Doubao Work Financial AI Agent: Automating A-Share Equity Research from Data to Contrarian Analysis
The author walks through a full equity research workflow on A-share optical communication stocks using Doubao Work's financial data plugin and built-in skills, demonstrating automated hotspot analysis, conditional stock screening, research report digestion, multi-stock financial comparison with formula-driven Excel output, daily monitoring, and a contrarian risk review via the Stock Analyst work partner, highlighting cross-device collaboration and source-cited rigor.
Overview
The author, a practitioner of AI-assisted investing, details a complete equity research workflow using the desktop version of Doubao Work. The platform's financial data retrieval plugin connects directly to authoritative Chinese data sources — Wind, Tonghuashun, and CNRDS — providing real-time quotes, financial statements, and macro indicators. Combined with a suite of built-in financial skills (market hotspot analysis, stock screening, industry research reports, multi-stock comparison, daily stock reports) and a pre-configured "Stock Analyst" work partner equipped with 13 research capabilities, the system functions like an on-demand private research team.
1. Hotspot Review: Understanding Market Movements and News
The test case focuses on the three leading optical communication stocks — Xin Yisheng, Zhongji Xuchuang, and Tianfu Communication (collectively known as "Yi Zhong Tian") — during the week of September 14–18. The author uses the Market Hotspot Analysis skill with a detailed prompt:
Use market hotspot analysis to review optical module market from September 14 to 18
Compare Zhongji Xuchuang, Xin Yisheng, Tianfu Communication with ChiNext Index and CSI 300
Provide period price changes and real data charts, noting dates and adjustment basis
Separate company announcements, broker views, news reports, and market rumors
List which rise reasons lack direct evidence and next-step research questionsThe agent runs for 13 minutes, verifies real market data, retrieves and cross-checks catalytic news, and delivers a structured review report. It extracts clear trend characteristics from pure data and categorizes catalytic events into a table: company buybacks, industry exhibitions, media reports, and broker opinions. Unconfirmed rumors and outdated news are filtered out. Notably, the agent generates charts without explicit instruction.
2. Conditional Stock Screening: Finding Needles in a Haystack
With a preliminary direction, the author invokes the Stock Screening skill, setting five composite filters:
Non-ST optical communication A-shares
First-half revenue YoY growth > 0
First-half net profit positive and YoY growth > 0
First-half operating cash flow > 0
Daily turnover >= 500 million RMB
The agent executes for 12 minutes 25 seconds, outputting a screening summary table and a documentation file. It identifies three candidates suitable for deeper research and lists key financial items for follow-up verification.
3. Reading Research Reports: Turning 'Bullish' into Specific Hypotheses
After selecting focus stocks, the Industry Analysis Report skill fetches publicly accessible broker research report PDFs, saves them locally, then reads each to extract core viewpoints and synthesize a logically complete consolidated report.
4. Multi-Stock Financial Comparison: The Most Painstaking Part
The author calls the Multi-Stock Comparison skill, requesting a side-by-side analysis using 2026 H1 reports as the financial baseline and September 18 closing prices for valuation. The prompt:
Compare Zhongji Xuchuang, Xin Yisheng, Tianfu Communication
Financials unified to 2026 H1 and 2025 same period, market data unified to September 18 close
Compare revenue, net profit, YoY growth, gross margin, operating cash flow
Add inventory and receivables changes, then total market cap and PE TTM
Cross-check numbers, units, sources, and page numbers against original filings
Deliver Excel with formulas and research report, noting non-comparable itemsThis heavy computation runs nearly 30 minutes, covering data extraction, unit conversion, cross-verification, and comparative synthesis. The most practical deliverable is an Excel workbook retaining all formulas, with every metric annotated with the source filing's page number for instant audit.
5. Daily Stock Review: Price Action, Capital Flow, News, Technicals, Risks
For ongoing tracking, the Individual Stock Daily Report skill (also available via the Stock Research skill) generates a local daily report covering price review, news tracking, and financial watchpoints.
6. Stock Analyst Work Partner: Multi-Device Relay and Contrarian Logic Check
Beyond single skills, Doubao Work offers a "Stock Analyst" work partner pre-loaded with 13 research modules (company due diligence, announcement interpretation, financial statement analysis, delivery quality gate, research closure ledger). The author tasks it with a contrarian analysis on Xin Yisheng:
Based on Xin Yisheng 2026 H1 report as latest financial basis
Juxtapose three common bullish views with strongest counter-evidence
Can revenue/profit growth be supported by cash flow? Why did inventory and receivables change?
What company disclosures exist for 1.6T volume ramp? What specific quantities remain undisclosed?
What do equity incentives and capital actions imply for future research?
Attach original text and page numbers, finally list what next quarterly report must proveThe analyst runs for 16 minutes 44 seconds, producing an interactive 10-chapter research webpage. Multi-device sync proves valuable: after desktop generation, the link opens in Chrome with full context preserved. A follow-up question on the receivables YoY basis triggers immediate re-calculation from the H1 report and updates the webpage in place.
Summary: Transforming Equity Research Efficiency
The end-to-end workflow demonstrates a stark efficiency contrast:
Comparison: Traditional Manual Research vs. Doubao Work Financial Agent
Data Collection : Traditional – switching across 3-4 terminals, manually downloading dozens of reports; Doubao Work – one-click access to Wind, Tonghuashun, CNRDS, auto-fetch and summarize.
Financial Comparison : Traditional – manual copy-paste into Excel, hand-written formulas, half-day effort, error-prone; Doubao Work – 20-30 minutes auto-deliver Excel with formulas and source page numbers.
Cross-Device Collaboration : Traditional – multiple devices, file transfer via WeChat, progress often lost; Doubao Work – desktop background run, browser cross-device seamless follow-up, invisible sync.
Rigor and Anti-Fabrication : Traditional – hard to filter noise and fake news, easily misled; Doubao Work – auto-distinguishes announcements, news, rumors; leaves blanks when data missing, refuses to fabricate.
The greatest gain is freeing researchers from tedious data grunt work. What used to take a team a full day now takes an agent ten-odd minutes. Formula-driven Excel, page-cited report summaries, contrarian challenge webpages – all ready for spot-checks.
However, the author emphasizes: AI handles real data retrieval, standard bottom-up drafts, source citations, and noise removal; but the core judgment of whether a company is worth buying or a business logic holds remains with the human.
If you often exhaust yourself hunting data, building comparisons, and digesting filings, the author strongly recommends downloading the desktop client and running a full cycle to experience the efficiency leap of an on-demand AI research team.
Search "Doubao Work" on the web to download the desktop version and claim a limited-time 30-day free subscription.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
Old Zhang's AI Learning
AI practitioner specializing in large-model evaluation and on-premise deployment, agents, AI programming, Vibe Coding, general AI, and broader tech trends, with daily original technical articles.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
