Lastware WhoBuys

Who would buy this company?

Enter any company’s website. Get the companies most likely to want it, ranked by what they have actually bought before, each with its precedent deals. A buyer brief reads the site and quotes it.

Try: · · ·

We read the public website once to match it; nothing you type is shared. 3 free checks left today.

Recently checked

  1. PostHogProduct analytics software that ingests and stores analytics, errors, session replays, and business data for querying and action by product teams. It integrates data from over 900 sources to provide a unified context layer for diagnosing issues and managing feature flags, experiments, and error tracking. → IBM bought Tealeaf Technology (2012)
  2. LinearProduct development software designed for teams to plan and build products with integrated AI agent workflows. It supports collaboration by automating issue routing, prioritization, and streamlining tasks from drafting product requirements to code deployment. → Citrix bought Wrike (2021)
  3. Modern TreasuryPayments infrastructure providing a single API to onboard customers, open accounts, and move money across fiat and stablecoins with built-in compliance and real-time ledgering. It supports multi-rail payments, automated workflows, and reconciliation for businesses and platforms. → GoDaddy bought Poynt (2020)

Or start from a buyer

4,583 past acquisitions by 159 companies · data as of 2026-10-08

  • Microsoft 347 deals since 1986 Latest: Osmos, Fintool, Mahi Labs
  • Google 280 deals since 2001 Latest: Prestige LLC, HTC's Vive XR Team, Wiz
  • Cisco 235 deals since 1993 Latest: Aura Asset Intelligence, NeuralFabric, EzDubs, Inc.
  • IBM 210 deals since 1984 Latest: Micromuse, DataStax, SeekAI
  • Sony 154 deals since 1975 Latest: Big Yellow Dog Music, F.A.M.E. Recordings, Egg Firm
  • Oracle 148 deals since 1994 Latest: AdiInsights, Verenia, GloriaFood
  • Apple 140 deals since 1988 Latest: Q.ai, MotionVFX, Color.io
  • Yahoo 119 deals since 1997 Latest: Artifact, Wagr, Commonstock
  • Intel 108 deals since 1988 Latest: Granulate Cloud Solutions, Siru Innovations, Codeplay
  • HP 108 deals since 1958 Latest: Humane Inc., Poly, HyperX
  • Gen Digital (Norton) 91 deals since 1984 Latest: Avast, Avira, Luminate
  • SAP 81 deals since 1991 Latest: Dremio, Reltio, Prior Labs
  • Autodesk 81 deals since 1992 Latest: MaintainX, Payapps, Innovyze
  • Amazon 78 deals since 1998 Latest: Fauna Robotics, Pellicano Italy, MX Player
  • AOL 78 deals since 1994 Latest: AlephD, RYOT, Velos
  • Broadcom (CA) 76 deals since 1981 Latest: SourceClear, Veracode, Runscope
  • Twitter (X) 68 deals since 2008 Latest: Laskie, OpenBack, Breaker
  • Electronic Arts 67 deals since 1987 Latest: TRACAB Technologies, Codemasters, Glu Mobile
  • eBay 66 deals since 1998 Latest: Depop, KnownOrigin, TCGPlayer
  • VMware 64 deals since 2005 Latest: Broadcom's End-User Computing, Broadcom Software Gro, Datera
  • Siemens 62 deals since 1980 Latest: ASTER Technologies, RapidMiner, ZONA Technology
  • Meta 54 deals since 2005 Latest: Moltbook, Assured Robot Intelligence, Manus AI
  • Citrix 53 deals since 1997 Latest: Unicon, Wrike, Cedexis
  • Adobe 50 deals since 1990 Latest: Film Impact, Semrush, Abstract Notebooks

Free: 3 new companies a day, the top 3 buyers of each, 3 briefs a day.

One payment unlocks every shortlist, CSV export and unlimited buyer briefs, forever.

How buyers are ranked, and how often it's right

Questions

How are buyers ranked?

By precedent: companies tend to buy what they have bought before. We collected about 4,600 acquisitions by 159 companies from Wikipedia: big acquirers like Google, Microsoft and Cisco, mid-size ones like Stripe, Datadog, Atlassian and PayPal, and buyers found in thousands of company articles. WhoBuys reads your website, writes a short plain description of what you do, and compares it by meaning with every acquired company. Buyers whose past deals are closest rank first.

What do the percentages mean?

If one of the 159 buyers in our data acquires your company, the chance it is this one. We calibrated them on 1,583 past deals: when WhoBuys says 10%, that buyer was the real one about 10% of the time. They don’t say whether you will be acquired at all; that depends on things no website shows.

How accurate is it?

Tested on every acquisition since 2010 in the data, with that deal hidden: the real buyer was ranked first 30% of the time and in the top 5 56% of the time, against 22% for "always name the five busiest buyers" and 3% by chance. Real websites are noisier than those test descriptions, so treat the result as a well-founded shortlist, not a forecast.

Why are big tech companies at the top so often?

Because they buy the most, across almost every category. Below the main ranking, Specialists shows smaller buyers that once bought a company very close to yours, and the buyer brief explains each fit and what could stand in the way.

Why isn’t my obvious buyer listed?

Our data covers 159 buyers with public acquisition records on Wikipedia; many active acquirers, private ones especially, aren’t there yet. The buyer brief can also name companies your own website mentions as a partner, platform or competitor.

Is anyone named here actually interested?

We don’t know, and we never say so. WhoBuys shows strategic fit from public precedents. It is not a rumour, a prediction of a deal or investment advice.

What do I get for free?

Three new companies a day with their top 3 buyers and specialists, any company someone already checked, and 3 buyer briefs a day. A one-time $0.99 unlocks unlimited checks, all 10 buyers, CSV export and unlimited briefs, forever. No subscription.

Is my company’s page public?

Yes: a check reads only the public website, and its result has a shareable page. WhoBuys is independent and not affiliated with any company named. It runs on Contextrie and Bref UI, both open source on GitHub, if you want to build something like it.

Open source underneath

Built with

WhoBuys is a thin layer on open-source packages. Star them, fork them, build your own micro.

Contextrie

60

Open-source context and state for AI: it keeps only what matters for the question.

In WhoBuys: Writes each buyer brief: reads the company’s website, keeps the passages about what it sells, to whom and who it works with, so every named buyer quotes a real source.

feuersteiner/contextrie

Bref UI

25

Open-source Svelte components for clean, accessible interfaces.

In WhoBuys: The website check, buyer shortlists, buyer pages and the unlock dialog.

feuersteiner/bref-ui

Also built with SvelteKit · Vercel AI SDK · Zod · libSQL

Lastware: software that assembles itself from shared state, capabilities and generated interfaces.