Platform Updates: February 2025

Introducing the Mosaic AI Analyst - A Resource for Building and Updating Mosaic Models Through GenAI

We're excited to reveal a new resource available to all users with AI-features enabled in Mosaic - the Mosaic AI Analyst.

Leveraging the power of GenAI, we have introduced an AI Agent into the Mosaic platform that can assist you in building or updating your Mosaic models through chat-style instructions capable of remembering the context of your modeling task and taking feedback / making updates as required:

Let's explore a few key use cases:

1. Subsequent Image Uploads to Vision. The idea for Mosaic's AI Analyst came from our users who wanted to upload additional financial information into Mosaic Vision past the initial financial forecast page. Below is an example of a user uploading NWC Balances information from another page in the CIM to update their Mosaic operating model.

2. Adding Structural Complexity to a Model. Mosaic's AI Analyst can also help you update existing operating models by making multiple changes at once that become tedious to do manually. For example, tasks that would require adding multiple rows, updating formulas, and applying growth and margin assumption changes across several years can be done more efficiently through chat.

Below is an example where we ask Mosaic's AI Analyst to update an operating model by splitting its revenue into three segments of different size and growth rates. In the example below, you can see that at first, the AI Analyst successfully applies the split, but fails to apply the growth as instructed. A quick follow-up message remedies it, however, as the AI Analyst remembers the context of the conversation you're having with it:

Data Clarity Unlocked by Attribution

As Mosaic matures from a top-of-funnel efficiency booster to a system of record for new deals analysis, multiple deal team members are collaborating on individual deal models and more sources of third-party data are being incorporated into Mosaic models (e.g., Capital IQ, FactSet, Mosaic AI, etc.).

To ensure perfect clarity into the source of each unique datapoint comprising a Mosaic model, we've borrowed a concept from the engineering world and brought it to the world of dealmaking.  The concept is "input attribution" - Mosaic now tracks for each datapoint in each model (i) who entered the data and (ii) when it was entered. At the time of writing, data could be entered into Mosaic by:

1. A Human Deal Team Member. An assumption entered by a user updating a Mosaic model will be tagged to them, displaying their avatar.

2. Your Mosaic Defaults. Firms and individuals on Mosaic can set defaults for minor assumptions (e.g., entry transaction fees as a % of TEV, corporate tax rate, etc.).

3. Mosaic Vision. Historical and projected financial data ingested by Mosaic Vision, our AI "CIM Reader" will be tagged accordingly.

4. Capital IQ & FactSet Data. Data sourced from Capital IQ and FactSet will not only reflect which data source the figure came from, but will also provide a link that will take you directly to the source "tearsheet" that the underlying datapoint was sourced from.

5. Mosaic AI Analyst. Any inputs set by Mosaic's AI Analyst (described above) will be attributed to it accordingly:

Let us know what you think 📣

Mosaic has been built over the years by the collective intelligence of the private equity industry - specifically, by you, our users - through your thoughtful product feedback.

Your feedback is instrumental in helping us build a product that you'll continue to love far into the future. Please reach out if there is anything we can do to make your Mosaic experience even better.

That's all for now! Shoot us a note at support@mosaic.pe if you can think of anything else you'd like to see in Mosaic!