01 / 12
Business Case

Atlas Onsite

One engine. Every brand, every market, every match.
Why we should scale AI-built betting content across every brand and market — seen from the CEO's and CFO's chair.
Structured on the Sequoia business-plan framework
CONFIDENTIAL
What we are
02 / 12
✦Company Purpose
One live feed in.
A complete betting experience out.
Atlas Onsite takes one live data feed and turns it into a full betting experience — articles, odds, insights and offers — in the right brand and the right language, for any league we point it at.
Build once
Serve everyone
Update everywhere
Why it hurts
03 / 12
✦Problem
Teams already use AI.
The process is still slow and manual.
Today's workflow Request→ Produce→ Hand off→ Publish→ Report — every step a request, a wait, a handover. Repeated per brand.
01✕

Everyone waits on everyone

Each step waits on the one before, across several teams. A new league or market still takes weeks.

02✕

Same for everyone

One size fits all — no personalisation, nothing in the moment. Every brand runs the same five insights.

03✕

We pay for the same work twice

Every brand and market funds its own version of the exact same job.

04✕

Hooked on bonuses

Without relevant content in the bettor's own language, we fall back on expensive welcome bonuses to win and keep players.

What we built
04 / 12
✦Solution
One standard for everything.
A new brand is a setting, not a project.
01✦

Data → everywhere

Sportradar + Kambi + Atlas mining in; on-site, video, radio, social, DOOH, push, CRM out — plus a real-time Recommendations API.

02✦

Brand is just the wrapper

Colours, assets and tone of voice differ. Everything underneath is shared.

03✦

Compliance is built in

Only our own odds, banned phrases blocked automatically, and every market gets its own rules and language.

04✦

Fix it once

Improve the engine once and every brand, market and language has it the same day.

The timing
05 / 12
✦Why Now
The technology got good enough in 2024–25.
01

AI got cheap and reliable

Text, speech and avatar video are now good enough — and cheap enough — to run real betting content at scale.

02

The window is tiny

The hour before kickoff plus in-play carry >70% of activity — a few hours per match, then it’s gone. Miss the moment, miss the bet.

03

Regulators favour control

Content that is governed, auditable and correct in every market turns compliance from a risk into an edge.

04

Margins are under pressure

Content cost and speed-to-market are now a competitive battleground for operators.

The prize
06 / 12
✦What changes for us
From a cost centre to a multiplier.
Time to market
Weeks and months
→
Seconds and minutes
Cost
A team + agency per market
→
Build once. Adding one is almost free
Headcount
A squad per brand
→
One small team runs them all
Output
What the team can make by hand
→
Unlimited — every match, market & language
Control & risk
Scattered, hard to check
→
One standard, compliant everywhere
The alternatives
07 / 12
✦Competition
Everyone has a piece. No one has the whole thing.

In-house, per brand

Slow and expensive, and rebuilt market by market. No personalisation, nothing shared.

Agencies and standard CMS

Not live, not connected to the data, and a human touches every single asset.

AI writing tools

They write words. No data, no odds, no compliance and no channels behind them.

✦Atlas's moat
We've wired the data, the odds and the content together under one set of rules, we serve every channel from one source, and we answer in real time. Nobody else has all of it in one place.
How it works
08 / 12
✦Product
Channels ask. Atlas answers — in real time.
→Request
A channel asks for the ideal content — and can name the product to push.
brand × market × surface × audience × moment × product
→
◆ Atlas core
  • Look updata · odds · mined gems
  • Decidepick the best · our odds only · right market, right language
  • Producetext · speech · images · avatar video · ready in advance or made on the spot
  • Deliverchecked · in the local language
→
Delivered →
The ideal, compliant, localized content lands on the requesting surface — instantly.
↩What worked flows back and sharpens the next request
Fed by Sportradar · Opta · Kambi (OBG) · KSP (OBG) · Atlas mining · DAM · competitor-odds & trend intelligence · compliance
Delivered to Product Pages · Onsite Reels · Social · Display · Online Video · DOOH · TVC/Live · Radio & Spotify · Push · CRM · Affiliate | × markets × brands · real-time
The economics
09 / 12
✦What it costs us
One cost base for everything —
and it gets cheaper per output as we grow.
01
One bill, not many — a small team and some servers run every brand, market and channel. We don't pay again for each new output.
02
Almost free to add one — a new brand, market, league or channel is just settings. No new team, no rebuild. A few cents of AI per match.
03
Cheaper as we grow — the more outputs on the engine, the lower the cost of each. Scale lowers unit cost — the opposite of normal content.

"The more we grow, the less each output costs."

Operating leverage
10 / 12
✦Team
3 → 100
sites without a bigger team. One small team runs the whole portfolio, because the shared standard does the heavy lifting.
01

One platform team

Owns the standard and the rules that let a single change roll out safely to every brand and market.

02

The engine does the squad work

Writing, compliance checks and translation run automatically. No editorial, design or dev team per brand.

03

Headcount stays flat

3 sites or 100 — same team. That's the whole point.

The model
11 / 12
✦Financials
Cost to serve is a rounding error.
The prize is the handle.
€1–3
all-in AI + compute per match
€1–2k
servers per year — shared across all tenants
<0.5%
turnover lift funds the whole platform*
01 — Cut the content bill

One engine instead of many teams

The editorial and agency spend in every market, folded into one engine.

02 — Speed

Days, not weeks

Launch a new league or market while the window is still open.

03 — Turnover uplift  ·  the big lever

Where it pays back

Break-even = platform cost ÷ (handle × GGR margin) — a fraction of one percent of turnover. Everything above that is profit. Finance has the real inputs.

*Break-even uplift, not a forecast — we don't invent revenue. Put in the group's real handle and GGR margin and the formula gives the exact figure. The content and headcount savings come on top.
The ask
12 / 12
✦The Ask
Turn a prototype into a proven business.
1

Measure what it drives

Track what each piece of content actually does for turnover. Prove the uplift — and give the engine something to learn from.

2

A legal owner for the ad rules

Each market's advertising rules written into the system, checkable and auditable. That's the gate that lets us scale safely.

Then scale from 3 to 100 — on one standard, one team, one update at a time.