Evidence-based AI
Sources we've referenced to build and improve Capmap for you.
- [1] Introducing the Model Context Protocol (MCP)
- Anthropic. (2024, November 25). Introducing the Model Context Protocol (MCP). Retrieved from https://www.anthropic.com/news/model-context-protocol
- https://www.anthropic.com/news/model-context-protocol
- [2] Next-Generation Database Interfaces: A Survey of LLM-based Text-to-SQL
- Hong, Z., Han, L., Niu, X., et al. (2024). Next-Generation Database Interfaces: A Survey of LLM-based Text-to-SQL. *arXiv preprint* arXiv:2406.08426.
- https://arxiv.org/abs/2406.08426
- [3] NL2SQL is a solved problem… Not!
- Floratou, A., Koc, M., Chang, L., et al. (2024). NL2SQL is a solved problem… Not! In *Proceedings of CIDR 2024*.
- https://www.cidrdb.org/cidr2024/papers/p74-floratou.pdf
- [4] SQL-PaLM: Improved Large Language Model Adaptation for Text-to-SQL
- Sun, R., Huo, K., Zhao, P., et al. (2023). SQL-PaLM: Improved Large Language Model Adaptation for Text-to-SQL. *arXiv preprint* arXiv:2306.00739.
- https://arxiv.org/abs/2306.00739
- [5] Startup Success Prediction and VC Portfolio Simulation Using Crunchbase Data
- Potanin, M., Trusov, B., & Beskrovny, P. (2023). Startup Success Prediction and VC Portfolio Simulation Using Crunchbase Data. *arXiv preprint* arXiv:2309.15552.
- https://arxiv.org/abs/2309.15552
- [6] A Fused Large Language Model for Predicting Startup Success
- Maarouf, A., Feuerriegel, S., & Pröllochs, N. (2024). A Fused Large Language Model for Predicting Startup Success. *EJOR, 322*(4).
- https://arxiv.org/abs/2409.03668
- [7] It's a people's game, isn't it?! Business Angels vs Machine-Learning Algorithms
- Blohm, I., Antretter, T., Sirén, C., Grichnik, D., & Wincent, J. (2022). It's a people's game, isn't it?! *Entrepreneurship Theory and Practice, 46*(4), 1054–1091.
- https://doi.org/10.1177/1042258720945206
- [8] The Power of Platform: How VC Platform Teams Contribute to Outsized Fund Returns
- VC Platform Global Community. (2024). *The Power of Platform: How VC Platform Teams Contribute to Outsized Fund Returns*.
- https://powerof.vcplatform.com/
- [9] Twitter Sentiment as a Weak Signal in Venture Capital Financing
- Tumasjan, A., Braun, R., & Stolz, B. (2021). Twitter sentiment as a weak signal in venture capital financing. *JBV, 36*(2), 106062.
- https://doi.org/10.1016/j.jbusvent.2020.106062
- [10] AI and Advanced Analytics in Private Equity & Venture Capital
- Boston Consulting Group. (2023). *AI and Advanced Analytics in Private Equity & Venture Capital*.
- https://media-publications.bcg.com/AI-in-Private-Equity-and-VC-BCG-2023.pdf
Implementation: The Model Context Protocol forms the standardised backbone of our agentic loop: tool calls, database queries and function executions are encapsulated as MCP messages.
Implementation: Defines the NL-to-SQL taxonomy used to translate investor questions into SQL queries.
Implementation: Reveals NL-to-SQL failure modes that drove our guarded decoding and fallback heuristics.
Implementation: Guided schema-specific instruction-tuning for accurate sector ranking queries.
Implementation: Serves as empirical grounding for our probability-of-exit model that powers forward-looking deal predictions shown in the product UI.
Implementation: Demonstrates multi-modal fusion (text + structured data) which we replicate when combining Crunchbase numeric metrics with founder-bio embeddings in our ranking pipeline.
Implementation: Provides evidence that ML-based scoring outperforms human judgement—supporting our decision to expose algorithmic rankings for both startups and VCs.
Implementation: Motivates co-investor network signals based on platform-team syndication graphs.
Implementation: Informs the social-signal extractor that augments LinkedIn talent flows.
Implementation: Sets benchmarks for sector dashboards and validates Historical Investment Patterns.