AI for Independent Authoring
using LLMs with augmented
knowledge bases
The Challenge
A leading platform for supporting independent authors attaining better outcomes for the books they publish, recognized the need for technology to scale their business. In particular, transforming authors into authorpreneurs was identified as one of the key deficiencies in an independent author’s journey that could have significant impact on their success as independent authors. The manual process of positioning a book was targeted for automation using AI. Tatras was given the responsibility to develop an automated, scalable, accurate and cost optimized solution for this.
Hypothesis
- RAG based LLM can be very effective to generate specific insights from a manuscript.
- Traditional NLP can be used to preprocess data for better LLM context.
- Report Generation can be distributed to help latency.
- Inclusion of Knowledge bases can reduce the chances of hallucination.
Execution
- Developed necessary Knowledge base using automated web scraping and traditional NLP.
- Created and fine tuned LLM prompts based on specific use cases for improved performance.
- Developed and deployed a complete UI and scalable Backend for consistent usability.
- Developed GenAI based solution for Book Graphics with minimum user interaction.
Outcomes
- Enhanced efficiency of report generation in terms of quality speed and cost.
- Dynamic Knowledge base support for queries.
- Robust and efficient UI for ease of access.
- Incorporating User feedback for improved suggestions.
- More independent authors are opting for creation of the AI based positioning reports.
- The platform has been able to scale their support to Independent authors resulting in better outcomes.
- Automation of these reports resulted in human effort now focused on deeper editorial inputs through efficiencies gained in positioning report generation.
Project Highlights
- Game changing AI service add to existing SaaS platform.