I have been creating podcasts on and off — mostly off — since 2007, when Paul Gillin and I came up with the idea to talk to each other about war stories from the tech PR world. (You can listen to the very first pod here.) That series would eventually evolve into several different pods that Paul and I would do over the years, with the most recent episode here. In between these shows, I would freelance pods to various clients and publications such as eWeek and various IDG ones.

I tell you this because today in the span of a few minutes, I managed to create some very credible podcasts out of previously just my written content, using a new Google tool called notebooklm.google.com. You upload your documents (PDFs or text files) and it converts them into two-host conversations that use some of it, along with using AI to bring in other information. The two “hosts” sound great: one is a male voice and the other a female voice. Call them Dick and Jane. The AI adds in almost the right amount of temporizing with “ums” and “likes” and back-and-forth byplay into the conversation. You can download the audio files here and hear it for yourself:

• I wrote a handbook for CSOonline recently about AI security posture management. Here is the pod:

• I also wrote an article for Internet Protocol Journal about the history of the Interop Shownet. Here is that pod:

I did almost no additional work to create these pods, other than search my own hard drive to find something that I wrote. Both of these samples are about ten minutes long. And while I wrote every word in both articles, the pods use examples that I never wrote (that were actually quite good) and bring in other information.) Using their ML routines to keep things more conversational works reasonably well and you almost believe that Dick and Jane are two live humans talking to each other about something that they “just read.” I could do with a few less inserted “likes” which seem to be the basic conversational building block of a certain generation. One thing that Google hasn’t coded into its system is to have the two hosts talk over each other, which I find annoying on other pods that have multiple human hosts.

Google’s tool is still very much in the experimental stage, but it is free to try out. In addition to creating podcasts, you can also query the content you upload just like any AI system, and it will also provide a summary and FAQs and other supporting things around your content. I would suggest that you don’t upload any private content however.

What does this mean for podcasters? Well, uh, things are going to get very interesting. While the Dick and Jane voices aren’t yet configurable, they are pleasant to listen to and seem 85% human. It also portends that my podcast business is probably dead in the water, not that I ever relied on it to produce any significant revenue. Given that I don’t cultivate any political outrage, or any outrage (other than from non-working tech or over-promised products), there was zero chance that my podcasting career would ever take off.

If you do produce some pods that you would like me to listen and compare to the original source materials, do drop a note in the comments.

I chat with my old colleague Keith Newman whom I have know for decades on his podcast The Look Back. We chatted about our overlapping times at CMP and Ziff-Davis and a few other places as well.

Paul and I spoke to Chris Matty, the co-founder and Chief Revenue Officer at Versium, Inc. His company is developing better B2B ad tracking technologies that will ultimately be used when the third-party web cookie finally bites the dust next year.

As with so many online technologies, replacing cookies might require a lot of work from advertisers and web publishers. This is because Google, Facebook, Apple and Amazon all have a vested interest in keeping customers within their “walled gardens” and not necessarily sharing their tracking data with others. The great cookie demise will bring about a series of consequences, some intended and some unintended.

For example, there will be an initial rush for advertisers to make use of first-party data (meaning data that they have collected over the years themselves) until they realize that this data is outdated or inaccurate and can’t really provide the sufficient quality or insights or a path towards eventual purchases that the old cookies had. There will also be an adjustment as advertisers realize that reaching B2B customers is a lot more difficult than reaching consumers because many business customers don’t necessarily identify themselves as such — think of all the LinkedIn accounts that carry Gmail addresses as an example.

The work-from-home movement has increased the complexity of the tracking business customers now have different IP addresses or are hidden behind VPNs, so all that geofencing and IP tracking data is out the window! Versium is attempting to resolve these issues by aggregating anonymous data from a variety of sources to profile website visitors without compromising their privacy. Resolving identity means collecting and matching deterministic data that allows a marketer to reach or contact a specific person, such as email, phone numbers, addresses and device IDs. For example, think of trying to ensure you have identified the same person when sometimes they call themselves Bob Smith, sometimes Robert Smith, and in other cases they show up as @rsmith. Versium believes that’s possible in many cases using independent, opt-in sources.

The company is working with a variety of independent publishers and advertisers to consolidate data assets to allow independent publishers and site owners to better compete with the internet giants. The goal is to achieve personalization with privacy protection.

Chris has written extensively on this topic here. “Companies that deploy identity resolution solutions to optimize and leverage data can take back the control they had once ceded to third-party cookies,” he asserts.

You can listen to our 16 min. interview with Chris here.