AGENT 01
Price Analysis
Judges current position from chart and trading flow.
- Technical indicator calculation
- Real-time price trends
- Support/resistance zone detection
R&D · AI AGENTSeven AI agents analyze
Seven AI agents analyze
one stock, each from a different angle
Lucidasoft's R&D lab researches multi-agent architectures. Instead of handing everything to one giant model, agents with different roles each analyze from their own angle, and the results are combined into a report. We validate this in a live Korean stock-analysis service, AI-Stock .
Try an Analysis Now
www.ai-stock.it · Just type a ticker and 7 agents analyze it instantly
7Analysis agents
Real-timeSSE progress display
100-ptComposite score
Liveai-stock.it
WHY MULTI-AGENT
Why Seven, Not One
Stock analysis mixes fundamentally different kinds of judgment. Reading a chart, reading a financial statement, and reading the tone of the news each need different data and different criteria.
Different Kinds of Judgment
Chart-reading criteria and financial-statement criteria are different. Cram both into one prompt and one of them gets diluted.
Bias Doesn't Spread
Because agents don't see each other's conclusions, one agent's misjudgment doesn't drag the overall conclusion with it.
Reasoning Is Traceable
Instead of just a final score, we keep each agent's score and reasoning, so you can see why a conclusion was reached.
Easy to Add Perspectives
Need a new analysis criterion? Just add another agent - no need to touch the existing ones.
AGENTS
7 Analysis Agents
Each agent runs independently, producing a score and opinion for its own domain. Since they don't reference each other's conclusions, one agent's bias doesn't spread to the rest.
AGENT 02
Financial Analysis
Assesses company fundamentals using DART disclosure data.
- DART annual results lookup
- Profitability/stability metrics
- Earnings trend assessment
AGENT 03
Sentiment Analysis
Reads how the market currently views this stock.
- Analyst consensus
- Key catalysts summary
- Key risks summary
AGENT 04
Risk Assessment
Separately evaluates downside probability and magnitude.
- Risk metric calculation
- Mitigating factors check
- Geopolitical risk factored in
AGENT 05
Sector Comparison
Looks at relative position within the same industry.
- Market position analysis
- Core competitive moat
- Competitive threats
AGENT 06
Valuation
Quantifies whether the current price is fair.
- Valuation metric calculation
- Key value drivers
- Over/undervaluation call
AGENT 07
News Analysis
Finds recent stories likely to move the stock price.
- News volume tracking
- Key theme extraction
- Issue impact assessment
FINAL
Final Report
Combines all 7 agents' results into a composite score and investment opinion.
- Composite score out of 100
- Score comparison across agents
- SWOT analysis summary
- Target price and investment opinion
HOW IT WORKS
See the Analysis Happen Live
You're not left waiting until it's all done. Each agent's progress streams live via SSE (Server-Sent Events), so you can see exactly what's being analyzed right now.
01
Pick a Stock
Type a ticker or company name and pick it from autocomplete. Analysis starts the moment you select it.
02
7 Agents Run
Each agent gathers the data it needs and analyzes independently, with live progress shown as cards.
03
Report Generated
Results are combined into a report with a composite score, investment opinion, and SWOT analysis.
✓ Price
✓ Financials
✓ Sentiment
✓ Risk
✓ Sector
Valuation
News
Generating report
[Price] Analysis started
[Price] Done · score 78
[Financials] Done · score 71
[Sentiment] Done · score 65
[Risk] Done · score 62
[Sector] Done · score 74
[Valuation] Analyzing...
TECH
What This Architecture Gave Us
AI-Stock was built around stocks, but the architecture itself is domain-agnostic. Any task with multiple, distinct judgment criteria can be moved into the same shape.
Agent Orchestration
We designed our own architecture for running multiple agents in parallel and merging their results into one conclusion.
Real-Time SSE Streaming
Progress is delivered instantly instead of making users wait through a long task - a practical safeguard against drop-off.
External Data Integration
Normalizes external data in different formats - DART filings, market prices, news - and feeds it to the agents.
Enforced Structured Output
AI responses aren't left as free text - scores and fields come back in a fixed structure, ready for the UI and stats immediately.
Externalized Judgment Criteria
Each agent's role and judgment criteria live in configuration, not code, so they can be tuned without a deployment.
Failure Isolation
If one agent fails, the rest of the analysis and report generation continue unaffected.
APPLY
Applicable Beyond Stocks
We turn research results into prototypes and consult with SMBs on putting them to work in real operations.
Manufacturing/QA
Multi-angle Defect Analysis
Agents dedicated to equipment logs, operator records, and material history each analyze to narrow down root causes.
Public Procurement
Bid Eligibility Assessment
License requirements, track record, profitability, and schedule risk are reviewed separately to decide whether to bid.
HR
Multi-Angle Application Review
Career fit, tech stack, and organizational fit are each scored to propose interview priority.
Contracts/Legal
Contract Risk Review
Poison-pill clauses, financial terms, and schedule terms are reviewed separately, with negotiation points summarized.