This is not a generic chatbot. It was built specifically for CSE investors — trained on over a decade of Sri Lankan financial news, SEC filings, earnings releases, and macroeconomic data. Here's exactly how it works.
At a Glance
What the model knows
📰
Training Corpus
40,000+ articles
Sri Lanka financial news, 2014–2026
🤖
Underlying Model
Claude (Anthropic)
Retrieval-augmented with CSE-specific context
📊
Market Coverage
CSE — All sectors
ASPI, S&P SL20, sector indices
🔄
Data Freshness
Updated continuously
New articles ingested as published
Data Sources
Where the knowledge comes from
The AI is grounded in a curated retrieval database built from the following primary sources. When you ask a question, it retrieves the most relevant documents before generating an answer — it does not rely on static training weights alone.
PRIMARY
EconomyNext & Daily FT — The two largest Sri Lankan financial publications. Every article published since 2014 is ingested, cleaned, and embedded into the retrieval index.
PRIMARY
SEC Sri Lanka Filings — Interim financial statements, audited annual reports, and material disclosures published to the Colombo Stock Exchange. Earnings data is structured and queryable by ticker and period.
SECONDARY
CBSL Publications — Central Bank of Sri Lanka monetary policy decisions, economic outlook reports, and interest rate announcements.
SECONDARY
InwestOut Research — Our own published equity research reports, quality-of-earnings analyses, and sector deep-dives, which are fully searchable by the AI.
MACRO
IMF & ADB Reports — International Monetary Fund Sri Lanka Article IV reviews and Asian Development Bank country outlooks, providing macro context for market questions.
Engineering
How it's built
The AI uses a Retrieval-Augmented Generation (RAG) architecture. This means it doesn't just rely on what a language model memorised during training — it retrieves fresh, specific evidence before every response.
1
You ask a question
Your query is converted into a semantic embedding and matched against 40,000+ indexed documents in our vector database.
2
Relevant documents are retrieved
The top-k most semantically similar articles, filings, and data points are surfaced — ranked by relevance to your specific question, not just keyword overlap.
3
Claude synthesises an answer
Claude (Anthropic's model) reads the retrieved documents and generates a grounded, cited answer. It cannot fabricate facts that aren't in the retrieved context.
4
Sources are cited automatically
Every substantive claim in the response is linked back to the original article or filing. You can verify every fact directly at its source.
Analytical Framework
Quality-of-earnings logic
The AI is tuned to apply the same analytical lens our research team uses when reading CSE financial statements. When given earnings data, it looks for the following signals:
💵
Cash Flow Quality
Operating vs reported
Does cash flow from operations track net profit, or is profit accrual-driven?
📦
Working Capital
Inventory & receivables
Is receivables growth outpacing revenue? Are inventories building without sales growth?
🏦
Debt Structure
Tenure & coverage
Short-term debt rollover risk, interest coverage ratio trends, debt-to-equity trajectory.
Transparency about limitations is as important as explaining capabilities. Do not use this tool without understanding the following:
⚠️
It does not provide buy, sell, or hold recommendations. The AI explains what the numbers show — it does not tell you what to do with them. All outputs are analytical observations, not investment advice.
⚠️
Knowledge has a cutoff. While the retrieval database is updated continuously, very recent events (within the last 24–48 hours) may not yet be indexed. Always check primary sources for breaking news.
⚠️
It can make mistakes. Language models can hallucinate or misinterpret data. Sources are always cited — verify important claims against the linked original documents before acting on them.
⚠️
Not a substitute for a licensed adviser. InwestOut is not registered as an investment adviser under the SEC Sri Lanka Act. Consult a SEC-registered financial adviser before making any investment decisions.
⚠️
Output quality depends on question specificity. Vague or very broad questions yield less precise answers. Questions that reference a specific company, period, or event get the best retrieval quality.
Full disclaimer: InwestOut AI is an informational research tool. All outputs are for educational and research purposes only. Nothing on this platform constitutes investment advice, a solicitation, or an offer to buy or sell any security. Past market sentiment, analyst commentary, and historical data do not guarantee future results. InwestOut.com and DamithInvest are independent research entities and are not registered as investment advisers or brokers under applicable Sri Lankan law. Always verify information against primary sources and seek advice from a qualified, SEC-registered financial professional.
Ready to use the AI analyst?
Click below to launch the chat widget — it's available on every page of the site via the button at the bottom-right corner.
