Boom Leverage
Live · free tier available

Search across every Thai company's file — compare years, check Opp Day against the filing

Management's own words are buried in tens of thousands of One Report PDFs and Opp Day sessions from ~918 listed companies. Ask in Thai or in English — “supply chain disruption” — and get the real sentences from every company that discussed it, laid side by side across years and against what was said at Opp Day, with the page or the minute behind every line. Not an AI summariser: what we cannot verify against the source, we do not show you.

What you can ask

Questions this actually answers

Write it the way you would say it. The engine searches every company and every year, then hands back the source passage and the filing link so you can check it yourself.

“impact of a weaker baht”

→ The companies, and the sentences where management discussed FX, with the fiscal year and the source filing beside each one

“weakening purchasing power at the low end”

→ Only the passages that discuss bottom-of-pyramid demand, pulled across many companies in one click

“rising energy and electricity costs (filtered to FY2022–2023)”

→ Who flagged it early and who flagged it late — the sequence of recognition, across years

“consumption outlook (filtered to CPALL, CRC)”

→ Each company's own words placed side by side — who said what, and where it doesn't match the numbers, every line cited. The conclusion is still yours

“higher ECL provisioning driven by NPLs”

→ The credit-risk narrative across the sector, with the source document for each claim

“shifted from caution to expansion (latest filings only)”

→ Tone changes between reporting periods, evidenced by the passage that changed

Three ways people use it

Scan the market, check a company, read back through time

1

Market scan

One question across every listed company. Boom Leverage AI lines up who gains, who loses and who hedged, one cited line per company — then ask the follow-up in chat.

2

One-company health check

Pick a ticker and get every finding proved against its filings, ranked by weight: the auditor's opinion first, then the risk signals, then the most recent year.

3

Year-by-year read

Walk one company back through time a year at a time, and see where the story management tells changed — the passage that moved, with its page.

Screenshot of a Terminal search for “impact of a weaker baht”: the Boom Leverage AI summary tables companies helped and hurt by a weaker baht, each line ending in a numbered citation
Market scan: “impact of a weaker baht” — every company's line side by side. Every [n] opens the passage it came from.
Screenshot of an Opp Day result: PRTR's management, in Thai, revising its revenue target down to 5–10% from 10–15%, with a button to open the recording at that minute
Opp Day: what management said out loud — here, a revenue target cut — linked to the minute in the recording.

Unedited captures from terminal.boomleverage.com, October 2026.

What you get

Not raw data. Evidence you can put in front of a committee.

Type a concept → real MD&A passages from every company, years back, in seconds
See who raised an issue first, and whose tone moved — across companies and across years
Every line shown points back to a page and a position in the original Form 56-1. What could not be matched is withheld, counted and explained
Hundreds of PDFs' worth of reading collapsed into a query
Who it is for

Built for three kinds of reader

Analysts and portfolio managers

Cover a sector or a theme in an afternoon instead of a fortnight, and screen the whole market qualitatively in one pass

Fundamental and deep-value investors

Interrogate the management narrative of every name you hold, against its own history and against its peers, immediately

Strategy, IR and advisory

Competitive intelligence — how rivals frame the same pressure, and the exact filing where their framing changed

How it works

Step 1

We index the whole market's One Reports, years deep

The MD&A section is extracted from public filings (Form 56-1 One Report / annual report) for every listed company and embedded for search

Step 2

You type the idea you are chasing

Vector search finds passages whose MEANING matches, across every company and every year — not just the ones that used your exact words

Step 3

You get the real passage, cited — and the unprovable is held back

Before anything renders, each line must match word-for-word against text in the original filing. What fails is withheld, and the count and the reason are shown to you. Nothing passes silently

A real capture of the Terminal's search screen: below the results, a notice states that 3 further findings were withheld because they could not be traced back to the source file word-for-word, and gives the reason
The search screen every user sees. The product declares what it withheld — not just the sales page.
Evidence, not just an answer

We do not show what we cannot prove

Every line you see has been matched back to real text in the original Form 56-1 — page and position on the page, openable immediately. What cannot be verified word-for-word we withhold, and we tell you how many and why. Not deleted, and not decided on your behalf: inside the terminal you can ask to see the near-misses yourself.

verified against the source filing(FY2021 onward)
98.9%
506,759 of 512,593 findings
withheld — never shown as a verified result
5,834
scanned image · no source file yet · text does not match word-for-word

Across 5 published axes, FY2021 onward · as of 20 Nov 2026, 09:41

beta · own denominatorOperating quality98.94%verified

How these numbers are measured

Measured over Results & financial position, Risk factors, Business & competition, Governance and Auditor opinion — the axes we publish coverage for, FY2021 onward. Material the terminal serves only on request is proved by a link to the publisher's original and measured separately; the full breakdown is on the coverage page.

As of 20 Nov 2026, 09:41

Newly opened · betaOperating quality

98.94%verified against the source · 111,556 of 112,756 findings

This axis only just opened, so its number is reported on its own denominator — not blended into the figure above — and every line clears the same verification gate.

FY2016–2020 archive · on the ALPHA desk licence

The earlier years are extracted, indexed and verified line by line against the source filings to the same standard as above — and they are included with ALPHA, the desk licence, which reaches FY2016 onward. The self-serve plans reach FY2021 onward. Any finding older than FY2021 is withheld unless its quote has been matched to the original filing.Archive coverage: 254,039 of 255,975 findings verified — a different denominator from the figures above, and not added to them

Request the coverage report and access to our pipeline dashboard

Screenshot of the Terminal's source viewer: a SIAM Form 56-1 One Report, page 10, with the quoted sentence about a stronger baht boxed on the filing's own page
Unedited capture from terminal.boomleverage.com · October 2026

Why us

  • Semantic vector searchmore

    you search meaning, not strings, so you find the company that said it in words you did not think of, in Thai or in English

  • An MD&A corpus covering ~918 listed companies, years deep, tens of thousands of documents, already indexedmore

    nothing else lets you query Thai filings this way

  • Continuously refreshedmore

    new and revised Form 56-1 filings are pulled from the public source twice a day, catching new quarters, late filers and amendments, and always serving the latest version

  • We do not show what we cannot prove. Every line is matched back to real text in the original filing before it reaches you; anything that fails word-for-word verification is withheld, counted and explainedmore

    not deleted, and not decided on your behalf

  • Boom Leverage AI lines the results up for comparisonmore

    who gains, who loses, who hedged, one line per company with a [n] back to the page or the Opp Day minute; a line without a source is withheld, and withheld output is not charged

  • Built by someone who ran risk and data inside a Thai bank, and knows what an MD&A is worth reading for

And what it is not

  • Not a buy or sell recommendation, and not a price target
  • Not a price forecast
  • Not an AI summariser, and not an AI that reaches a verdict for you — Boom Leverage AI puts what management wrote next to what they filed, one cited line at a time, and withholds any line it cannot tie to a filing, so the conclusion stays yours
  • Findings derive from documents the companies published themselves. Check the source before you act on it

Pricing

Pick the plan by how far back you need to reach. Move up or down at any time.

Start free

FREE · evaluation

Free

Latest filings only (at least the last 5 quarters, guaranteed) — enough to find out whether this changes your work before you pay for it · no export

Opp Day: the last 2 quarters

100 credits/day

Try it free

DELTA · 4 years of history

~฿799 / mo

≈ $25 / mo

Put what management says today next to what they said 3 years ago · 50-row Excel/CSV export

Opp Day: the past 4 years

150 credits/day

Subscribe
Most popular

GAMMA · 6 years of history

~฿1,399 / mo

≈ $43 / mo

The deepest self-serve window — follow a story across a full cycle · 50-row Excel/CSV export

Opp Day: the past 6 years

350 credits/day

Subscribe
Team / desk licence

ALPHA · desk licence

Contact us

10 years of history, FY2016–2026 · multiple seats · 100-row Excel/CSV export (DELTA and GAMMA export 50) · API / integrations as a design partner · 1,500 credits/day — priced per team on seats and scope, contact us · the auditor axis is open to DELTA and GAMMA too, charged in credits

Opp Day: the past 6 years

1,500 credits/day

Talk to us

Priced on seats and scope of use

1 credit = one whole-market search, or a filtered search over up to 25 companies · beyond that, +1 credit per additional 25 companies (filtering 50 companies costs 2) · the quota resets daily

What each action costs in credits, and AI limits per plan

Need the raw data in your own stack? (quant / systematic / enterprise)

A REST API that pulls the whole market's MD&A in bulk into your own models. Developer & Enterprise plans (contact sales) are on the terminal.

See the API plans

Rather learn it hands-on first? Boom Field Lab runs short workshops on real Terminal cases, in Thai — selected workshops include a month of GAMMA.

⚠️ This service is a search and analysis tool over public disclosure documents (MD&A), built to accelerate research. It is not investment advice, it is not a solicitation to trade any security, and it does not warrant the completeness or accuracy of the source filings. Verify the original document before acting on anything you find here.

No time to run the search yourself? Commission a report.

On-Demand Research: we point the same engine at one company or one question and write the memo, every claim cited to the filing page — Quote on request (priced per scope — one name, a sector, or one question).

Commission a report
Live

Ask the whole market your question, right now

Sign in with Google and start searching, no card required. The free plan covers the latest filings on 100 credits a day; upgrade when you need to reach further back, or search more often.

Open the Terminal, free

Frequently asked

How is this different from Google, or from Ctrl+F?+

We use semantic (vector) search — meaning, not literal characters. Type “energy costs” and it surfaces companies that wrote “electricity tariff pressure”, “fuel price escalation”, or the Thai equivalent, because the same multilingual embedding model is used to index and to query. You can ask in English and land inside a Thai paragraph, or the reverse. Ctrl+F searches one file you already have open; this searches every company, every quarter, every year, at once.

Why not just upload the filings to ChatGPT or Claude myself?+

You can, if the job is one company and one document. The job here is tens of thousands of documents across ~918 companies and several years, simultaneously — many multiples of any chat context window, and uploading file by file does not scale. We index the market ahead of time and search it by vector in seconds. But the difference that matters more than speed is this: a chat answers from what it remembers, and we answer from what we can prove. Every line must match back to text in the source document before it reaches you; anything that cannot be matched is withheld rather than guessed. And the corpus refreshes twice a day, instead of being frozen inside a model's training data.

How much is covered — how many axes, and how far back?+

The Terminal searches six axes extracted from the real documents: financials (MD&A) · risk factors · governance · business & competition · auditor opinion (the audit report and its Key Audit Matters) · operating quality (beta — six lenses over the MD&A: customer concentration, FX, inventory, working capital, labour, litigation). Search them together or one at a time. How far back depends on the plan you hold: the self-serve packages reach FY2021 onward, and ALPHA, the desk licence, reaches FY2016 onward — ten years. The two eras are measured and reported separately and are never blended into one figure: the headline verified share on our pages covers FY2021 onward, and the earlier years carry their own coverage number. An axis that has not cleared our verification standard is not opened for search, even when the data is already in the system.

What language are the source documents in, and what happens with scanned files?+

We prefer the English filing. When a company publishes both, the ranking picks the English document first and falls back to the Thai one only when there is no English version — and you can still query in either language, because the embedding model is multilingual. For filings that are pure scanned images with no text layer, we do not scrape the raw text: the text comes from reading the page with a vision model, and OCR is used only to recover word coordinates so the quote can still be boxed on the original page. The decisive gate is not the parser, it is the verification: every line must match back to text in the original Form 56-1, and anything that fails is withheld from the results (measured 16 Aug 2026 across the whole index as it stood then — a different scope from today's headline, which is on /research: 370,329 of 373,534 lines passed = 99.14%; the remaining 3,205 were withheld). We are not asking you to believe our parsing is perfect. We are giving you a way to check it, every time.

Where does the data come from, and is this legal?+

Both the MD&A and the auditor's report come from the Form 56-1 One Report / annual report — documents the company is legally required to publish — and every result links back to the source file so you can open it yourself. We sell the search and the analysis over those public documents. We do not resell any vendor's proprietary data.

Why price on years of history, and how are credits charged?+

The further back you can reach, the more of a trend you see and the more clearly a shift in management's tone stands out — so you buy the depth your work needs. Free covers the latest filings (100 credits/day) · DELTA 4 years (150/day) · GAMMA 6 years (350/day) · ALPHA, the desk licence, 10 years back to FY2016 (1,500/day). Credits are charged per ACTION, not per minute: one search = 1 credit (filtering beyond 25 companies adds 1 per additional 25) · opening the source file with the passage boxed on the page = 2 · a full single-company health check across all axes = 3 · scanning your portfolio = 2 · unlocking one company's auditor opinion = 5 (the price is on the button before you press it, and it is charged once per company per day) · following a link out to the regulator's site = free · re-running a query that is still cached = free.

Is there a free tier, and what does it include?+

Yes — the free plan searches the latest filings (at least the last 5 quarters, guaranteed) on 100 credits a day, with no card. Enough to establish whether the tool changes your work before you pay for it. The same daily budget covers every action, including auditor detail: unlocking one company's auditor opinion costs 5 credits (the price is on the button before you press it), so a free day can open several names alongside your searches. What the free plan does not give you is depth or export — it stops at the latest filings, and reaching back years or exporting results across a book is what the monthly plans are for. What your account can reach right now is shown in the box above the pricing, read directly from the system.

Is it live? How do I start?+

It is live today — sign in to the Terminal with Google and search immediately, no card required. The entitlement and history depth attached to your account are shown in the pricing section above.