August 2024
App Release Date: 13-Aug-2024
Release Calendar
The release calendar for August 2024 is as follows:
TDE (Trusted Device Enrolment)
Live on Test
31-Jul-24
Mentions Overtime
Live on Test
30-Jul-24
Auto Notifications
Live on Dev
06-Aug-24
Behavior Influence in Intent
Live on Test
31-Jul-24
Co-Pilot Explore
Live on Test
30-Jul-24
AdGenius & AdGenius Tailored
Live on Test
30-Jul-24
AI Twin Enhancements
Live on Test
30-Jul-24
In-product assistance & announcements
Live on App
24-Jul-24
Brief Explanation
We will upload the supademo once the features are ready. We will also update the release notes then. For now, the intention of this release note is to convey what's coming up, to all.
TDE (Trusted Device Enrolment)
This feature will allow the user to enable the device authentication to log in to consumr.ai. This means that if a user uses face recognition to login to the laptop, as long as the device stays the same, he/she would be able to login to consumr.ai using the face recognition feature. This would enable us to increase the trust towards the user and enable a TOTP based authentication once in 3 or 6 months. The benefit is that the users will then be able to login to the platform quickly and will be more motivated.
Mentions Overtime
Like Behavior trends (a.k.a Behavior over time) we will introduce a scheduling option for Mentions insight.
Auto Notifications
We will be adding an auto notification feature that will be auto triggered with a condition is satisfied. It will then send an email to the users motivating them to login to the platform.
The objective is to increase logins by 3x from end of August to Mid Sept.
Behavior Influence in Intent
In advanced options of an intent report, you can add a behavior report which will "influence" the intent report. In the sense it will tilt the keywords more aligned with the behavior report attributes. For example, if the behavior report has primarily young females interest in music, then keywords like: concert shoes, women shoes will pop up more.
Co-pilot Explore
Co-pilot has 3 modes: Analyze, Investigate and Explore. While Analyze and Investigate has gone live on July 16th, 2024, Explore is still pending. If you don/t know what explore is, refer to the last release notes.
AdGenius & AdGenius Tailored
We did a press release on this last month. While the algorithms were developed and some portions were integrated into the front-end product not all were.. Let's understand what this is and what it means for our clients.
AdGenius: This is a fine-tuned proprietary model of consumr.ai. It has been developed keeping in mind how high performing ads on social platforms perform. This model is not trained on any of our client's data. It's dataset is scraped posts/ads across social platforms. This will enable us to create better creatives in planning and improve. This was already pushed from the backend and in fact the TCH creatives were born of this in the backend.
AdGenius Tailored: This is a client-specific fine-tuned model. This is found under "Creative AI models" in the manage assets section of the organization. This is a paid feature. Once the client initiates this, we create a bespoke fine tuned model specific to the best performing creatives based on the actual performance of their ads run in the past year. This enables a significant uplift in the new creatives being created right off the gate.
AI Twin Enhancements
Better chat memory
Now supports persona creation from TikTok, GA4, Snap in addition to Meta.
Enhancements in the Sidebar
In case of Meta Advanced AI reports, AI Twin names will match the name in the audience summary card for a more consistent experience. We will strive (not a guarantee) to have de-duplicated names in the audience summary cards for a particular portfolio.
To increase trust and transparency, we have added two options that can be accessed via the "three dot" menu next to each answer. These two options are: Check consistency & See Trace. Check Consistency: This helps you cross check if the answer provided by the AI Twin, is consistent with its persona and memory. See Trace: This helps you ascertain how "real" and "grounded" the answer is based on the reports it was built on. So if you want to verify a claim that it has made you can use this. Of course, its silly to use this for tracing which song you like, cause that wont be present, however, it might trace it to a particular music genre possibly.
Last updated