Oops! Something went wrong while submitting the form.
Researches market size, growth, players, buyers and trends, cites the source for each number, and marks separately anything that could not be verified at a primary source.
The recording is a real session. The sheet on the right is what it produced.
Sai opens each profile, pulls the signal, and writes the row, live, in a real browser.
Eight columns, sorted by score, with a source link behind every claim.
The market or category to size A time window for trends A delivery format — branded report or Google Doc
Market size and growth figures, each with its source Where estimates differ, the figures side by side rather than averaged Main players and positioning Buyer segments and reported pricing Trends over the stated window A section listing claims that could not be verified at a primary source
Run it monthly to re-check the sizing and record what changed since the previous edition.
Market sizing is the practice of estimating how much revenue a market represents and how fast it is changing. It is normally expressed as a current figure, a forecast figure, and a compound annual growth rate connecting them.
The standard framing separates three quantities. Total addressable market is the revenue available if every potential buyer bought. Serviceable addressable market narrows that to the segment a given product can serve. Serviceable obtainable market narrows it again to the share realistically winnable. These are estimates built on assumptions, not measurements.
Market sizing appears in investment memos, board materials, business cases and product strategy documents. In each case it functions as a premise — a number that other conclusions are built on top of.
That is what makes its reliability worth examining. A sizing figure rarely appears alone; it appears as the foundation of an argument.
Two credible research firms will size the same market at figures that differ by several times. This is normal, and it is usually not an error by either.
They are counting different things. A market defined as software licences is smaller than the same market defined to include services, infrastructure and adjacent spend. Where a category is new, no standard definition exists yet, and each firm draws its own boundary.
The base year differs. A fast-growing market sized in 2024 and in 2026 produces different starting points, and forecasts anchored to different bases diverge further with each projected year.
Forecast horizons compound. A figure for 2030 and a figure for 2035 are not comparable even at identical growth rates. The further out the horizon, the more the terminal number reflects the assumed growth rate rather than any observation.
Methodology differs. Bottom-up sizing from unit counts and prices produces different results from top-down sizing from adjacent market shares.
The practical consequence: a spread between estimates is information about definitional disagreement. Averaging them removes that information and produces a number that no source supports.
There is a more useful distinction than which forecast to trust, and it is between figures that were measured and figures that were projected.
An example of the first. Menlo Ventures' 2025: The State of Generative AI in the Enterprise, published 9 December 2025, reports that enterprise AI spend reached $37 billion, up from $1.7 billion in 2023 — split roughly $19 billion on applications and $18 billion on infrastructure and model APIs — and that this represents around 6% of the global SaaS market.
That is a figure about spending that has already happened. It can be wrong through sampling or definition, but it is not a projection.
A 2035 terminal value is a different kind of object. It is an assumed growth rate applied to a base, and its reliability degrades with distance.
Both belong in a sizing report. Presenting them as the same kind of number is where the report starts to mislead, and the near-term measured figure is usually the one a business case should rest on.
Founders raising or planning. The sizing slide is scrutinised, and a figure that cannot be traced to its source is a liability in diligence.
Product and strategy teams evaluating a category. The question is whether a market is large enough to enter, and the answer depends on which definition of the market is used.
Corporate development and investors. Several markets need sizing on a comparable basis, and inconsistent definitions between them produce inconsistent conclusions.
Consultants and analysts producing client work. Every figure will be checked, and unattributed numbers are the ones that get challenged.
Desk research by hand scores full marks on all three quality columns. An analyst who reads the primary sources and records what they could not confirm produces the most reliable output in the table, and the constraint is that it takes days and is repeated from scratch each quarter.
The third column is where the automated methods differ from each other. Citing a source is now common. Recording what was rejected is not.
Sai researches the market you name and produces a report covering size and growth, the main players and their positioning, buyer segments and reported pricing, trends over your stated window, and where the gaps are.
Each figure carries its source. Where estimates conflict, they appear side by side with their definitions rather than reconciled into one number.
The report also contains a section for what was excluded: claims that appeared only in secondary citations or search summaries, and figures that are widely repeated but not published at a primary source. These are listed rather than used.
Output is a branded report or a Google Doc.
Market sizing has a short shelf life. New estimates are published, base years shift, and figures that were unverifiable acquire primary sources — or fail to, and become more doubtful for having circulated longer without one.
Running the task quarterly produces a new edition against the same definitions, and the comparison between editions is often more informative than either edition alone: which estimates moved, which players changed positioning, and which flagged claims were eventually sourced.
Keeping the market definition constant between runs is what makes the comparison valid. A changed definition produces a changed number for reasons that have nothing to do with the market.
For competitive and trend research beyond sizing, AI market research covers the broader workflow.
For positioning one property against named competitors, comparative market analysis is the narrower exercise.
For recurring competitive tracking rather than a point-in-time report, a weekly competitive intelligence report in Google Docs covers that cadence.
