Amazon and PBS have entered a partnership that will bring content from the public media operation to Prime Video. More than 150 local PBS channels and the PBS Kids Channel will launch as a free ad-supported TV (or FAST) offering on Prime Video over the coming months. The from PBS noted that this is the first time this collection of programming will be available on a major streaming service for free.

PBS Distribution is also launching two new FAST channels that will be available exclusively on Prime Video for a limited time beginning November 26. These channels are PBS Drama and PBS Documentaries. It seems Amazon is looking to focus on a lineup of FAST channels within Prime Video for free viewing, since the company announced that it is .

Having yet another place to watch public media content is a happy development for PBS fans. The broadcaster recently launched a FAST channel with Roku called PBS Retro, specifically with shows from the 1980s and 1990s for those of us who want to dive into a little nostalgic escapism.

At this year’s annual BUILD conference, data architecture giant Snowflake went all in to give its customers advanced capabilities, including some long-previewed features, to easily mobilize their datasets to build and share powerful AI applications. 

The company debuted new tools for Cortex AI, its fully managed offering for developing conversational AI apps grounded in enterprise data hosted on its platform.

It also announced Snowflake Intelligence, enabling users to create ‘data agents’ that could not only answer questions related to structured (organized in tables) and unstructured data (PDFs, documents, etc.) on the platform but also take action across third-party platforms like Salesforce and Google Workspace using the generated answers.

Below is a rundown of all major announcements:

Cortex AI enhancements

Ever since its introduction last year, Cortex AI has been receiving regular updates from Snowflake to simplify how developers create and run AI apps.

At BUILD, Snowflake continued to bolster this offering with new multimodal input support for apps in development, managed connectors to integrate internal knowledge bases to the apps, and knowledge extensions to tie third-party documents, like news articles, to the services.

The company also announced Cortex Chat API combining structured and unstructured data into a single REST API call for fast-tracked RAG and agentic app development; observability for the developed AI apps (building on the TruEra acquisition); and support for SQL Joins and multi-turn conversations in Cortex Analyst to unlock richer insights from structured data.

Snowflake Intelligence

Using the enhancements to Cortex AI, including integration with internal knowledge bases, the company announced Snowflake Intelligence, a unified platform enterprises users can use to build ‘data agents’. T

The agents will use Snowflake-hosted business intelligence data as well as that connected via third-party platforms to provide users with instant answers to their business questions. 

Further, once the insights are produced, the users can ask the same agent to act on them across integrated third-party tools.

This could involve a wide range of tasks across third party apps, from automatically and autonomously creating an editable form in Google Workspace using the generated insights to modifying an entry in Salesforce CRM.

Open Catalog, Document AI enhancements

Back in June, during its flagship summit, Snowflake and its industry partners unveiled Polaris as a vendor-neutral catalog implementation for indexing and organizing data conforming to the Apache Iceberg table format.

The offering has since been open-sourced and donated to the Apache Foundation. At BUILD, the company took a step ahead and debuted a fully managed, hosted version of the catalog called Snowflake Open Catalog. T

Now generally available, Polaris helps enterprises grow and evolve by integrating new engines and applying consistent governance controls. 

In addition, Snowflake also announced the general availability of Document AI, the product it offers to let users extract data from unstructured documents like invoices, on AWS and Microsoft Azure.

Threat prevention and security monitoring

In light of the recent customer data breach, Snowflake has taken multiple steps to bolster the security of its users, including enforcing multi-factor authentication by default.

At BUILD, the company continued this work with the introduction of Leaked Password Protection, a capability that will automatically detect and notify customers if their Snowflake credentials have been exposed on the dark web (much like Google).

According to Christian Kleinerman, the EVP of product at Snowflake, the company may even go and disable the accounts with compromised credentials for account protection.

In addition to this, Snowflake announced a new Threat Intelligence Scanner Package for its Trust Center, the place where users see how well their accounts are configured.

The feature will provide users with a risky user view, giving them the ability to detect when a potentially risky user is active along with the best steps to deal with the situation.

Snowflake’s Trust Center is also getting extensibility, which will enable third-party partners to leverage Snowflake’s native app framework and add additional checks and assessments to the dashboard.

Snowflake BUILD runs from November 12 to November 15, 2024.

A 2018 Supreme Court ruling eliminated the requirement that an e-commerce retailer needed a physical location in a state in order for said state to be able to collect sales tax on purchases made by residents. While the decision was a boon for states, it created a headache for e-commerce sellers.

Kintsugi is looking to offload and automate calculating and filing sales tax for companies. The San Francisco-based company’s AI technology connects to a company’s billing and payment systems and figures out in which states they are liable to pay sales tax. It then registers users in the correct states. From there, the system can automatically calculate and remit what a company owes in sales tax to keep companies in compliance.

Kintsugi raised a $6 million Series A round earlier this year led by Link Ventures that valued it at a $40 million post valuation in April. The company has since reopened its Series A round, taken on additional $4 million in capital led by Airwallex, and has doubled its valuation to $80 million.

Pujun Bhatnagar, Kintsugi co-founder and CEO, said that he got interested in the sales tax space while working as a senior machine learning engineer at Meta in 2018. Bhatnagar told TechCrunch that both his father and grandfather worked in taxation their entire careers. Bhatnagar found himself in 2018 wondering what he wanted to do with his life. It just so happened to be around the same time as the Supreme Court ruling, which opened up a whole new market that was worth exploring, he said.

“It’s basically an amalgamation of 52 different types of little countries, which have their

own laws and jurisdictions when it comes to local governments,” Bhatnagar said. “And 48 of these jurisdictions have sales-tax-related laws.”

To really understand the problem, Bhatnagar said he started doing sales tax for e-commerce and SaaS companies by hand for a year and a half to really understand the pain points before writing any code. He made Kintsugi’s first few employees calculate sales tax by hand, too.

From there they built a platform and algorithm to modernize and automate sales tax compliance. Bhatnagar said that building the model in-house has made their results more accurate than competitors that rely on large, all-encompassing language models. He said that the company keeps humans in the loop to monitor for accuracy, too.

The company was formally founded in 2022 and launched its website in August 2023. Bhatnagar said inbound interest was immediate, and Kintsugi has been able to grow its customer base to more than 1,100 users in the past year. It has earned $1 million in revenue.

Kintsugi isn’t unique in wanting to modernize the sales tax process for companies. Competitors include Anrok, which has raised more than $50 million in venture money, and CereTax, which has raised $19 million in venture capital, in addition to numerous legacy companies that outsource the process to folks in countries like India. This is the same type of work that Bhatnagar’s family had worked on.

Bhatnagar thinks that part of the reason demand has been so high for Kintsugi is its approach to landing customers. The company allows potential customers to sign up for free and test out whether they like it. If they choose to proceed, they can pay $100 per tax filing or create a custom plan. Bhatnagar added that some of their competitors charge hefty fees just for onboarding to their platforms.

“We are the only company in the space that has a ‘get started’ button, that has a ‘we will do your sales tax analysis for free [button],’” Bhatnagar said. “And that’s not going to be just done once. You can create a free account, and every seven minutes the report is going to be updated for you. And that’s a value prop that we want to provide to founders for free, even if they decide to not pay a single dime for Kintsugi.”

Kintsugi plans to use its new capital to keep expanding its tech and to help the company gear up to expand into Canada and Europe.

“We are a bunch of nerds,” Bhatnagar said. “We are not trying to sell any snake oil. Connect your data, see the results.”