With the 19.3% average growth rate per annum in global AI R&D spending (Statista statistics, 2023), the AI technology iteration cycle has been condensed to 6-8 months, which poses a severe challenge directly to Notes AI founded on intelligent notes. For instance, in 2023, OpenAI launched GPT-4 Turbo, its context window was enlarged from 32k to 128k, and the speed of long text processing improved by 300%, while the semantic analysis model of Notes AI only handled 50k tokens during the same time, and the response lag ended at 2.3 seconds (industry average of 1.5 seconds). Exposed gaps in real-time processing capabilities. However, Notes AI mitigated the impact of technology generation to some degree by reducing user data encryption costs by 40% (compared to cloud solutions) by simplifying localized deployment solutions and achieving 27% daily active user growth in the first quarter of 2024.
At the market level, Microsoft Loop and Notion AI, by virtue of extensive integration with GPT-4, have gained the market at an annual cost per user of $12 (Notes AI is $18), reducing the latter’s market share in the SME market from 15% in 2022 to 11% in 2023. But Notes AI, by virtue of its differentiation strategy, has introduced multimodal input (voice to text accuracy 98.7%) and cross-platform synchronization speed (500MB/min) to the industry leader in the top 5%, and doubled the size of user annotation data using federated learning technology, improving personalized recommendation accuracy by 22%. As per Gartner, the usage level of AI capabilities in intelligent note-taking software has hit 64% in 2023, but end-user fears regarding the risk of privacy violations have turned Notes AI’s ISO 27001 compliance certification into a differentiator, which has pushed its enterprise customer renewal rates to 89%.
Another technology adoption bottleneck is computing power demand. For example, Meta’s Llama 3 model is trained with 5.2GWh (equal to the annual electricity usage of 50,000 households) while Notes AI quantizes model parameters from 175B down to 7B by utilizing quantitative compression technology, reducing inference power consumption by 76%, such that its mobile package size for installation is limited to 85MB (average for similar products is 120MB). Apart from that, its adaptive learning engine has reduced content classification mistakes from 8.4% to 3.1% via processing 1 billion user behavior logs, and achieved 5 million interactions per month in education (Coursera Partnership case 2023). Despite an 18% year-to-year increase in the price of AI chips (TechInsights statistics), Notes AI has been able to maintain operating costs constant at 32% of revenue via a hybrid cloud model and a profit margin of 14.5% in 2023, slightly higher than the industry average of 12%.
In the longer term, however, the rampant growth of generative AI (a 430% year-over-year growth in global generative capacity in 2023) is revolutionizing user behavior. For example, ChatGPT’s “Live meeting summary” feature speeds up information processing to 1,200 words per minute (compared with the human rate of 200 words), while the equivalent Notes AI function manages a mere 600 words and needs to be manually invoked. To counter this trend, the company has increased its R&D spend to $120 million in 2024 (28% of revenue), will include reinforcement learning frameworks, and will increase auto-tagging coverage from 35% to 80% by 2025, and API call latency to 0.8 seconds. If such metrics are achieved, Notes AI can surpass current technology constraints and become worth more than 15% of the market value in the Red Sea AI tool (compared to current 9%).