Future-proof your AI strategy

Greetings Cohere Community,

Time for a quick reality check. Generative AI for enterprise holds great promise, but realizing its potential requires thoughtful implementation. What may work for small proof-of-concepts, may not work at scale or for the long haul. Consider how run-time costs, latency, hosting, and data security impact the overall solution and ROI. As AI technology continues to advance rapidly, businesses need the flexibility to adapt solutions to their evolving needs.?

At Cohere, we take a pragmatic, holistic approach, so companies can deploy generative AI without being locked into any one model or cloud solution. Our large language models (LLMs) run on all major cloud platforms, including AWS, OCI, GCP, and Azure (coming soon) , and our suite of fine-tuning capabilities offer businesses more flexibility for building secure custom AI applications.?

But how do we ensure that this technology is safe regardless of where it’s deployed??

Cohere’s Head of AI Safety Seraphina Goldfarb-Tarrant, and technical staff member Maximilian Mozes, attempt to answer this question in The Enterprise Guide to AI Safety. This blueprint examines seven potential harms and the latest mitigation methods to combat imminent threats from AI.?

Biases and inaccuracies plague current generative AI, highlighting limitations and the necessity for greater transparency into these models before broader implementation. Research lab Cohere For AI is tackling the data transparency crisis with the launch of the Data Provenance Initiative , a multi-disciplinary collaboration with 11 institutions. The group focuses on the poorly documented and geographically skewed datasets used to train foundational models. The initiative launched the single largest audit to date of AI datasets and created the Data Provenance Explorer , a tool to explore over 2000 open datasets that have been tagged to include original data sources, numerous re-licensings, creators, and other data properties.?

So, how then do we build and deploy AI responsibly?

To further efforts, governments continue to set boundaries for how AI should be built and deployed responsibly . Alongside other leaders in the AI space, Cohere’s President Martin Kon attended the signing of President Biden’s Executive Order on AI. While across the pond, Cohere’s CEO and Co-Founder Aidan Gomez attended the UK’s AI Safety Summit. Speaking with Bloomberg News , Gomez doubled down on the need to focus our attention on the immediate and very real threats of AI instead of sensationalist theoretical futures. These first regulatory steps aim to strengthen governance, advance innovation, and manage risks. More is sure to follow.

And crucially, how do we implement these concepts in practice today??

As the AI ecosystem evolves and enterprises adopt generative AI, success will hinge on practical orchestrated execution. For example, creating a robust, enterprise-scale retrieval system involves advanced LLMs, like Embed v3 for effective search, and capabilities like fine-tuning for optimal performance. Companies such as BlueDot have significantly improved accuracy , exceeding 90% in real terms, by leveraging these integrated, fine-tuned AI models.?

At Cohere, we believe this holistic approach to building AI solutions for enterprise will steer AI in a direction that is scalable, responsible, and reliable.

For more, check out our latest stories or read on for this month’s highlights and upcoming events!

Product?

Our most advanced embedding model, Embed v3 , is now available! Offering state-of-the-art performance per trusted MTEB and BEIR benchmarks. Try it now. ?

For Business

Ready for enterprise-grade fine-tuning? Our comprehensive suite of fine-tuning capabilities lets businesses deliver custom-made AI solutions efficiently. Learn more .?

Developers

Learn how to build a chatbot from scratch using Chat with retrieval-augmented generation (RAG) in our latest LLM University course .?

Research

Last chance to contribute to Aya, an open-source multilingual model led by the largest open science AI initiative involving 100+ researchers, including our independent research lab, Cohere For AI . Contributions will be accepted until Dec.15th. Contribute today .

Company

What’s in the news? Cohere is gearing up for the new year and announced plans to double our London office headcount in 2024. Visit our newsroom. ?

Upcoming events with Cohere?

• Nov 27 - Dec 2 [Las Vegas]: Cloud enthusiasts at AWS re:Invent swing by Booth #124 or book a meeting with our team.
• Nov 29 [Online]: Join Cohere For AI in a fireside chat with Google Deepmind Senior Staff Research Scientist Orhan Firat.
• Dec 6 - 10 [Singapore]: Empirical Methods in Natural Language Processing (EMNLP) will cover all things NLP. Drop by Booth #11!
• Dec 10 - 16 [New Orleans]: The Neural Information Processing Systems Foundation is back with NeurIPS . Join us at our booth #1109.

For all upcoming events, explore cohere.com/events .?

Explore what's possible in Cohere's playground. Try it now.