Triente publishes three free, interactive tools. Two of them read the U.S. banking industry. The third reads the national mood. From the outside they look like three different projects on three different subjects, and it would be fair to file them as unrelated demonstrations.
They are the same method applied to different public datasets. The subject changes; the method does not. This note describes that method, and uses the one tool that leaves banking entirely, the U.S. Sentiment Signal, to show it working outside the firm's home field.
The method, in four steps
The method is four steps, and the order matters.
1. Start from authoritative public data. We don't generate the underlying facts. We start from data a public body already collects and stands behind. For the banking tools that is the FDIC's quarterly Call Reports, filed by every federally insured bank. For the Sentiment Signal it is federal economic series from the Federal Reserve's FRED API and the better-known national surveys. The first decision in every build is the same: find the most authoritative public source for the question and build on it, not on anything proprietary of ours.
2. Wire it through modern APIs into a clean, queryable layer. Public data is usually published for the record, not for use. The FDIC's own interface is a dated web form; FRED is an API, but each survey publisher formats its release differently. The work is plumbing. Pull each source through its API or a reliable scrape, normalize it onto a common shape, and cache it so a reader gets an answer in seconds instead of an afternoon of cleaning. Modern APIs make this cheap enough that a small team can do in days what used to need a data vendor.
3. Add a small amount of AI, for synthesis rather than facts. AI comes in last, after the data is already assembled and checked. It writes the short plain-language narrative on the banking dashboard and helps summarize where a series stands. It does not produce the numbers. Every figure traces to a public source; the model's job is to read the assembled data back in sentences, under the same provenance rules as everything else. A tool that lets a model invent the facts is worse than no tool.
4. Keep provenance attached to every number. Each value carries a tag for where it came from and how solid it is: measured from an official source, scraped from a publisher, entered by hand, or modeled as a baseline where no public reading exists yet. The Sentiment Signal applies this most fully, marking every point on every chart this way. This is the step most dashboards skip, and it is the one that decides whether a reader can trust what they are looking at. Naming the modeled points is also how we stay honest about what the tool does and does not know.
The Sentiment Signal is the test
The U.S. Sentiment Signal is the clearest evidence the method travels, because it has nothing to do with banking.
The two banking tools sit on the firm's home field, where the data and the questions are familiar. The Sentiment Signal applies the same four steps to a different question entirely: how the country is feeling about the economy, the government, and its own direction. The starting point is still public data — federal series from FRED plus the major national surveys. The plumbing is the same — ten series normalized onto a single 0-to-100 scale, updated monthly. The use of AI is the same and just as restrained, a summary laid over numbers it does not invent. The provenance discipline is the same, every point marked measured or modeled. The subject moved from bank balance sheets to national mood, and the method carried over without modification.
That is the useful test. A method that only works on the data it was designed for is a one-off. A method that produces a banking explorer and a national-sentiment index from the same four steps is something a firm can re-point at a new question.
What this means, and what it doesn't
The method generalizes across industries. The firm's practice stays specialized.
These tools are demonstrations built on public data, not the work itself. They exist to show, in public, how Triente turns a messy external dataset into something a decision-maker can read in a few seconds. What we sell is the same method applied to a client's own data, at a depth a public tool cannot reach, anchored in banking and the financial sectors closest to it where the firm's experience runs deepest.
The two claims sit together without contradiction. The method is general; we have pointed it at bank Call Reports and at national survey data with the same four steps. The offer is specific; an engagement goes deep on one organization's question, on its own numbers, with senior people doing the work. The public tools are the evidence that the method holds. They are not an invitation to treat us as a generic data shop.
Who it is for
You do not need to be hiring us to use the tools. They are free, they cite their sources, and they are built for anyone who has to read an industry or the country without a data team behind them — journalists, researchers, students, and operators. If you are weighing whether to bring us in, the tools are also the most honest brief we can give you. This is how we work, on data you can check yourself.
The three tools are free and need no account: The U.S. Banking Industry Data Explorer, The U.S. Banking Industry Simulator, and The U.S. Sentiment Signal. If you want this method pointed at a question of your own, get in touch.