Farsight AI automates and augments financial services and insurance workflows with cutting-edge automation technology. We particularly focus on complex, nuanced use cases where risks of AI hallucination are particularly high. We offer unmatched accuracy, speed, and customization in mundane tasks directly within your native processes. Our proprietary tech, combined with your firm’s knowledge, opens new efficiencies in investment banking, private equity, hedge funds, and insurance. We drive significant time savings and improved employee efficiency through seamless integration with any data type or workflow.
External link for Farsight AI
2 Park Ave
20th Floor
New York City, NY 10016, US
Farsight AI reposted this
Fantastic New York Tech Week event with Mark Castleman, Greg Lavender (Intel), Frank Nydam (Tausight) and YJ Lin (Dell for Startups) Enjoyed speaking about AI adoption, use cases and cautions within financial services organizations, and building startups in general. Special thanks to citizenM hotels for the picturesque venue and for graciously hosting us! #NYTW #FarsightAI
Excited to be a part of this amazing cohort! Great first week with Mark Castleman Sharad Garg, PhD Reed Frerichs Kelsie Pallanck https://lnkd.in/dDgU6rMi
Incredibly excited to continue working with the team at StackAware. Their emphasis and expertise in security is paramount to responsible and secure LLM deployment (especially in financial services and insurance)
I help AI-powered companies manage cybersecurity, compliance, and privacy risk | NIST AI RMF and ISO 42001 expert | Harvard MBA | Marine veteran
Today I’m thrilled to announce StackAware’s partnership with Farsight AI. Farsight AI is a team out of MIT and Wall Street specializing in cutting-edge generative AI applications for financial services and insurance. A leader in large language model (LLM) optimization and AI agent implementation, Farsight helps AI-powered teams: - Efficiently tackle nuanced AI agent use cases like financial modeling, slide creation, and insurance submission prioritization - Reduce hallucinations and other undesired outputs - Ensure system reliability and accuracy across thousands of requests Teaming up with StackAware makes perfect sense considering our mission to help companies build their AI governance programs by managing risks related to: - Cybersecurity - Compliance - Privacy PROTECTING DATA INTEGRITY FROM ALL THREATS The accuracy of LLM outputs remains a key concern slowing enterprise AI adoption. Whether corrupted through malicious activity like prompt injection and data poisoning or by accidental vectors like hallucination, questionable data integrity can halt AI deployments in their tracks. StackAware helps build AI governance programs that ensure desirable output by: - Designing security programs to minimize the risk of corrupted model use - Penetration testing LLM-based applications - Developing policy guardrails for AI training Farsight AI complements these capabilities perfectly, optimizing artificial intelligence deployments through: - Retrieval-augmented generation (RAG)-based feedback - Adversarial synthetic data generation - Agent-based workflow optimization TOWARD LLM TASK AUTOMATION FOR FINANCIAL SERVICES ENTERPRISES Farsight AI CEO Samir Dutta added that “our ability to ensure LLM and AI agent reliability in complex, multi-step and nuanced workflows - combined with StackAware’s holistic approach to AI governance and security - will deliver huge value to customers in financial services. Accuracy, explainability and reliability are paramount in any tech solution, but even more so for those used in financial services.” Especially with increased scrutiny of AI use by the U.S. Securities and Exchange Commission (SEC) and other regulators, ensuring compliant and responsible deployments is table stakes. Ready to start building your AI governance program? https://lnkd.in/gyyu58_v Want to get in touch with Farsight? https://lnkd.in/g6M8Qwna
💀 Prompt engineering is a necessary evil. We recently open-sourced our in-house method for automating it (based on Google DeepMind research). "Large Language Models as Optimizers" proposed using LLMs as optimizers to identify prompts that maximize task accuracy. We've open-sourced our toolkit implementing this methodology and an article analyzing how well it generalizes to optimizing via GPT-3.5-Turbo. Perfect prompts for your LLMs with just one line of code, in much less time. Key Findings: https://lnkd.in/d7mhGkRE GitHub: https://lnkd.in/dbnBuhcg DeepMind OPRO Paper: https://lnkd.in/dXDY-Mzv
🚨Open-source models are now beating GPT-3.5-turbo on tasks in the financial domain. This is monumental for enterprise adoption of meaningful apps within financial services and insurance. At Farsight, we created an automated benchmark to evaluate LLMs’ question-answering abilities on company financials. We find: 🤗 Open-source model SOLAR-10.7B now eclipses GPT-3.5 ❌ LLMs have a 20-50% error rate, even when provided gold contexts 🤖 Calculations remain a major weakness for LLMs (GPT-4 has >40% error) Farsight AI leverages proprietary technology to boost performance of these models to production-grade quality in many vertical financial services and insurance applications. Check out our full post here: https://lnkd.in/eUmsYvUU Contact us at info@farsight-ai.com to learn more #AI #OpenSource #Finance #GPT4
The strongest sentiment in the AI community this week: the underwhelming nature of the GPT Store's public launch. We believe part of the disappointment is because most of these GPTs aren’t fully harnessing their most potent feature – external, self-hosted API integrations, or “actions”. This might be due to the recency of the launch, the sparse documentation available and a difficult debugging process. In any case, we've crafted a comprehensive tutorial to guide you through integrating GPTs with your own APIs, seamlessly and efficiently. This step-by-step guide is designed to demystify the process and unlock the full potential of GPTs in your projects. We leverage OpenAI’s GPT Builder, Flask and EC2 (or Heroku) to demonstrate how to deploy your own API and have a GPT reliably interface with it. Dive in to further explore how to make the most out of these amazing tools! #GPTStore #AgentFramework #FarsightAI https://lnkd.in/euBWW7XZ