← Back to guides

How Cifra's AI analyzes each sector

A generic artificial intelligence model applies the same questions to a soft drink company as to a cybersecurity firm. At Cifra, the AI is governed by a multi-agent system with specialized financial rules that tailor the inspection to each industry's business model.

1. Cifra's multi-agent architecture

When you request the analysis of a 10-Q or 10-K report filed with the SEC, three specialized agents intervene in sequence:

1. Sourcing Agent (SEC EDGAR): Download and validate the official documents filed with the SEC (10-Q, 10-K, and 8-K exhibits: press releases and earnings presentations with guidance). It ensures that the figures come from audited sources and not from third-party summaries.
2. Sector Agent: Identify the company's industry and subsector, loading the corresponding specialized knowledge framework (critical ratios, recurring costs and typical accounting traps for that sector).
3. Financial Analyst Agent: It processes the official XBRL tables and the report text, structuring the analysis into two time horizons and three mandatory quantitative blocks.

2. The two mandatory time horizons

A common mistake when reading quarterly results is judging a company solely by the year-to-date cumulative figures, ignoring whether the latest quarter has suffered a sharp deceleration. Cifra splits the analysis into two horizons:

• LAST 3 MONTHS: It shows exclusively the performance of the reported fiscal quarter. It allows immediate detection of margin losses, cost acceleration or slowdowns in sales volume.
• FOR THE FULL YEAR (YTD Cumulative): In quarters Q2, Q3 and the annual close Q4, analyze the cumulative period (6, 9 or 12 months). This neutralizes business seasonality and allows judging compliance with the annual guidance offered by management.

3. Adaptation of the analysis by sector

Each stock market sector presents particular financial dynamics. Below are the rules and focus areas that Cifra's AI applies across the main industries:

A. Consumer Staples sector (food, beverages, personal care)

Companies like Coca-Cola, Procter & Gamble, Kraft Heinz or Molson Coors sell everyday necessity products with inelastic demand. The AI focuses its analysis on:

• Volume vs Price/Mix Breakdown: Are sales growing because the company is selling more units (healthy growth) or only because it is passing on inflation by raising prices? If volume persistently falls while price rises, the AI warns about the exhaustion of pricing power against the distributor's private labels.
• Gross Margins and Input Costs: Sensitivity to agricultural commodities (coffee, sugar, cocoa, wheat) and packaging materials (aluminum, cardboard, glass).
• Recurring Free Cash Flow and Dividends: Defensive companies tend to pay a generous dividend. The AI calculates the cash conversion rate (*FCF Conversion*) and the payout on free cash flow to verify whether the dividend is comfortable or whether it compromises balance sheet strength.
• Deleveraging after Acquisitions: History of net debt / EBITDA after past brand purchases or mergers.

B. Technology and Software Sector (SaaS / Cloud / Big Tech)

In the technology sector, recurring revenue models and executive compensation policies distort traditional financial statements:

• The critical Stock-Based Compensation (SBC) filter: Many tech companies boast a high adjusted profit because they exclude the payment of stock-based salaries to their employees. Cifra's AI considers share dilution as a real operating cost that must be deducted from true shareholder performance.
• Annual Recurring Revenue (ARR) and Net Dollar Retention (NDR): Measurement of cross-selling capability (*upselling*) among existing customers and the churn rate (*churn*).
• Cash Conversion vs. Accounting Margins: Examination of operating cash flow deducting deferred revenue from multi-year contracts to avoid temporary cash mirages.
• R&D intensity (R&D): Percentage of sales reinvested in research and development to maintain competitive advantages.

C. Industrials and Capital Goods Sector (Industrials)

Machinery, components, logistics and heavy equipment companies operate with long cycles and high fixed costs:

• Backlog and Book-to-Bill: The relationship between orders received and orders delivered. A ratio above 1.0 signals demand expansion for the coming quarters.
• Operating Leverage: Small percentage variations in sales produce pronounced fluctuations in operating income due to the need to amortize large-scale factories and infrastructure.
• Maintenance vs Growth CAPEX: Differentiation between the investments necessary so that the machinery does not become obsolete and the capital allocated to opening new plants.

D. Consumer Discretionary and Retail Sector

Companies that depend directly on consumer disposable income (fashion, electronics, restaurants, leisure):

• Comparable Sales (Same-Store Sales): Isolation of organic growth from mature stores relative to the net opening of new establishments.
• Management and Days Inventory Outstanding (DIO): The anomalous increase in inventory on the balance sheet often anticipates forced discount markdowns (*promotions*) and inventory impairments that erode gross margin.
• Customer Traffic and Average Ticket: Breakdown between the number of physical store / online visits and the average amount spent per purchase.

E. Energy Sector (Energy: oil, gas, and midstream)

Compañías de exploración y producción (E&P), refino, servicios petroleros y transporte de hidrocarburos cuyo resultado depende del precio del crudo y del gas:

• Production and Realized Price: Production in boe/d and realized price against WTI and Henry Hub benchmarks. The AI distinguishes whether profit grows from more volume, better prices, or lower costs.
• DD&A y Coste de Extracción: La amortización de activos petroleros (depletion) es muy elevada, por lo que el EBITDA cobra protagonismo; se vigila el lifting cost y el coste de reposición de reservas (F&D).
• Fee-Based Midstream: In transportation and storage with take-or-pay contracts, high debt and stable cash flow are structural; interest coverage is reviewed, not commodity volatility.
• Reserves and Capex: Evolution of proved/probable reserves, maintenance vs growth capex, and dividend sustainability (often variable) through the cycle.

F. Materials Sector (Materials: chemicals, mining, packaging)

Producers of chemicals, metals, packaging, paper, and construction materials, with margins tied to raw material spreads:

• Raw Material vs Product Price Spreads: Margin depends on the spread between inputs (naphtha, gas, iron ore) and outputs; the AI explains the spread, not just the price.
• Price vs Volume and Utilization: Breakdown of growth between price increases and real volume, with capacity utilization as a cycle signal.
• LIFO Inventories and Environmental Provisions: Effect of LIFO liquidations in chemicals and steel, and environmental remediation and closure liabilities (not financial debt).
• Mining: Proved reserves, capitalized stripping, royalties and streams (sales of future production), and TC/RC costs in smelters.

G. Health Care Sector (pharma, biotechnology, medtech, services, distribution)

Pharmaceuticals, biotechnology (including pre-commercial), medical devices, hospitals and laboratories, wholesalers and pharmacies, life sciences, and animal health. The AI adapts the inspection to each business:

• R&D as maintenance investment: In pharma, biotech and medtech, R&D remains a cost (it is not excluded from profit) and is explained in its own note: amount, percentage of sales and evolution. It is the real investment that sustains the business and does not appear in CAPEX.
• Intangibles de producto e IPR&D: La amortización de productos comercializados y patentes y el deterioro de IPR&D permanecen como coste en la columna Ajustado; solo el deterioro de fondo de comercio se revierte con nota. El desgaste del producto protegido es real.
• Companies without earnings (pre-commercial biotech): Visible warning and focus on liquidity: burn rate, cash/runway, convertibles and dilution (ATM, capital raises). The pipeline is neither evaluated nor valued, and expected losses are not read as business deterioration.
• Subsector peculiarities: in pharma, gross vs net revenue (rebates, 340B, chargebacks), channel and distributor inventories, loss of exclusivity (LOE) and litigation; in medtech, razor/blade model, elective procedures and recalls; in services, payer mix and volumes (admissions, tests); in distribution and pharmacies, thin margins, working capital and reimbursements; in life sciences, recurring consumables and orders; in animal health, pets vs production.

H. Utilities Sector (regulated power, gas, water, and IPP/merchant)

Regulated power and multi-utilities, regulated natural gas distribution, water supply, and independent power producers and merchants (IPP). Their results depend on the rate base, approved rates, and, in the unregulated portion, on power prices:

• Rate base, authorized ROE, and rate cases: Regulated profit is the authorized ROE applied to the rate base; the AI explains the capex plan, rate cases, and the approved rate of return. All of a utility's capital revolves around the rate base and the dividend.
• Structurally high debt and negative FCF: Rate base capex is financed with debt and equity issuances: high debt and negative free cash flow are the business model, not a warning sign on their own. Interest coverage and maturities are monitored.
• Dilution (ATM/DRIP) and shareholders' equity: In the 10-K, Cifra adds a deterministic "Equity" section with the evolution of shareholders' equity, book value per share, total assets, and share count change (ATM/DRIP dilution); the AI interprets it without recalculating it.
• Regulated vs merchant mix: In IPPs/merchants (VST, CEG, NRG), the unregulated portion depends on power prices, capacity markets, and contracted hedges; it is separated from regulated results when the filing breaks it out.

I. Communication Services sector (telecom, cable, media, publishing, and streaming)

Telecommunications, cable and pay satellite, broadcasting, publishing, and film/streaming/entertainment (studios, cinemas, live entertainment, sports, music, and satellite radio). The AI adapts the inspection to each business:

• Cash vs accounting and the cost of content: the core of the business is recurring subscriber revenue (telecom, cable, streaming, satellite radio) against the real costs of content and network. The amortization of produced or licensed content is a real operating cost and is not added back; content spend versus its amortization is the key streaming metric.
• Spectrum, licenses, and brand intangibles: the amortization and impairment of spectrum, licenses, customer relationships, and brands remain as a cost in the Adjusted column (as in technology and health care); only goodwill impairment is reversed with a note.
• Structural telecom debt and network capex: high debt and negative free cash flow after network capex are not warning signs on their own (as in utilities): capex is financed with debt and the subscriber service pays it back. Spectrum auctions are non-recurring capex and tower leases are not financial debt.
• Advertising and the biennial election effect: in broadcasting and publishing, advertising revenue is cyclical and distorted in election years (even/odd effect); advertising is separated from recurring retransmission, subscription, or affiliation revenue, and digital subscribers, ARPU, churn, and audience are read.

J. Real Estate / REITs Sector (Real Estate: listed equity REITs)

U.S. listed equity REITs (industrial and logistics, retail and net lease, residential, offices, data centers, self-storage, health care, hotels, towers, timber/farmland and specialty). The challenge is not classification: a REIT's GAAP profit is not useful because property depreciation drags it down. The AI analyzes with the sector lens:

• The FFO/AFFO lens: el FFO (Funds From Operations, estándar NAREIT) suma de vuelta la depreciación de los inmuebles y los deterioros de inmuebles depreciables y resta las ganancias por ventas. Cifra calcula una aproximación determinista (Beneficio Neto + D&A total + deterioros − ganancias) y, si el filing publica un FFO con reconciliación, la fila y la nota prefieren la cifra reportada. El AFFO/CAD no es estándar (cada empresa lo define) y solo se muestra si el filing lo publica: el sistema no lo recalcula.
• The dividend is judged against FFO: the obligation to distribute at least 90% of taxable income makes a payout on EPS above 100% normal; sustainability is measured with the payout on FFO and, if available, on AFFO.
• Occupancy, leasing spreads and same-store NOI: operating portfolio metrics (physical vs economic, total vs same-store, RevPAR/ADR for hotels) explained as published by the company. In the 10-K, Cifra adds a deterministic "Portfolio Occupancy" section (10 years) when the XBRL series is reliable; if not, it is omitted and not invented.
• Book value and price: the 10-K includes book value per share versus the closing price with the P/B; book value is historical cost less accumulated depreciation and is not NAV (which is not calculated).
• Equity raises and financing: ATM/DRIP programs are the sector norm; the deterministic section quantifies the shares issued and their implied average price versus book value (issuing above creates book value per share and below dilutes it), new debt with its reported rate, and property purchases. Structurally high debt (4-6x Net Debt/EBITDA) is the business design, not an alarm by itself.

K. Financials Sector (banking, consumer finance, insurance, capital markets, and payments)

Commercial and investment banks, consumer finance (cards, consumer credit, non-REIT mortgage), insurance (including managed care or health plans), capital markets (brokers, asset managers, exchanges, data/ratings) and payments. Here the business es the balance sheet and the standard engine (sales with gross profit, working capital, FCF) does not apply as-is: the AI uses its own lens, the balance sheet lens, and reserve the standard format for fee-based businesses (asset managers, exchanges, data/ratings, brokerages, and payments).

• Block 1 — MARGIN: net interest income, fees and other income, total revenue, provision for credit losses (always as a cost and never adjusted), operating expenses, EBT and net income; for insurers, earned premiums, investment income, loss ratio and DAC. The provision is analyzed against net charge-offs (NCO) and nonaccrual coverage, using the definition published by the bank.
• Block 2 — BALANCE SHEET AND QUALITY (replaces classic Cash Flow; RESERVES AND FLOAT in insurance): loans and deposits with their prior close, LDR, 90+/nonaccrual delinquency, ACL/NPL coverage, net charge-offs, tangible book value per share and AOCI. The CET1 and regulatory ratios are only shown if the filing publishes them: the system does not calculate them. Deposits and funding are not debt-as-risk: they are the balance sheet model.
• Block 3 — CAPITAL AND RETURN: capital reconciliation (Net Income ± change in AOCI + issuances − dividends − buybacks − acquisitions = retention) cross-checked against the change in balance sheet equity; dividends and buybacks are read as constrained by regulatory capital.
• Deterministic 10-K series: "Capital and profitability" sections (ROE on average equity, NIM on average assets, efficiency ratio, and tangible book value per share) and "Credit quality" (nonaccrual loans and coverage) with 10-year charts; for insurers, "Book value and ROE" with the reported combined ratio or MLR.
• Ratio honesty: the definitions of delinquency, coverage, and NCO vary by bank and are cited as-is, without normalization; the company's "operating EPS" figures are cited separately and never confused with the Adjusted column.

4. The Three Immutable Pillars of Analysis

Regardless of the sector, every Cifra analysis verifies three quantitative blocks to ensure that no figure is left out of context:

Block	What it analyzes	Why it is decisive
1. Sales and Margins	Revenue, Gross Profit, Operating Income, EBT and EPS (Normal vs Adjusted).	Evaluates the quality of growth and exposes the cosmetic adjustments management uses to inflate earnings.
2. Free Cash Flow	Operating cash flow less CAPEX (capital expenditure).	Accounting profit can be manipulated with depreciation and provisions; cash generated in the bank doesn't lie.
3. Capital Allocation	Dividends paid, net share buybacks, change in net debt and acquisitions.	Determines whether management creates long-term value or wastes resources on expensive buybacks or destructive acquisitions.

Investor summary

Artificial intelligence should not be limited to summarizing text. It must apply the rigor of a senior financial analyst: contextualize the figures across two time horizons, verify consistency between profit and free cash flow, and apply the right questions to the specific sector in which the company competes.

Frequently asked questions about Cifra AI

Why does Cifra AI analyze companies differently depending on the sector?

Because every industry has completely different profitability drivers, cost structures, and risks. While pricing power and volume elasticity are key in consumer staples, revenue recurrence (ARR) and share dilution cost (SBC) come first in technology.

What are the two mandatory time horizons in Cifra analyses?

Quarterly Q2 and Q3 analyses and annual Q4 analyses are strictly divided into two independent horizons: 'LAST 3 MONTHS' (the strict quarter to see the recent trend) and 'FULL YEAR' (the year-to-date cumulative to check the annual trajectory and smooth seasonality).

Which mandatory pillars does the AI evaluate for each company?

Three immutable pillars: 1) Income statement (sales, margins, and adjusted vs GAAP EPS), 2) Free Cash Flow (real cash generation after capital investments), and 3) Capital Allocation (dividends, buybacks, debt, and acquisitions).

Educational content from Cifra. It does not constitute a buy recommendation or financial advice.