The analyst that reads the filing and walks the floor.
cubby ai sizes markets, decomposes supply chains, and reads disclosure the way an ops lead reads a P&L. Sourced output, English and Japanese, every figure traceable.
§ 01 · Capabilities
Built for analyst-grade research with an ops bias.
Reads, computes, cites. Every number ties back to the primary source it came from.
Reads the long documents
Thousands of pages of 10-Ks, S-1s, annual reports, and earnings transcripts — extracted into the rows and columns you actually model on.
Sized, not consensus-fed
Bottom-up TAM/SAM/SOM with assumptions you can pressure-test. No magic numbers from secondary aggregators.
Supply chain, decomposed
Inventory turns, supplier concentration, lead-time drift — modeled from disclosure, not from the deck the company wants you to read.
Two markets, one workflow
US SEC filings and Japanese yuho disclosures — same agent, one output, audit trail attached.
§ 01 · MARKET SIZING
Industry & market sizing
TAM / SAM / SOM, segment momentum, and the primary sources behind every figure.
Estimate TAM/SAM/SOM for warehouse management software targeting 3PLs in Southeast Asia, bottom-up: warehouse count by country, average WMS spend per facility, addressable share for cloud-native solutions. Show assumptions explicitly.
A 19% CAGR through 2029 is forecast for last-mile delivery software in Japan. Pressure-test with 3 supply-side and 3 demand-side inputs. What would have to be true on labor cost and e-commerce penetration for the number to hold?
Within industrial automation, name 3 sub-segments with the strongest 2024 to 2026 momentum across funding flow, hiring velocity, and regulatory tailwind. One paragraph thesis each.
Our company sells inventory software to mid-market retailers. List 5 adjacent verticals with high overlap in the buyer's job-to-be-done, and the criteria that would justify expansion into each.
I am sizing the global cold-chain logistics market. Rank the 5 most credible primary sources, what each is good for, and where they disagree on base year and definition.
Map 8 generative-AI sub-categories on a 2x2 of hype level by defensible moat. Justify each placement with one named example and the moat type it relies on.
§ 02 · COMPANY DEEP DIVE
Company deep dive
Financial models, business-model decomposition, KPI walks tied to disclosure.
Draft a conservative 5-year DCF for Toyota Motor Corp at current consensus. State WACC, terminal growth, and the 2 sensitivities that matter most. Flag any input treated as assumption versus disclosed.
Sony's operating margin moved 280 bps over the last 3 fiscal years. Decompose the walk by segment, price/mix, and one-time items based on their MD&A and segment footnotes.
Build a quality screen for global large caps: ROE > 15%, FCF margin > 12%, Net debt / EBITDA < 1.5x, 5-yr revenue CAGR > 7%. Output the criteria and 3 names this captures across US and Japan.
Reconstruct SaaS unit economics from public ARR + customer count + S&M spend: 200m ARR, 10,000 customers, 80m S&M. Compute CAC, payback period, LTV/CAC, and stress-test with 20% churn.
Among Nikkei 225 companies that announced buybacks in the past 12 months, rank top 5 by buyback yield, completion rate, and EPS accretion. Note whether any was funded by new debt.
Argue the strongest 3-point bear case for Nintendo at current valuation. Use only points a skeptical long/short analyst would lead with, sourced from the most recent disclosure.
§ 03 · SUPPLY CHAIN & OPS
Supply chain & operations
Inventory, suppliers, lead times — modeled from filings and operations footnotes.
Decompose Shopify's effective inventory exposure across its merchant base from disclosed merchant GMV, take rate, and Shopify Fulfillment Network footprint. Highlight the 2 segments most exposed to a 20-day lead-time spike.
From Apple's most recent 10-K and supplier list, build a concentration view: top 10 suppliers by named-component category, geographic concentration, and the 3 single-source nodes most at risk.
Track lead-time language across the last 4 quarters of Datadog's and Snowflake's earnings calls. Where are they signaling drift on infra capacity, talent, or international rollout?
Walk Toyota's COGS over the last 3 fiscal years, decomposing material, labor, energy, and one-time items. Tie each line to the segment footnote that disclosed it.
Across the global top 10 contract manufacturers, rank working-capital efficiency by cash conversion cycle, payable days, and inventory days. Flag any name where receivables grew faster than revenue 2 quarters running.
TSMC announced a new Arizona fab phase. From the disclosed capex, equipment lead times, and historical ramp curves, estimate when meaningful wafer output hits — month and quarter — with the sensitivities.
§ 04 · CAPITAL MARKETS & MACRO
Capital markets & macro
Rates, FX, policy transmission, sector rotation — read through an operator's lens.
Compare the latest FOMC statement to the prior one word-for-word. Identify hawkish/dovish shifts and map each to a likely market reaction in 2-yr rates, USD, and the 3 most exposed S&P sectors.
Decompose the latest US CPI print into goods, services ex-shelter, shelter, and energy. What does each say about the Fed's reaction function over the next 2 meetings?
Build 3 BoJ policy paths over the next 12 months (status quo, gradual normalization, sudden hike). For each, project USD/JPY range and the 3 most exposed Japanese export sectors.
Explain the current shape of the US Treasury curve, what is implied for 12-month forward growth and inflation, and which historical analog it most resembles. Note where the analog breaks down.
Across the past 4 US recessions, which equity sectors led the recovery in the first 6 months after the trough? Show base rates and one structural reason per sector.
Pull the latest Baltic Dry, container-rate, and US trucking-tonnage indices. What signal do they collectively give for Q4 GDP nowcasts and for retailer inventory cycles?
§ 05 · FILINGS & DISCLOSURE
Filings & disclosure
Structured extraction from primary documents — 10-K, S-1, yuho, transcripts.
From the attached 10-K, extract the risk-factors section and rank by materiality given the company's current business mix. Quote the sentences supporting each rank.
Summarize the MD&A of the attached 10-K in 200 words, capturing revenue drivers, margin drivers, and any forward-looking guidance, with paragraph references.
Identify any accounting-policy changes, restatements, or critical estimate revisions in the attached annual report. Explain what each implies for earnings quality and comparability.
From the financial-statement footnotes, surface 5 items most investors miss: contingent liabilities, off-balance-sheet exposures, related-party deals, segment reclassifications, and one of your choice.
From the attached S-1, extract growth metrics, cohort behavior, customer concentration, and any disclosure that contradicts the headline narrative on the road show.
From the attached earnings-call transcript, identify the 3 questions analysts pressed hardest, summarize management's answers, and rate answer quality 1 to 5 with one-line justifications.
§ 99 · How it runs
How it runs
Frame the question. Get an audit trail.
You frame the question
Market size, model, filing digest, supply-chain walk — any analyst-shaped ask.
It reads and computes
Pulls primary sources, parses disclosures, runs the spreadsheet logic, cites as it goes.
You get sourced output
Tables, walks, citations. Auditable. Ready for the memo or the deck.
§ END · Open a session
Hand the agent the research question.
Free to try. In English and Japanese.