will withdraw once this is disproven
  • $5permonth\ast 200people,$10 per month * 100 people, $20permonth\ast 50people,$100 per month * 10 people, $1000 per month * 1 person

  • Assuming a toC, the goal is achievable

    • Assuming readers of academic papers as customers, the goal is achievable

      • Feeling a bit off
        • Academic papers are generally highly specialized, so summarizing them appropriately requires high LLM performance
        • Processing PDF papers properly and delving into them is quite challenging
          • Especially with formulas
      • But about 20% use them that way

    • Assuming readers of internet articles as customers, the goal is achievable

      • This seems more plausible
        • Even from current usage, about 30-40% fall into this category
      • There is an issue with Sharing Functionality Leads to Increased Awareness, but Seems to Conflict with Copyright
      • A straightforward approach like “Providing website summarization functionality with a Chrome extension” might be the most reasonable
        • Seems promising (tkgshn)

    • Assuming people who summarize meeting transcripts as customers, the goal is achievable

      • This seems quite plausible
      • About 10-20% of users are using it in this way
      • It’s possible even with chat logs
      • Since this use case doesn’t seem widely recognized, I’ll tweet and see the response
      • I want to grasp more specifically what it means to be a “person summarizing meeting transcripts”
        • I run an online salon for educators, and we have regular meetings on Zoom, so I plan to try using it there (kawahiii)
          • I tried summarizing last week’s meeting video with Fractal Summary (kawahiii)
            • The accuracy of the transcription wasn’t sufficient (this is a glasp issue), and the lack of shared materials made it a bit hard to read (kawahiii)
              • It might be good to delve into what kind of transcription would be ideal (nishio)
                • The example of Social Hack Day uses Zoom recordings uploaded privately on YouTube and uses that transcript
            • Currently, it might be better to have meetings that don’t handle materials in a dialogue format? (kawahiii)
            • I’ll try using it for now and see the reactions of the customers (kawahiii)- “When sharing materials on the screen, just listening to the audio might not provide enough information,” said nishio.

• “Oh, that’s helpful,” replied blu3mo.

  • “Indeed, content that can be understood solely through text might be more appropriate,” added blu3mo.

• Regarding YouTube and podcasts

• Besides the above, there may be other good customers.

  • Although not deeply explored, potential customers could include those who read legal literature or use cases in the education sector.