AIR Platforms
Research · Product

The AIR Platforms June Release is Here

AIR now lives inside the ChatGPT and Copilot Agent Stores, our specialized bank model goes live, and AIR Agent supports follow-up questions on completed scenarios — built for the speed of modern credit.

Before the Fourth of July weekend takes everyone's attention, a quick note on what we shipped this month, with a few things I think you'll actually want to use.

Most of my career has been focused on ensuring myself and my team members are able to make data-driven decisions that drive value for the business.  One thing that always stuck with me is how much analyst time gets burned not on judgment but on plumbing. Pulling financials, reconciling names, re-running the same scenario from scratch because you thought of one more question after you closed it.

What raises the stakes is who is asking now. It is no longer only the analyst running the model. The pointed questions are coming from LPs, and from the top of the house, where conviction on a name gets pressure tested from every seat that carries risk or answers for it, and none of them are willing to wait weeks for an answer. For insurers the clock is even less forgiving, where a downgrade can move capital charges and a quiet deterioration you did not catch shows up later as a reserve problem. On the energy side it is the same pressure from a different angle, where a counterparty you have traded with for years can deteriorate quietly until the exposure is already on you. Nobody wants to be the one holding a First Brands when it turns, and depending on your seat the consequences run from capital and reserve adequacy to fiduciary duty to your investors. That is the problem we keep chipping away at, and June was a good month for it. We are building AIR to give you that answer in the moment you are asked, with work you can stand behind.

AIR Everywhere: ChatGPT, Copilot, and Your Own Stack

This one came straight from you. Enough customers asked us to get AIR inside the LLM environment they already run, Copilot, ChatGPT, Claude, that we built a way to meet you there rather than asking you to come to our UI. That request has now grown into something bigger. We are piping AIR out as a direct data feed, so the Risk Scores and signals land inside your own models, your own pipelines, your own stack, no interface required. One customer wired it in and surfaced deterioration on a name they would never have thought to pull up on their own, simply because the signal came to them inside the workflow they were already in. That is the whole point. The risk you miss is usually the one you did not know to look for. Here’s AIR directly in the OpenAI/ChatGPT and Co-pilot Agent Store:

AIR Credit Intelligence listing inside the ChatGPT Agent Store.
AIR Credit Intelligence in the ChatGPT Agent Store — elite credit intelligence inside the LLMs your team already uses.

Worth being clear about why that matters. A general model on its own is built to give you a confident, agreeable answer, and it will give you a slightly different one each time you ask. That is fine for drafting. It is not how you want to assess credit. The AIR feed puts a tested, reproducible risk engine underneath whatever you have built on top, so the number holds up the same way twice and you can show your work. The language model handles the interface. The model of record handles the decision. The teams already wiring AIR in this way are the ones moving fastest on names before they make the tape.

The Specialized Bank Model is Live

We expanded coverage to banking institutions that file US regulatory financials, including Call Reports and Y9s for bank holding companies. For names that already carried an AIR Risk Score, we refined the methodology to account for bank specific factors like loan portfolio composition. If you cover financials, this is the one to try first. Regional bank stress is exactly the kind of deterioration AIR is built to flag early, and the desks watching it through AIR right now have a head start the rest of the market does not.

AIR Risk Score time series and credit risk breakout for Bank of America Corp under the specialized bank model.
AIR Risk Score and key driver breakout for Bank of America Corp, generated under the new specialized bank model.

Iterative Scenarios in AIR Agent

You can now return to a scenario you’ve already run and ask follow-up questions against it, rather than starting over — helpful for digging into the drivers behind a projected Risk Score.


AIR Agent scenario chat showing a tariff scenario with a follow-up question about which companies will be most impacted.
Ask follow-ups against a completed scenario without restarting the run.

Quality-of-Life Upgrades

We shipped a set of smaller upgrades this month based on direct customer feedback:

  • Enhanced Portfolio View. Surface which names are moving, aggregate shifts in implied ratings, top movers (largest decliners and improvers), and a 1Y migration summary across the portfolio.
  • Data-First Search. Search now prioritizes companies with the most recent filings, so the names you need surface first.
  • Document Visibility. Track how the documents you've uploaded are being used across the platform.
  • Private Company Renaming. Rename private companies to match your own internal naming conventions.
Portfolio overview for a Renewable Energy book showing top movers, sector distribution, and AIR implied rating migration.
Portfolio view now surfaces top movers, aggregate rating shifts, and a 1Y migration summary.

We'd love to walk your team through these new capabilities. Reach out to schedule a session and see how to get more value out of AIR today.

Get started

Stay on par. Or stay ahead.

A continuously updated view of credit risk on every issuer that matters to your book.

  • 11mo median lead time
  • Private and Public issuers covered
  • SOC 2 Type II