replient.ai
DEEN

Case study · International expansion

fonio.ai

fonio.ai: 27 languages. Two people in marketing. No translation tool.

fonio.ai advertises in nine markets at once. The comments underneath arrive in German, English, Polish, French, Italian – and 22 more languages. This case study shows how a two-person team keeps that under control.

27languages in the inbox
×30ad volume in four months
96.5%of disruptive comments caught automatically
6.4working days back in 69 days
  • Markets

    AT, DE, CH, FR, IT, PL, BR, UK, US

  • Channels

    Facebook, Instagram, LinkedIn, TikTok

  • Team at fonio

    2 people, no moderation team

  • Effort in the tool

    1 automation

All figures from the replient.ai production database, 2 June – 9 August 2026 against the 69 days before. Customer-side metrics are deliberately reported in relative terms only. Methodology at the end of the page.

About the brand

Austria’s fastest-growing startup

fonio.ai builds the AI phone assistant for businesses: an agent that answers calls, books appointments and routes requests on. Founded in Vienna in 2024, the platform now counts more than 10,000 customers – from diagnostic centres and hospital groups to Volkswagen’s used-vehicle business.

It sells mainly through paid social: short, fast reels, pushed broadly. Every new market means new ads, new audiences – and a comment section in one more language.

Company figures: fonio.ai’s own published figures on fonio.ai (as of August 2026).

  • 2024

    founded in Vienna

  • 9

    markets

  • #1

    software tool of the year 2026

  • 10,000+

    customers

These posts are loaded from instagram.com and set cookies. Please accept marketing cookies to display them here.

View fonio.ai on Instagram

Recent reels from @fonio.ai – the material the comments land under.

Every live ad in the Meta Ad Library

Try the AI yourself

On fonio.ai you can call the AI phone assistant for free and test it live in conversation – no sign-up needed.

Open fonio.ai

Chapter 1

Polish is the third-largest language in the inbox

In March, comments arrived in two languages. By July it was 22. Across the whole window, 27 languages came in – 73.5% of all comments are not German.

The distribution maps the expansion more precisely than any org chart: Polish sits third with 20.5%, ahead of French and Italian – exactly the markets fonio.ai is moving into.

And this is where international scaling gets tedious. Not because nobody could translate, but because every single comment would have to be copied, translated, judged and then decided on – hundreds of times a day, in languages nobody on the team speaks.

Comments by language
  • German
    26.5 %
  • English
    23.9 %
  • Polish
    20.5 %
  • French
    12.1 %
  • Italian
    7.8 %
  • Dutch
    3.7 %
  • Spanish
    2.5 %
  • Other (20)
    3 %

"Other" bundles 20 languages below 1% each – including Portuguese, Arabic, Romanian, Turkish, Russian, Ukrainian and Georgian.

Languages per month

2

March 2026

22

July 2026

detected in the inbox

2 secis all replient.ai needs to detect a comment’s language, sentiment and topic – no matter which language it was written in

Chapter 2

Every new market brings its own trolls

You can only enter a market if you understand the comment section underneath. At fonio.ai, 91% of all comments sit under a paid ad – and there the share of critical comments runs about eight times higher than under organic posts.

That is not a fonio problem, it is the mechanics of paid social: ads reach people who neither know the brand nor went looking for it. With an AI product, part of the audience writes against the technology on principle.

So exactly one rule runs in replient.ai: when the system detects a clearly brand-damaging comment, it is hidden under the ad – after a 20-minute grace period. Hidden is not deleted: on Meta the comment stays visible to the person who wrote it. Criticism and real feedback stay in the inbox; only what helps nobody is gone.

Share of critical comments
under paid ads58 %
under organic posts7 %

Hide disruptive comments automatically

active since 11 Jun 2026

active

When

Comment is brand-damaging

Then

Hide under the ad

Grace period 20 minutes

79%of all moderation actions run through this one rule

The same comment section, once without and once with the rule

On the left the ad runs unmoderated: trolls sit between an honest question and real customer praise. On the right the rule has fired – question and praise are still there.

without the rulehow the ad gets delivered
Comments
not.a.fan.here2h

More AI junk. Soon nobody will talk to a human at all.

Reply
41
practice_dr_seibert2h

Does this also work when patients call in Polish?

Reply
6
justsaying_20111h

🤡🤡🤡 total scam

Reply
3
brunner.painting48m

Been using it for three months. No missed calls any more, genuinely strong.

Reply
12
with the ruleafter the 20-minute grace period
Comments
not.a.fan.here2h

More AI junk. Soon nobody will talk to a human at all.

Reply
41
Hidden
practice_dr_seibert2h

Does this also work when patients call in Polish?

Reply
6
justsaying_20111h

🤡🤡🤡 total scam

Reply
3
Hidden
brunner.painting48m

Been using it for three months. No missed calls any more, genuinely strong.

Reply
12

Illustrative example. The comments shown are written for this page and do not come from fonio.ai’s channels – we never publish real comments.

Chapter 3

Nine markets do not keep office hours

63.7% of all comments arrive outside office hours – evenings, nights or weekends. Across nine time zones there is simply no hour left where everything lines up.

And that is exactly the part software does better than a rota: the ads keep running, and the space underneath keeps being looked after – at 11:40 pm, on a Sunday, in Polish.

When comments arrive
63.7 %off-hours
  • outside 8am–6pm / weekend63.7 %
  • within office hours36.3 %

Office hours = Mon–Fri, 8am–6pm (Europe/Vienna).

The result

A new market no longer costs a new hire

When a human hides a comment, the median gap between comment and click is 2 hours 46 minutes. The automation takes 20.4 minutes – and 20 of those are a self-imposed grace period. That is 88% less reaction time.

  • 6.4

    working days

    back in 69 days

  • ~45

    minutes

    saved, every day

  • –88%

    reaction time

    versus doing it by hand

  • 0

    new hires

    needed for moderation

Time saved uses the product formula: 40 seconds per automated action, 30 per reply, 20 per hide or like – the same number fonio.ai sees in its own dashboard.

In their words

Benedikt Brauner runs performance marketing at fonio.ai. In a LinkedIn post about his AI toolkit, he adds as a postscript:

We’re also currently testing replient.ai to help manage all the comments we receive across different channels and countries. A huge recommendation for international marketing teams!
Benedikt BraunerBenedikt BraunerPerformance Marketing Manager, fonio.ai

This LinkedIn post is loaded from linkedin.com and sets cookies. Please accept marketing cookies to display it here.

View the post on LinkedIn

What a clean comment section does for ad performance we measured in an A/B test with two identical campaigns: +48% ROAS, +54% conversion rate. Those figures come from the Zauberfein case study, not from the fonio.ai data.

Read the Zauberfein case study

Postscript

Congratulations: fonio.ai passes $10M ARR

While we were finishing this case study, founder Daniel Keinrath announced the next milestone: from $500k to $10M ARR in twelve months, seven new markets entered, the team grown from 5 to 60 people. Next goal: $100M ARR by the end of 2027.

For this case study that means one thing above all: nine markets will become more. And every new market brings one more language into the comment section.

Daniel Keinrath · CEO & Co-Founder, fonio.ai

This LinkedIn post is loaded from linkedin.com and sets cookies. Please accept marketing cookies to display it here.

View the announcement on LinkedIn

Methodology

Where the numbers come from

  • The basis is every customer comment on fonio.ai’s channels in the replient.ai production database. Two windows of identical length are compared: the 69 days from go-live on 2 June 2026 against the 69 days before, both cut at midnight Vienna time.
  • For the before window, channels and ad accounts were fully re-crawled so that volume is genuinely comparable.
  • Customer-side metrics are reported in relative terms only – as a share or a factor. We do not publish absolute comment, ad or campaign counts for fonio.ai.
  • Reaction times come from the replient.ai activity log, measured against the comment’s own timestamp. Medians throughout.
  • Deliberately not shown: a before/after comparison of sentiment or reply rate. Those fields only exist for comments that went through replient.ai and would not be sound for the before window.
🚀

Ready to save time?

Start your 14-day free trial and save up to 80% of the time spent on comment management!

Start now. It's free.

no credit card required · 100% risk-free