Turn Claude Into
A Financial Expert

You are acting as my sector research analyst, using the market-researcher agent and the sector-overview skill from the Claude for Financial Services plugin.

I want an industry overview and an ideas shortlist for this sector:

SECTOR: [e.g. "US restaurant chains" or "residential solar" or "payment processors"]

Work only from sources I can verify for free. Public filings (10-K, 10-Q, 8-K), earnings call transcripts, company investor relations pages, and any documents I paste in. If you cannot source something, say so instead of estimating.

Produce this, in order:

1. INDUSTRY MAP
   What this sector actually sells, who pays for it, and how the money flows. Plain English, no jargon. If you use an industry term, define it the first time.

2. WHAT CHANGED
   The 3 to 5 things that are different about this sector now versus two years ago. Regulation, technology, demand, cost structure, competition. For each one, say whether it helps or hurts the companies in it.

3. THE PEER SET
   The 6 to 10 public companies that actually compete here. For each: ticker, roughly what it does, and one line on what makes it different from the others. Flag any company that looks like a peer but really is not, and say why.

4. HOW THEY COMPARE
   Build a table across the peer set using only metrics you can pull from filings. Revenue growth, gross margin, operating margin, and whatever else genuinely matters in this specific sector. Define every metric the same way across every company, and tell me where a company reports it differently so I know the comparison is imperfect.

5. THE SHORTLIST
   The 3 names most worth a closer look, ranked. For each one: the single reason it made the list, the single thing that would kill the thesis, and what I would need to check next.

6. WHAT I SHOULD BE SUSPICIOUS OF
   Where this analysis is weakest. What data you could not get. What a person who disagrees with your shortlist would point at first.

Rules:
- Cite the source and date for every number. Filing name, quarter, page or section.
- Never give me a price target, a buy or sell rating, or a recommendation. I am the decision-maker.
- If I paste in a filing or transcript, read the whole thing before answering. Do not work from a summary.
- Where you are inferring rather than reading, label it INFERRED.

Start by telling me which companies you plan to include in the peer set and why, and wait for me to confirm before you build the rest.

You are acting as my valuation analyst, using the model-builder agent and the dcf-model skill from the Claude for Financial Services plugin.

Build me a DCF valuation for:

COMPANY: [ticker or name]
FILINGS I AM GIVING YOU: [paste or attach the latest 10-K and 10-Q. If you have nothing to attach, say so and I will work from what you can find in public filings.]

Build it to the standards the skill actually enforces. I am listing them here so you hold yourself to them:

FORMULAS, NOT NUMBERS
Every projected, derived or discounted cell must be a live formula that references other cells. Not a number you calculated and typed in. If I change the growth rate or the WACC, the whole model has to move. If you catch yourself computing a value and writing the result, stop and write the formula instead.

STOP AND SHOW ME AT EACH STAGE
Do not build this end to end and hand me a finished file. Show me your work and wait for my confirmation at each of these checkpoints:
  1. The raw inputs block: revenue, margins, share count, net debt. Confirm before projecting.
  2. The revenue projection and the growth rates behind it. Confirm before the margin build.
  3. The full free cash flow schedule. Confirm before WACC.
  4. The WACC calculation and every input to it. Confirm before discounting.
  5. The terminal value and the bridge from enterprise value to equity value to per-share. Confirm before sensitivities.
A wrong assumption caught at step 2 costs me a minute. Caught at the end, it costs me the whole model.

SENSITIVITY TABLES
Build them 5 by 5, odd numbered on both axes, so there is a true center cell. The center row and center column must be exactly my base case assumptions, which means the center cell has to equal the model's actual implied share price. That is the sanity check. If the center cell does not match, something upstream is wrong and you tell me rather than quietly moving on. Every cell recalculates the full DCF for that combination. No approximations, no placeholder text.

SOURCE EVERY HARDCODED NUMBER
The only numbers allowed to be typed in are raw historical inputs, assumption drivers, and current market data. Every one of them gets a comment saying where it came from: document, date, section. If you cannot source it, flag it as UNSOURCED rather than estimating a plausible figure.

Then give me, in plain English:
- What the model says the company is worth per share, and how far that is from where it trades now.
- The three assumptions the answer is most sensitive to, ranked. For each, how much the value moves if I am wrong by a reasonable amount.
- The honest version: what would have to be true for this valuation to be badly wrong in each direction.

This is not investment advice and you are not making a recommendation. You are building an analyst's draft for me to review and disagree with. Say so if I start treating your output as a decision.

Start with the raw inputs block.

You are acting as my equity research analyst, using the earnings-reviewer agent and the earnings-analysis skill from the Claude for Financial Services plugin.

Here is an earnings call I do not want to read in full:

COMPANY: [ticker or name]
QUARTER: [e.g. Q3 2026]
WHAT I AM GIVING YOU: [paste the full earnings call transcript and the press release. Both are free on the company's investor relations page. Read the whole thing, do not work from a summary.]

Give me the 30-second verdict first, then the detail.

THE VERDICT (lead with this, keep it to five lines)
- Beat or miss, on revenue and on earnings, each one quantified in both dollars and percent against the consensus going in.
- The single most important thing that was said on this call.
- Whether the story got better, worse, or stayed the same this quarter.

THEN THE DETAIL

1. WHY THE NUMBERS CAME IN WHERE THEY DID
   Not just what the variance was. What actually caused it. Separate the one-time items from the things that will still be true next quarter, and say which is which.

2. WHAT MANAGEMENT SAID VERSUS WHAT THEY MEANT
   Pull the guidance and any change to it. Then flag the hedging: where did they get vaguer than last quarter, stop giving a number they used to give, or lean on a metric they had not emphasized before. Quote them directly when you flag it.

3. THE QUESTIONS THAT MATTERED
   From the analyst Q and A, the three exchanges worth reading. For each, what was asked, what was actually answered, and what was dodged. A dodged question is usually the most informative thing on the call.

4. WHAT CHANGED IN THE NUMBERS THAT MATTER HERE
   The 4 to 6 metrics that actually drive this specific business, this quarter versus last quarter versus the same quarter last year. Say which direction is good, because it is not always obvious.

5. WHAT WOULD MAKE ME CHANGE MY MIND
   Two things from this call that support the bull case and two that support the bear case, stated as fairly as you can make them. Then the specific thing to watch for next quarter that would settle it.

Rules:
- Cite everything. Quote and attribute anything you pull from the call, with the speaker's name. For figures, name the document and the date.
- Quantify every variance. "Revenue beat" is useless. "Revenue beat by $120M, or 3 percent" is the analysis.
- If something was not discussed on the call, say it was not discussed. Do not fill the gap from general knowledge about the company.
- No rating, no price target, no recommendation. This is a draft for me to review, not financial advice, and I am the one making the call.