Conversation Recommendations

Our interaction with technology is changing. It is becoming more ambient, and more objective focussed. The way we search has drastically changed in the last 3 years (Remember the last time you went into an article to look for information and did not directly ingest it from the AI synopsis). Very soon, our agents will be helping us work across different platforms: ordering food, navigating maps, booking cabs, and every other thing.

And as much as it is going to help humans, it is going to change ads. Change its shape, structure, and the involved entities. Too early to comment on how it will reshape the entire industry. But we are surely the transformation will be led by the following two domains:

1. Changing Computing Interaction → A change in human’s computing interaction changes the real estate and the delivery of ads. Search shall no longer be the biggest trigger of ads.
2. Ever-Denser World Model → With tighter integration between humans and machines, computers shall have more data to target the promotions enabling high intent.

One major space where this change ripens possibilities is the conversational bot segment. Tons of these bot applications came up after they were popularised by Character AI, and Replika. Many humans already treat these bots as their girlfriends. These bots because of their dense understanding of the human can place promoted recommendations in the chat. These promotions are of high intent, and appeal personally because they come from a human’s favorite pass time friend - the bot.

Types of conversational ads:

1. (Observation-Based) - You've been skipping meals again. I found this service that brings you the kind of comfort food that makes you smile - you know, like that noodle place you love?
2. Explicit Discussion - Sorry to bother you, but remember that writer's retreat we talked about? They just opened registration for their virtual sessions. It's the one with ocean-sound meditation between writing sprints. It's early, but spots are filling up. I've already checked - the schedule works perfectly with your work calendar. Should I grab you a spot?
3. (Question-Based) - (on being asked about restaurant recommendations) I've got several great options. There are three restaurants that are promoted in my search - Osteria, La Maison, and Sushi Roku. They all have excellent reviews, but I should mention they pay for priority placement. There are also some wonderful non-promoted places like that small Vietnamese restaurant you loved last month.

‍

One other previously unheard area where ads will emerge is spaces. We envision that all our spaces: cars, retail stores, cafes, offices, warehouses, and home will have their own OSs which orchestrate the working of different systems in the space. These OSs work together with Humans to achieve the respective objectives of the spaces. These OSs have a very deep perception of their own world and are best equipped to place promotions for these areas. Imagine in a meeting room, a problem is discussed and the OS of the meeting room suggests promoted recommendations for the problem to the person in charge.

We have built our own ad engine to serve this use case. Though started by serving ads for our meeting room product, we soon realized it could be placed across any conversational agent. It is a privacy-first SDK that integrates with any chatting platform and enables them to put context-based recommendations. It is currently working with different parties to reach beta testing.

Though this is an experiment, we are proud to say that we are amongst the early ones to experiment in the agentic ad space.

‍

----------

Previously published at https://meera.3102labs.com/memo/conrecs-memo

‍