An AI assistant that answers the phone and WhatsApp for small businesses. No pivot, no category invention, no enterprise sales motion. What fonio actually built was a go-to-market engine that funded itself — €300 of prepaid credits paying for the ads that brought the next customer — and then changed the pricing model at the exact moment the product had earned it. This page takes it apart step by step.
When fonio comes up, everyone goes straight to the technology. How the AI phone agents work. How hard they are to build. Where the technical moat is.
Anyone can get a phone agent working in a demo. Getting it to handle 2 million real calls a month is a different problem entirely. Matthias Gruber built what most AI voice companies never reach: they are not sitting on top of telecom infrastructure, they are inside it, and GDPR-compliant with it. That alone is hard.
But the real advantage isn't technical. It's focus, customer obsession, and a GTM motion that compounds at every step — where each stage pays for the next one instead of needing a round to fund it.
Of the five companies researched, fonio is the only one whose growth is driven by paid performance marketing into an SMB base — the same shape of problem we solve. Sierra, Fin and Ada are top-down enterprise sales motions where CAC is a salesperson and a nine-month cycle. Their playbooks are interesting; fonio's is transferable.
Sequence is the strategy. Each step is only affordable because the previous one worked.
Before building anything, Daniel Keinrath tested market pull by selling competitors' solutions. Real customers, real conversations, real rejection.
They weren't building in a vacuum — they learned what the market wanted before writing a line of code in that direction. The fonio premise was sold before the product existed. They did not go to investors, accelerators or advisors first. They went to buyers.
No sales team. No playbook. Keinrath on LinkedIn every day talking to SME owners and closing deals himself — demo calls 14 hours a day while the first MVP was still being built. Outreach automated with Dripify to raise conversation volume, every interaction logged in Attio. He was filling in the CRM for two hours after a full ten-hour day of calls.
Two outputs: revenue in the bank, and deep familiarity with customers' language, pain and objections — organised in a CRM rather than in someone's head. That phase can't be outsourced. They didn't try.
This is the step that still stands out. Instead of burning ad spend to acquire customers at a loss, they engineered a loop: sell €300 of prepaid fonio credits, self-serve, no sales call. Prepaid credits covered ad spend. Ad spend brought more customers. More customers bought more credits. The loop closes.
They were spending around €40,000/month on ads and running profitably — before raising a single euro from VCs. They are at over €400,000/month now, run by Benedikt Brauner.
Most startups raise a pre-seed to afford their ad spend. fonio used its ad spend to fund itself into the pre-seed.
Only after the self-serve loop was working did they hire sellers. Steve Gruber took that over, refining the system and building the first sales team at the same time.
Those sellers walked into something rare: a product already converting, data on what worked, an automated pipeline, and referenceable customers on day one. They weren't hired to figure it out. They were hired to scale what was already figured out. The team still runs on that pipeline ("Sales Pipeline 2.0") and has closed 3,000+ clients on it since.
Until February 2026 fonio ran on prepaid credits. Credits made sense to optimise cash flow at the start. Once the GTM motion was set up, contractually committed subscriptions became necessary to properly monetise a growing base.
Subscriptions require product trust and retention confidence. fonio earned the trust first, then monetised it. Growth has averaged over 30% month-on-month since. Lukas Pulling owned making sure every customer could actually experience the product fully — which is what makes a subscription defensible rather than a churn machine.
Most companies hire "country managers" who execute a centrally produced playbook. fonio did the opposite. Each GTM Lead owns their market entirely — website, ads, hiring, local partnerships. They are treated as entrepreneurs, not employees.
They started with France (one of Europe's largest SME markets) and now have offices in New York, Munich, Milan, Paris, London and Warsaw, active across Europe and Brazil, expanding into the US.
With the self-serve motion, sales motion and ads strategy all running, the team started adding channels: SDRs calling people who had run a test call on the website, cold-email infrastructure, and a heavily expanded partner network with a partner programme designed by David Leibovitz.
Every channel was added at the moment the base could carry it — not because it was trending.
| Period | Customers | Added | Note |
|---|---|---|---|
| Q3 2025 | 1,700 | — | Founder-led + self-serve credits |
| Q1 2026 | 4,000 | +2,300 | Includes 450 from Fluently; subscription pivot lands Feb |
| Q2 2026 | 7,000 | +3,000 | First full quarter on subscriptions |
| Aug 2026 | 9,000+ | +2,000 | $10M ARR milestone |
All prices excl. VAT. Published openly — which is itself a segment signal: the two SMB players in this research (fonio, My AskAI) publish prices and use them as a marketing weapon; the enterprise players (Sierra, Ada) publish nothing.
ACV ≈ €1,200–6,000/year. That is a volume business: 9,000 customers at roughly €1,100 average annual revenue. It only works because acquisition is paid-and-profitable rather than sales-led — the exact opposite structure to Sierra's ~$180k–350k first-year contracts.
SMBs, sold horizontally across: hotels, tradespeople, property management, car dealerships, government agencies, auto repair shops. Main market DACH — Germany, Austria, Switzerland. Biggest named clients: TKE, Mahle, YouPower, Holcim, Bolt, Volkswagen, Magnum.
Entrepreneur who had founded both B2B and B2C businesses before fonio. Ran the pre-product selling, then the founder-led sales phase — ~20 demos a day, 14-hour call days, ~1,000 clients and ~500 partners closed personally.
Started as an engineer and moved fast: opened his own business after three years, then became CPO, then co-founded fonio in 2024. Owns the part most AI voice companies never reach — being inside telecom infrastructure rather than on top of it, and GDPR-compliant.
Sigma Squared Society may be a crucial factor in both the start of the business and its partnerships, alongside AustrianStartups (Markus Raunig, Hannah Wundsam). Gruber has publicly credited both as what "enabled me to do what I love on a daily basis now". Worth treating as a live hypothesis to test in conversation rather than an established fact — but if it holds, the replicable version is which community you enter before you need it.
A live embed of ad-library video is not possible. Meta, Google, LinkedIn and TikTok ad
libraries have no public embed API, and all four send frame-blocking headers, so an
<iframe> of them will render blank on this site. Deep links work and are below.
If you want the creatives physically on the page, the workable route is to capture them — download or screen-record the ads once, then host the files here as assets we control. That's a manual capture step, roughly an hour for a set of 8–12 creatives, and gives us stable assets that don't break when fonio rotates its ads. Say the word and I'll do that pass.
Not a guess — fonio's channel mix tells us where to look. Paid social is 5.99% of all traffic (the highest in the peer set) and their social traffic is 68.0% Facebook, 22.0% LinkedIn, 6.0% YouTube, 4.0% Instagram. Paid search is 12.84% and display 5.34%. So:
Ad-library links are live queries — results change as fonio rotates creative, and an empty result usually means the country filter, not the absence of ads. Meta's library only guarantees full historical archives for political/social ads; commercial ads show currently-active only, which is another argument for capturing what we see.
fonio's paid search is buying competitor and infrastructure brand terms, not category
terms: twilio (6.3%), retell (5.8%), vapi (5.2%),
retell ai (4.9%), placetel (3.1%). They are intercepting people who were about
to build it themselves on Twilio/Vapi/Retell, not people shopping for an AI receptionist. If the
creative matches that intent, it's the single most interesting thing in their ad account — see
the conquest analysis.
Selling a €300 credit pack self-serve made the ad budget self-funding. Any packaging where cash arrives before delivery turns CAC payback from a constraint into a non-issue.
Bidding on twilio and vapi targets people mid-decision on building, where the
alternative is six engineer-months. Highest-intent, lowest-competition traffic in the category.
SDRs ring people who already ran a test call on the site. The product manufactures the intent signal; the SDR just follows it. Cheap, and it needs no data purchase.
Credits → subscription only once retention justified it. The 30%+ MoM after Feb 2026 is the payoff for waiting. Worth a guardrail metric before any pricing change we make.
A GTM Lead owning site, ads, hiring and partnerships per market produced +30–66% growth per country. Localisation without the ad account is just translation.
Reselling a competitor's product is the cheapest possible demand test — and it's the one habit shared by four of the five companies in this research.