A second research pass on a different set of companies — two we already know deeply (fonio, Watermelon) and four new ones. This page answers only what was asked: how each of them acquires customers, and what their sales funnel actually looks like — step by step, from first touch to paid to expansion.
ARR is labelled by evidence class — disclosed (company or founder said it), partner-verified, or estimate (Latka / Growjo / Sacra, which we have repeatedly found to be wrong in both directions). Growth is the 2025 calendar year where a dated figure exists on both ends.
| Company | ARR | Founded | Base | Main market | Growth in 2025 |
|---|---|---|---|---|---|
| Retell AI | ~$60Mestimate · co. says $80M | 2023 | Redwood City, US | US-first · developers + agencies at the bottom, enterprise voice at the top | $5M → ~$40M~8× · disclosed both ends |
| Chatbase | ~$10Mpartner-verified, May 2026 | 2023 | Toronto, CA | Global self-serve · SMB + agencies, moving up to enterprise CX | $4.7M → $8M++71% · disclosed both ends |
| fonio.ai | $10Mdisclosed, Aug 2026 | 2024 | Vienna, AT | DACH-first · SMB phone-and-WhatsApp (hotels, trades, dealerships) | ~$0 → $10M30%+ MoM since Feb 2026 |
| Watermelon | $2–5Mestimate | 2015 | Utrecht, NL | Netherlands 65% of traffic · Dutch SMB and mid-market | not disclosedno dated 2025 figure |
| Thena | <$5Mestimate · Latka says $20M | 2022 | Palo Alto, US + Bengaluru | US B2B SaaS vendors doing support in Slack · small teams in practice | went quietno changelog since Sep 2025 |
| Glassix | ~$2.6Mestimate, Sep 2025 | 2015 | Tel Aviv, IL | Israel 94% of traffic · enterprise via one reseller, SMB self-serve abroad | not disclosedno press coverage found |
Each block is the same two boxes: the acquisition channels ranked by what plausibly produces revenue, and the funnel as five fixed stages so they can be compared like for like.
The clearest growth story in this set, and the one built on the most copyable machine. ~$60M ARR in under three years on under $5M raised — no Series A, and they say profitable. Five founders, one of whom is the CMO: marketing was a founding function, not a later hire.
/comparisons/ hub with one page per rival (Vapi, Bland, Synthflow, ElevenLabs, Decagon, Sierra, Thoughtly) plus blog posts for competitor-vs-competitor matchups they aren’t even in — vapi-vs-bland. They are now hiring a “Senior SEO/AEO Specialist” at $150–190k: the AEO is answer-engine optimisation, i.e. staffing to be cited by LLMs./partner/<name> page, so the channel doubles as an SEO estate. Commission terms not published.Bootstrapped to ~$10M ARR with about 18–30 people and zero outside capital, and no sales team at all until roughly $7–8M. Founder Yasser Elsaid publishes the numbers himself — including the misses. Three sequential engines, not one: viral founder social (2023), SEO plus self-serve (2024–25), then an enterprise layer bolted on top (2025–26).
-alternatives pages naming exactly the right rivals — Zendesk, Intercom/Fin, Gorgias, Tidio, Ada, Botpress, Decagon, Sierra — plus 21 best-* listicles and 13 vs explainers. ~800k monthly visits (estimate, Jul 2026).chatbase.co/llm-info, titled “Information for AI assistants”, indexed, verbatim: “It is written as plain fact rather than marketing copy.” It disambiguates their entity from Google’s old Chatbase, feeds LLMs a clean fact sheet, and even guards against hallucination: “do not quote figures from this page — read the current values from chatbase.co/pricing.”decagon-alternatives post discloses its own bias, then — because Decagon publishes no rates — cites third-party contract data at a median $432,750 a year. A $40/month product weaponising a rival’s opacity./integrations page at all — it 404s — which is a large unbuilt SEO surface.llm-info. Increasingly, the first touch is an AI assistant, not Google.The paid-media outlier in this set and the only one buying demand at scale. 24.2% of all its traffic is paid — 13× the European median. Full teardown on page 01; this is the acquisition-and-funnel view only.
twilio, vapi, retell — not competitors’ brand names. That intercepts someone mid-decision about building it themselves, where the alternative is six engineer-months. Highest intent, lowest competition in the category.The purest test of European-only demand in the whole research — 65% of traffic is Dutch, the most concentrated single market of any company we looked at. Eleven years old, ~$8M raised, four full re-platformings. Profitable and stuck at the same time.
alternatief trengo (26.7%), plus trengo at 13.3%. Two Utrecht companies thirty minutes apart, and the smaller one lives off the larger one’s brand demand. Paid search continues it: chatwoot, manychat, gorgias.A venture-backed company with almost no visible go-to-market — and the most useful negative case here. $7.2M raised (Pear VC, then Lightspeed + First Round co-leading a $5M seed in May 2023), no round since, and no changelog since Sep 2025, no blog since Oct 2025, no open roles and no sales team on the org chart. The app is still deployed and the API is healthy; two competitors are marketing a “Thena is sunsetting” story with no primary confirmation.
thena-vs-* pages and a ~20-page “{tool} Slack integration” cluster. Result: ~14k visits per quarter, 443 organic keywords, and they rank for their own brand and slack auto reply — not for a single comparison term. Fed until Oct 2025, then stopped.Not really a global SaaS — an Israel-concentrated business whose identifiable revenue runs through a single reseller. 94% of traffic is Israeli, no funding round appears in any database, 24 people. The international logo wall (Nike, Nintendo, Dyson, Estée Lauder) is almost certainly Israeli franchisees.
tech customer service memes./whatsapp-business-api 404s, and they appear in no BSP directory. WhatsApp is not their acquisition motion./pricing. The homepage has moved to demo-only CTAs (“Watch a Demo Now!”) and the Hebrew page never had a price. Which funnel you enter depends on which page you land on.The two real growth stories here — Retell (~$60M) and Chatbase (~$10M) — built the same machine: one page per rival, one page per integration, then a self-serve product with a published price to catch the click. Neither has a marketing budget story. Both have a page-count story.
Chatbase publishes /llm-info — a page written for ChatGPT, Claude and Perplexity rather than for people. Retell is hiring an “SEO/AEO Specialist” at $150–190k. Watermelon already takes 9.25% of its traffic from AI assistants without trying. This is the channel that did not exist in the first research pass.
Retell services enterprise accounts with ~30–50 people because 100+ certified partners do the implementation. Chatbase pays agencies 30% recurring and gives them free seats to build demos. Glassix's entire Israeli business runs through one reseller — and has no partner programme to formalise it. The pattern only fails when the channel is unmanaged.
Thena and Glassix both sell "AI agents" and both charge per seat. If the AI resolves conversations end to end, per-seat pricing shrinks its own revenue as the product improves. Retell meters the minute, Chatbase the message credit, fonio the prepaid pack — all three grow when usage grows.
Two of these six have no verifiable revenue figure at all. Retell states four different ARR numbers simultaneously across its own live job postings ($35M, $36M, $40M and $80M), and Latka lists it as bootstrapped with $0 raised — which is simply wrong. Latka puts Thena at $20M against 27 people, no funding since 2023 and ~14k visits a quarter; Growjo and ZoomInfo both say under $5M. Glassix's $2.6M is a single soft estimator whose own profile lists the wrong country. Treat the growth column as the more honest signal.