Envision tapping into your data’s full potential with AI so advanced, it transforms how you operate. Deploying #LLMs with #RAG isn’t just a step forward—it’s a game-changer. However, while initial results can be impressive, there’s an underlying complexity that most overlook—model drift, variance between different models, and the challenges of ensuring accuracy over time. Our experience with these issues has led us to pioneer solutions that combine the power of probabilistic AI with the precision of traditional engineering—a concept we call RAG-V. Explore more in our CTO Joe Chang's latest blog: https://lnkd.in/gePiSd5w
Primer.ai
Software Development
San Francisco, CA 28,937 followers
Decision-ready AI for the world's most critical organizations
About us
Primer exists to make the world a safer place. We do this by building trusted decision-ready AI for the world's most critical organizations. Our software enables leaders, operators, and analysts to better understand the changing world around us in real time, and make informed decisions when the stakes are high.
- Website
-
https://primer.ai/
External link for Primer.ai
- Industry
- Software Development
- Company size
- 51-200 employees
- Headquarters
- San Francisco, CA
- Type
- Privately Held
- Founded
- 2015
Locations
-
Primary
244 Jackson Street
201
San Francisco, CA 94109, US
-
3100 Clarendon Blvd
Suite 925
Arlington, Virginia 22201, US
Employees at Primer.ai
-
Joe Chang
CTO
-
Jason Franklin
MD at WVV Capital, ex-Investment Partner @ a16z, Staff Scientist @ Stanford CS, CMU CS Ph.D., Allen Newell Award Winner
-
Brent Williams
Technology and Security Leader | Strategist | Board Member | Advisor
-
Robert M.
Proven team leader, strategist, multi-disciplinarian. Pragmatic AI / ML for government, national security, & defense. Multiple startups, multiple…
Updates
-
Excited to welcome Matthew Macnak as our new Senior Vice President of Customer Solutions Engineering! Matthew started his career as a US Navy Submarine Officer, where he completed 6 Strategic Deterrent Patrols aboard the USS ALASKA GOLD and served his final tour on OPNAV N3/N5 staff. After transitioning from active duty, he led engineering teams in the defense industry, focusing on the design, testing, and deployment of #C4ISR systems for the U.S. Department of Homeland Security and the US Army. Most recently, Matthew led teams at DataRobot, Amazon Web Services (AWS), and Instabase, helping government agencies adopt AI/ML technologies to enhance mission outcomes. Join us in welcoming Matthew!
-
-
Primer.ai reposted this
Two weeks ago, National Institute of Standards and Technology (NIST) published a report on risk management of generative AI. The report lays out a dozen risks that cover everything from the risk of insulting people with offensive language to helping terrorists design bombs. The one that stands out is the risk of hallucinations. As the report defines, it is "the production of confidently stated but erroneous or false content." NIST is spot on. Fixing hallucinations is the top priority for the AI community, both industry and academia. Thousands of papers have been published on the subject this year alone. My own team at Primer has been laser focused on detecting and mitigating them. It's a solid report and very readable even for non-experts. I especially like the subtle points made about how hallucinations can sometimes be desirable, for example if you're asking the generative model to be creative. One detail did catch my eye. NIST uses the term "confabulation" rather than "hallucination", noting that some commenters say "the terms 'hallucination' and 'fabrication' anthropomorphize [generative AI]." It's a fair point, though its debatable whether "confabulation" is any less anthropomorphic. We should all settle on whether to change the term soon. Even Merriam-Webster Inc. has weighed in, defining hallucination as a type of AI error. There may be no going back at this point. NIST report: https://lnkd.in/g4db_SNR Hallucination papers: https://lnkd.in/gxgNKB3G Primer's latest progress on hallucination mitigation: https://lnkd.in/gAR23c3G Merriam-Webster definition of hallucination: https://lnkd.in/gv3jsy_w
Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
nvlpubs.nist.gov
-
More and more #AI software rely on large language models (LLMs) hosted in the cloud, but this can be a significant limitation for users in denied, degraded, intermittent, or limited (#DDIL) environments. These users need reliable AI tools just as much as—if not more than—typical military users. Many tasks performed by #LLM such as classification and summarization can be effectively handled by a smaller language model trained for the task. Most of the work done by AI software does not require an LLM at all. Learn about the advantages of small language models: https://lnkd.in/g3sakjQw
AI robustness for denied, degraded, intermittent, or limited (DDIL) environments
https://primer.ai
-
Last month, the Army Chief Information Officer issued crucial guidance on #GenerativeAI and Large Language Models, emphasizing the importance of human oversight to mitigate AI hallucinations. Human-in-the-loop review before any significant decision is vital. But Gen AI systems should also leverage human-inspired critical thinking. "Chain-of-thought" (CoT), a method from a 2022 Google Research paper, forces Gen AI models to think through problems step-by-step. This has two advantages: first, it reduces hallucination errors, making recommendations more accurate. Second, it aligns with the US Army CIO's guidance by providing clear reasoning behind #AI outputs. Learn more about CoT in our latest blog by John Bohannon, VP of Data Science at Primer: https://lnkd.in/g4hGxirH
Chain-of-thought is a secret weapon against hallucinations
https://primer.ai
-
Imagine relying on #AI that fabricates information 5-10% of the time. For intel analysts and decision-makers, this isn't just a nuisance—it's a huge risk. At Primer, we understand the stakes and developed an automated fact-checking pipeline that catches over 99% of errors before they reach the user. See how Primer ensures reliable AI and prevents #hallucinations: https://lnkd.in/gJUGX33D
How Primer prevents hallucinations
https://primer.ai
-
We're thrilled the US Navy has added AWS Marketplace to its Enterprise Software Licensing program, simplifying the experimentation and procurement of next-generation #AI technologies. At Primer.ai, we're committed to bringing trusted, reliable AI to our Sailors. For agencies looking to experiment and derive insights from unstructured data, a cloud-based version of our AI-powered search and discovery tool, called Primer Delta Lite, is available in AWS marketplace. With Primer Delta Lite, you can explore your unstructured text data within the comfort and security of your AWS environment without needing to enter into a contract. We greatly value our partnership with Amazon Web Services (AWS), which enables us to deliver AI solutions with the scalability, security, and reliability that our customers require. Read the full announcement of the US Navy's AWS Marketplace initiative: https://lnkd.in/gzJWyMqd Check out Primer Delta Lite on AWS marketplace: https://lnkd.in/gHxyk9HF
The Department of the Navy adds AWS Marketplace to its Enterprise Software Licensing program | Amazon Web Services
aws.amazon.com
-
We’re thrilled to welcome Erin O'Brien Hawley as our new Chief Revenue Officer and Joe Chang as our new Chief Technology Officer! Erin brings over 30 years of experience in federal tech sales, including leadership roles at Instabase Gov, DataRobot, and Cloudera. Joe brings over 25 years of experience leading engineering teams in the creation of new products based on emerging technologies, and previously served as CTO of Uber Freight. Prior, Chang co-founded Marin Software, a publicly traded SaaS provider of real-time analytics and machine learning algorithms. Join us in welcoming Erin and Joe! https://lnkd.in/gRHkHEY9
Primer names new CRO & CTO to executive leadership team
-
Attending Army Data Summit next week? Join us at booth #512 to see how our AI solutions help you harness public and proprietary data at scale and in real-time —with unprecedented speed, power, and accuracy. . . Meet our team members: Brian Gershkoff, Sarah Girard US Army, Army Chief Information Officer
-
-
Today marks the launch of our new website! At Primer, we're dedicated to building trusted #AI to help decision-makers, operators, and analysts gain insights faster for superior decisions. While our mission remains the same, our rebranded website simplifies navigation, learning and exploration of our AI solutions while offering a fresh new look. We hope you like it. Explore the new site: https://lnkd.in/gXshy3zT
Explore primer.ai
https://primer.ai