Hire an AI company earnings adjustments and reliability reviewer

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• A table of one-off items and each year's ordinary earnings
• Accounting choices that lift profit, sized, with a clean, adjusted or avoid call
• A rating of how far the record can guide the future
• Low, normal and high earnings per share, each with a reason

How it does the job well

• Reconcile to source documents, such as bank statements, ledgers and invoices, before reporting.
• Show the calculation, inputs and period for every figure.
• Label actuals, estimates and forecasts clearly.
• Send tax, legal and regulatory questions to a qualified professional.

Mistakes it avoids

• Mixing cash and accrual figures, or different periods, in one comparison.
• Rounding or formula errors that do not tie back to totals.
• Giving personal investment advice.

What a good result looks like

An example, not your task: A company's latest profit looks too good, or too poor, to be typical.

• Step 1 one-offs removed. One-off items by year and ordinary earnings by year, with the ordinary and all-in averages.
• Step 2 padding. Accounting choices that lift profit, sized, restated earnings and a clean, adjusted or avoid call.
• Step 3 how far the record can be trusted. Six tests of the record and a guide, partial guide or not a guide rating.
• Step 4 normal earnings. Low, normal and high earnings per share, each with a reason, and the limits.
• Hand-back summary. Each step's file and every MISSING line in one list.

How the work is checked

• Each step passed its own checks, or the run stopped at the step that failed and says why.
• The steps ran in order, and each one used the result of the one before.
• Nothing was invented to fill a gap. Every gap is a MISSING: line.
• Gains and losses of the same kind were paired, and an avoid call puts no number on dishonesty.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

A second opinion

Before it hands back, your AI reviews the draft the way Benjamin Graham would. This is a way of reviewing the work. It does not mean they wrote or endorse it.

Capabilities

• Removes the one-offs. Lists gains and losses that will not repeat, pairs the ones booked in different places and rebuilds each year.
• Looks for padding. Checks write-downs, parked costs, depreciation, subsidiary income and per-share effects, and sizes each.
• Rates the record. Tests the scatter, the nature of the business, temporary profits and what the loss years say about survival.
• Works out normal earnings. Averages five to ten years, caps the figure at what a normal year has shown and gives a range.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id real-earning-power only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/real-earning-power/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• Five to ten years of income statements, with the notes
• Balance sheets and cash flow statements for the same years (optional)
• A few lines on the business and its industry (optional)
• Share counts by year (optional)
• Years you think were unusually good for the business (optional)

Helpful context (optional)

• The company's annual reports, with the notes
• Cash flow statements for the same years
• Share counts, splits and buybacks by year

Tools and access

• Claude: Reads many lines across several years and keeps each step's table for the next.
• A spreadsheet: Lets you check the year-by-year table and the averages.
• Any good chat AI: Reading statements and building a table is text work. It cannot fetch the filings for you.
• A spreadsheet: Handy for checking the adjustments by year.
• A licensed financial adviser: Needed for a large or personal decision. This skill is a method, not advice.
• Any good chat AI: Reading statements and linking years is text work. It cannot fetch filings.
• A spreadsheet: Handy for the restated earnings and share counts.
• Any good chat AI: Judging a record against a business description is text work.
• A spreadsheet: Handy for the average, the spread and how profit moves with volume.
• Any good chat AI: Choosing a window and building a range is text work.
• A spreadsheet: Handy for averages and the worst and best three-year runs.

Functions and deliverables

• Company earnings adjustments and reliability reviewer: Share several years of statements and notes. Get one-off items, accounting questions, history checks and an estimated range for normal earnings.
  • A table of one-off items and each year's ordinary earnings
  • Accounting choices that lift profit, sized, with a clean, adjusted or avoid call
  • A rating of how far the record can guide the future
  • Low, normal and high earnings per share, each with a reason
• One-off earnings adjustment reviewer: Find one-off gains and losses in several years of income statements, then see each year's ordinary earnings without them. Gains and losses are paired.
  • A table of one-off items by year, with where each one sat
  • Gains and losses of the same kind paired, with their net
  • Ordinary earnings for each year with every adjustment shown
  • An ordinary average and an all-in average
  • A list of items to question
• Reported earnings adjustment reviewer: Review accounting choices that may change reported profit. Get sized findings, an adjusted earnings view and questions to resolve before relying on it.
  • Each finding with year, amount, direction and share of reported profit
  • Reported and restated earnings by year
  • A read of the pattern
  • A clean, adjusted or avoid call
  • Questions to put to the company
• Earnings history reliability reviewer: Review how much weight to give past earnings. Get a record check, business risks and a guide, partial-guide or not-a-guide rating.
  • The record with its scatter
  • A finding for each of the six tests
  • A rating of guide, partial guide or not a guide
  • How wide the next range should be
  • What facts would raise the rating
• Normal earnings estimator: Share cleaned earnings and a reliability rating. Get candidate figures and a range only when source and order checks pass.
  • The window and share count used
  • Averages with loss years in and at zero
  • A statement of whether the latest year was used alone, and why
  • Candidate low, normal and high earnings per share and in total, each with a reason
  • A range-order result and a validated range only after all source and order checks pass
• me.md builder: Answer a few questions, or let your AI use what it remembers. Get one private file that tells any AI who you are.
  • One me.md file with up to ten dated sections
  • A list of facts taken from memory, for you to check
  • A short note of what it removed for privacy
  • The two next steps, to save it and set it up
• Benjamin Graham: Learn to separate price from value using Benjamin Graham’s recorded frameworks. Check the evidence, the downside and the room for error.
  • Your problem worked on the way Benjamin Graham would: their questions, their method, their push-back
  • The answer or the work, in Benjamin Graham's voice, using their ideas by name
  • Anything that goes beyond their public record, marked
  • One suggested next step

Tasks you can use it for

• A company's latest profit looks too good, or too poor, to be typical.
• A big write-down was followed by a very good year.
• You are valuing a business and need one earnings figure you can defend.
• You are reading an annual report and want to know which profit is for show.

Limits to check before starting

• You want a price or value for the company: run this first, then Stock valuation margin-of-safety reviewer.
• You are lending to the company, not owning it: use Bond and loan repayment risk review.
• You want to know if the business itself is any good: use Long-term business quality reviewer.
• You have fewer than five years of statements, or the company is new. The record cannot guide, and the run will say so.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Download the handoff pack. It's a zip with HANDOFF.md and every skill file this workflow uses.
• Give the folder to Claude (in Claude Code, or as files in a Claude Project) and say "Follow HANDOFF.md".
• Paste the income statements and the notes. Add cash flow statements and a note on the business if you have them.
• It runs four steps and saves each one. A missing figure is marked MISSING, never guessed.
• Read the call on padding first. If it says avoid, stop there. If not, use the normal earnings range.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/real-earning-power for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.