ZIFF aims to help you pave a path toward greater automation within your organization.
SILICON SLOPES, Utah (PRWEB) March 28, 2018
ZIFF, Inc. announces an Unstructured Ai Database for image, audio, and video data with built-in indexing, search, training, and inference. ZIFF anticipates data consumption by automatically building out Ai models using the image, audio, video, and metadata you feed into it. An alpha release is already used in several deployments and will be made generally available in the second half of 2018.
ZIFF is unique in its approach to Ai. By focusing on empowering product visionaries and software engineers, ZIFF can help organizations fully unlock the insights and automation trapped within their unstructured data.
“Previous to starting ZIFF, Ben Taylor [Co-founder] and I had front-row seats to watch our previous CEOs fall in love with their businesses again when Ai freed them to think expansively about opportunities in the marketplace. At ZIFF we want to continue to help CEOs transform their business with Ai. We want visionary leaders to feel empowered to unlock the potential of their business using the data, talent, and skills their organizations already possess. That’s why v3 of ZIFF is literally a database; something all engineering folks know how to use, something so practical and useful that it can almost be forgotten,” said David Gonzalez, CEO of ZIFF.
He continued, “We have a problem in machine-learning, especially around unstructured data and deep-learning. In 2018 there are only two paths to Ai: pre-built models or point-and-click platforms. Pre-built models do not provide enough value to transform an organization, and point-and-click platforms still require too much expertise to manage the ceremony around generating accurately labeled datasets, training and validating Ai models, and figuring out APIs and deployment.”
“If you stop and think about it for just a second you know that most of the data running your business are unstructured and the highest value data in your products are unstructured. Think conversations, images, documents–data used by real people to do real work. ZIFF aims to help you pave a path toward greater automation within your organization. Because with Ai it’s conceivable that you could cost-effectively QA everything. If you could QA every important touchpoint what new opportunities would open up for you in your organization? Ai also holds the keys to making your best workers faster and more accurate. We want CEOs considering how Ai will transform their businesses when their teams move at 2x, 10x, or even 100x their current rate. Finally, Ai allows your products to do even more work for your customers–letting you remove friction points that impede product growth and adoption.”
“We have been working with product teams for a year now and watching them work to integrate Ai and what we find is no surprise; there is a dyad of productivity for getting Ai to deliver value within an organization, and they are the same folks who have been shipping value for decades and decades: product visionaries plus software engineers. These folks know how to deliver value, and we want to get out of their way. Most teams do not have a dedicated DBA [database administrator] because that role is not needed to create world-class applications and we are seeing that same trend play out with Data Scientists especially in the context of building products and core-services on unstructured data. That’s why this next version of ZIFF gets rid of everything that gets in the way of product or engineer productivity.”
“The situation with machine-learning and structured data is encouraging but structured data and machine-learning have a history of successes especially when you think about financial data, clickstream or e-commerce data, and transactions plus it is building off of 15 years of BigData efforts. Unstructured data is essentially BigData 2.0, and it almost feels like starting from scratch. Efforts to organize and pull value out of it are not very mature. Deep-learning makes it possible, but it is easy to feel overwhelmed. Even skilled Data Scientists get bogged down trying to figure out the GPU hardware, drivers, frameworks, and network architectures required to do deep-learning, and most projects never see the light of day,” said Ben Taylor, Chief Ai Officer at ZIFF.
Taylor continues, “Since we’re dealing with BigData 2.0 then we need Data Mining 2.0; that looks like deep indexing, search capabilities, clustering and grouping, scalable architectures; everything we count on today for structured data but able to digest terabytes or even petabytes of image, audio, and video data. ZIFF v3 finally addresses all the major pains we’ve encountered in delivering value with Ai in our careers, and we couldn’t be more excited.”
“At ZIFF we want to help executives push their business toward the transformations they dream about and to do that confidently. In my previous role as Chief Data Scientist, I had to address how to prevent models trained on biased data from propagating that bias. Thinking about Ai at that level does not happen on the first or second attempt in an organization, and we can help organization short-circuit that learning curve as well; this version of ZIFF is super flexible while enforcing best practices for training and validation.”
This new version of ZIFF will build on the success the company has had working with product companies and teams.
“ZIFF’s deep-learning based data curation is the indispensable tool I never knew I needed,” said Caroline Rowland, CEO New Moon
“HireVue has a world-class AI team and, HireVue's pre-hire assessments are a market-leading AI solution for identifying the very best talent – built from the ground up – but we regularly find opportunities to scale out our AI efforts using ZIFF. We have been able to do more iterations on problems and explore new ideas using Ziff that would otherwise be delayed,” said Loren Larsen, CTO HireVue
"Chatbooks helps people hold on to their memories by offering an easy way to turn digital images into photo books. Thanks to ZIFF, we’re able to make our apps so much more intuitive, smarter, and user-friendly. ZIFF has been helping our users quickly turn mountains of photos into beautiful books containing the images they most want to remember, and almost effortlessly. Together, we’re helping people celebrate all those little everyday moments that add up to an extraordinary life in print," said Nate Quigley, CEO & Co-Founder Chatbooks
"Chatbooks strives to simplify the otherwise complex task of creating photo books. ZIFF helps our engineering team create a fantastic user experience that powers next-generation applications, keeps us ahead of the competition, and perfects the details that matter most to our customers. With ZIFF, photo books are now built so seamlessly within the app that the experience can only be described as ‘magical,’” said Steve Bentz CTO & Co-Founder Chatbooks
With v3 of ZIFF, all the frustrating ceremony required to deliver value with Ai will be done automatically when you insert data or query it out. As a database, organizations can look to leverage ZIFF similarly to how they use their existing databases. Additional functionality specific to improving Ai time-to-value include:
Ai Assisted Curation
Product leadership can turn to existing domain experts to catalog and label massive datasets more securely, quickly, and accurately with Ai Assisted Curation than by turning to the crowd
Deep-Indexing & Unstructured Search
- Deep-indexing does for an organization's unstructured data what standard indexing does for structured data; it makes finding exact records or similar records fast
- Unstructured Search APIs give software engineers the ability to surface robust Image/Frame or Audio search with exact match or rank-ordered results
Image, Audio, Video + Metadata
Adding data to ZIFF is simple and all data ingested can be easily labeled with ground-truth/fact metadata (e.g. one or many target objectives) as well as structured metadata (e.g. relevant context) supported metadata types include:
- Binary, Multiclass, Continuous, or mixed targets
- Spatial for image data
- Temporal for audio data
Auto-tagging with pre-built Ai models
every instance of ZIFF comes with pre-built models that provide accurate tags for Faces, Things, even English Speech automatically and auto-tags can be used to filter queries and add additional value to your unstructured data.
SQL or JSON API with Real-Time, Batch, and Pub/Sub
- ZIFF Ai components work in the background presenting engineers with a familiar SQL interface compatible with existing software tools like their favorite ORM
- Data can be inserted real-time or batch
- Results can be queried real-time or asynchronously
- Network-layer heat mapping allows you to see and understand what is driving Ai models so you can better diagnose unwanted biases, facilitate human-in-the-loop efforts, or to create plain text or even unstructured data reason codes
- Fully managed model training & validation ensures best practices will be followed to train Ai algorithms
Image/frame and audio data can be intentionally anonymized through Ai managed downgrading that still allows for targets to be learned
With this release, ZIFF becomes the world’s only Unstructured Database. ZIFF is powered by Real Intelligence and Ai.
If you would like more information about this topic, please contact David "Gonzo" Gonzalez, CEO at ZIFF, Inc. - 801.669.2617 or email at gonzo(at)ziff.ai.
ZIFF is the world’s only Unstructured Database. Powered by Real Intelligence and Ai ZIFF empowers organizations to rapidly deliver the insight and automation trapped in their unstructured image, audio, and video data. ZIFF does not capitalize the “i” in Ai (Artificial Intelligence) because Ai still has little intelligence and at ZIFF we trust that when Ai earns it, it will capitalize itself. Previous to starting ZIFF Co-Founders David “Gonzo” Gonzalez and Ben Taylor worked automating machine-learning for Citizen Data Scientists and delivering superhuman HR insight with Ai at Big Squid and HireVue Respectively. Headquartered in lovely Utah, ZIFF is in the heart of Silicon Slopes. ZIFF was founded in 2017.