☀️July 3rd 2025

The following features will be released on July 3rd on app:

1.	AI Twin Versioning	Live on Test
2.	Conversation Filters	Live on Test
3.	Calendar Interface	Live on Test
	Meet Scheduling	Live on Test
5	Proactive Tasks	Live on Test
6	Deleting an AI Twin	Live on Test
7.	AI Twin API	Live on Test

Ai Twin Versioning

As we know that AI Twins draw their intelligence from a Persona ( Behavior report) and Memories ( Intent & Mentions reports). However, the reports/assets that we create with consumr.ai are created at a moment in time. This also means that AI Twins used to have static intelligence till a few month ago. Hence, we came up with updatable twins that auto updates, since the user won’t know when and how to update a twin.

However, our interface was not designed to handle or manage these updates. To give you a few examples of situations: a. If a user has invested time in conversing with an AI Twin and it gets updated. Should the user continue the conversation with the updated twin, even though the history of the conversation was with an older version of the twin? If you have to ask the same questions with the new twin, the responses may be different.

b. If a twin in a group conversation is updated, how do we insert the new twin in the group conversation? c. If a user prefers to move on to chat with the newest version of the twin, should there be 2 conversations that shows up with the same name? Would it not create chaos? This release introduces a new versioning framework to address exactly those scenarios — allowing users to retain continuity, avoid confusion, and switch versions with intent. The system now tracks and displays which version of an AI Twin is active within every conversation. You’ll be informed when a Twin has been updated, and you’ll have the choice to either continue with the current version or initiate a new conversation with the updated Twin — without losing your previous threads or insights.

Here’s how it works for you:

• Version Switch Prompt: When a Twin is auto-updated, you’ll see a prompt in ongoing chats with the option to switch to the latest version.

• Conversation Branching: Choosing the new version will automatically branch the conversation — preserving history with the older version and starting a new thread with the updated one.

• Group Twin Sync: In meetings, if one or more Twins are on mismatched versions, the platform will suggest syncing all to the latest available version for consistency.

• Version Tags: Each Twin now carries a visible version tag within the chat or meeting context so users always know which version they’re interacting with.

• Clean Navigation: Even if multiple versions of the same Twin exist in different threads, the UI ensures they don’t show up as duplicate conversations, keeping your workspace tidy.

AI Twin Versioning now ensures that evolving intelligence never compromises your context, collaboration, or clarity. It’s a foundational improvement designed to support the dynamic nature of your consumer insights — while giving you full control over when, where, and how you interact with your Twins. Go ahead and take it for a spin. Let me know if you have questions.

Conversation Filter

In conversation, we have a list of AI twins that appear as names. However, we know that twins are not people that you spend time with, as colleagues or clients. Hence, it is not possible for a user to understand what the name stands for. In the past, we had put in efforts to create AI Twin cards, that show up when you hover on them. Yet, it is difficult to search for them just by their names or hover on each twin to know who they are. This conversation filter feature will help a user search and filter the twins as per their need.

As shown in the screenshot above, clicking an ‘expand’ icon next to the plus sign of conversation, will open a page wide pane, that will show the Twins, Agents and conversations in detail. You can use the search at the top to search for a name of a characteristic of a twin, agent or conversation. Alternatively, you can also use the hard filters below the search bar as filters.

The Twins, conversations and agents are grouped in accordion to allow you to better use the space and avoid a scroll. This prepares us to allow our clients to create as many twins as possible, their way and give them easy access to spot them with ease.

Calendar Interface

When we switched to the new design, our intent was to reduce the training hours for the users and to blur the lines between a real person and AI twins/agents. Since meetings could be executed adhoc and real time, the idea of calendar sounded like a forcefit. Products like outlook and teams have calendars to schedule and block people’s time for a meet and to provide all the information about the meeting within an event.

Since we came around to allow scheduling meetings and tasks (details coming ahead), calendar was not longer a misfit but a requirement. Hence, we will not be getting rid of the meeting tab on the menu and replacing it with Calendar, that will be the new house of Tasks and Meetings (scheduled or adhoc). Here you will be able to, view and edit a task or meeting, create a new meeting or task, view concluded meeting or task all in a unified Tab called “Calendar”

This is the biggest update of the release.

a. The Problem statement

There were 3 things that we were looking to accomplish:

ii. Since we are talking about scheduling, we wanted to extend the scheduling function to meetings as well, that can enable a user to execute a meeting and then build follow ups.

iii. We wanted to build in proactive tasks of varied natures, that lets the users create a task to be executed by our agents and schedule them for a foreseeable future recurring on weekly, fortnightly or a monthly level.

iv. We did not have an interface to integrate all the requirements seamlessly without impacting the trainability aspect for the users.

b. The solution

Drawing inspiration from MS Teams again, we realized that we could bring in the calendar tab in the menu and get rid of meetings. This is the new home for all meetings, scheduling and proactive tasks that happens. In short, let me explain how this will work: i. On the home page of calendar, you see the most obvious, A Calendar. On it will be events. Every event of the past will be an executed task or meeting of the past. Click on the event chip and the report of that event will appear right there. Any event scheduled in the future will show up on the future dates. If you click on the event chips, it will open the settings of the schedule. You will only see scheduled event chips out here.

ii. The new event button is where you can create a new event. It can be a meeting that we have in the current interface or a task.

Task is any activity that the system can conduct on behalf of the user, while the user is not actively logged in to the system. For example: Proactive recurring notifications. The future of tasks is what excites us all. Currently, we will only be launching one kind of task, called proactive task. Here the user will be able to select the scope of the tasks from, Regulatory, Competitor, Industry, Seasonality, Events and Randomized. Define and schedule it. These tasks will then be executed as scheduled and the user will be informed about it via emails.

Meet Scheduling

Meeting was an attempt to put a group of Twins and agents in one room and give them a problem with some guardrails to discuss and watch the solution unfurl in a discussion by the twins and agents using Observed and deterministic datasets that is created by consumr.ai. This process was a result of design and critical thinking by the team to replicate the nature processes of an organization and how meetings are conducted. Thus, this was Phase 1.

In Phase 2, we touched upon the next level. Recurrence and follow ups. Most productive meets have action items and follow ups. Objectives that require more bandwidth, focus and more number of meetings are not just meetings but Projects. In Phase 2, we wanted to replicate a natural process of smaller objective driven recurring meets and follow ups. Hence, Scheduled meeting was born. In this, you simply can visit an already conducted meetings and schedule it to recur repeatedly for a finite future for a similar objective or create a meeting and schedule it for later dates, before the first meeting concludes (irrespective of the results).

Proactive Task Scheduling & Delivery

AI platforms are striving to become proactive. Platforms like Perplexity shares daily and weekly notifications through the discover feature of their app. Though they are by and large just news reports, inviting users back to their platform. GPT tried this with a different method, by introducing tasks that a user can schedule, and we found them to be useful though limited. Inspired by there movements in the market, we wanted to extend the scheduling feature beyond scheduling of behavior reports and extend them to tasks that a user can follow up on, within the scope of consumr.ai. The most important objective of this feature is to provide the users with important nuggets of information weekly, fortnightly, monthly or perhaps even daily as follow up of a question the user had asked or an objective the user wants to accomplish. This will be done by 3 methods: 1. Create a proactive tasks from the calendar, with an objective in mind and a type of task and then schedule it to a recurring proactive task. You then wait for consumr.ai to look in to its knowledge bank and get the answer for the user frequently as per the recurring schedule set and deliver it straight to their inbox. This is also to give the user a feeling that while they were asleep, consumr.ai’s copilot was hard at work to bring him up to speed.

2. Going ahead if we see adoption of this feature, we will ensure that we are able to provide the users the option to do that for most things done on consumr.ai. Like a report, question or even a group discussion can be added as a recurring task, results of which can be directly sent to the user’s inbox as their daily, weekly or monthly dose of intelligence.

Deleting an AI Twin

In the converse tab of the menu, sits a list of AI Twins in conversation or AI Twins created but not conversed with in addition to Archived Twins. However, it is very easy to realize that some twins may lose its importance after making it or gets wrongly created. In such situations, it is important to reduce the clutter and give the ability to delete the created twins, as was the requests coming in from various clients. However, this feature comes with 2 conditions:

Twins created more than 24 hours ago

The bundles of manage, we have to define the limits on no.of AI twins that can be created per organization. If a twin is created more than 24 hours ago, is now needs to be deleted, an Admin can do so however, it will not replenish the Twin limit by 1.

Twins created less than 24 hours ago

If the admin of the organization tries to delete the twin in 24 hours of its creation, he/she would still be able to do it and the count of AI Twins will be added back to the AI Twin balance.

We have introduced this feature to ensure that user don’t create twins, interact with them, delete them and create another one as a process or a hack.

AI Twin API

This is our first attempt to create public REST based APIs that will allow external systems to integrate with our datasets. Through the AI Twin API, external systems will now be able to:

• GET a list of Twins in a portfolio

• GET Persona details of a Twin

• GET the underlying reports of each Twin

• Converse with a TWIN

What they cannot do:

• Create Organizations and Portfolios

• Create AI Twins

• Conduct Meetings

• Do Tasks

The API documents will soon be available on this link [Watch out for this space]

The APIs will also be available as a remote MCP server for external agents to interact with. The pricing for the API will also be published soon. API will need to be enables and the credits per organization can be assigned through the bundles in an organization and a report of its usage will be available in the activity log report.