About

Three people,
and a factory

We built the system that makes games before we made one. We own the game; the casino owns the player, the wallet and the licence.

Talk to us The games

People: 3
Games built: 8
Operators live: 0
Laboratory certificates: 0
Player records held: 0

3 people
1 on strategy, 2 on the build, no layer between
1 factory
one generator, one engine, one cabinet, one evidence pipeline
8 games
built, 4 of them playable with no account and 4 measured against their own evidence

Getaway Gaming is 3 people. One decides what gets built and on what terms, and two build it. There is no layer in between, so a decision here takes a conversation, not a quarter. A studio this size usually has one game. We have 8, and 4 of them are measured against their own evidence, because the first thing we built was the system that makes games. None of them carries a laboratory certificate yet, and the register says so.

We started by asking what it would take for a new title to cost almost nothing to make. The answer was a shared toolkit: one generator, one measurement engine, one cabinet, one motion layer, one evidence pipeline, and a vocabulary of mechanics that games are assembled from. Every hour spent on one title goes into the layer the next one inherits. The library is the asset and the games are the dividend.

Getaway Gaming is a game provider. We build slot games and license them to casino operators, who embed them on their own sites. We are not a casino: we hold no player accounts, take no deposits and pay out no withdrawals, and we have no gambling licence of our own to offer anybody a game under.

That boundary is built into the architecture. The only player identifier that reaches our systems is an opaque reference the operator chooses. There is no name, no email, no address, no date of birth and no IP address anywhere in our schema, and adding one would take a legal decision.

What we own is the half nobody sees: the game logic, the random number generator, the maths, round settlement, the licensing chain that decides whether a given player in a given market may open a given build, and the runtime the game draws itself in.

Our games reach players through somebody else's server. They are carried by an established remote gaming server, and we do not offer a direct integration. That is the fastest route to a player, because operators already run against it, and a new studio needs shelf space before anything else. What we built runs behind our games on that server.

Why the next game costs less than the last

These are not reskins. Eight different mechanics come out of one toolkit. A new title inherits the maths vocabulary, the cabinet, the motion layer and the evidence pipeline, and each of those has a check that fails when a game drifts away from it.

Inherited · 1

The maths, as a vocabulary

30 mechanic families with a declared method for pricing each one, a calibration engine that solves a game's free parameters against a published return, and a second engine that re-measures the answer. A new game declares which families it is made of and does not re-derive what a cluster, a respin or a collect is worth.

Held by: each game declares its families, and the declaration is checked

Inherited · 2

The cabinet, whole

One shell every game mounts: the reel law, the cascade law, the fitting rules for a desktop, a phone upright and a phone turned, the settings dialog, the loading screen, the opening screen, the share card and the demo URL. The next game inherits all of it from one registration.

Held by: a route missing any of the six is a failing audit that names the file and the fix

Inherited · 3

Motion, time and money on screen

Curve roles that each skin publishes in place of hand-written easing, one celebration clock that times a win's announcement by the money won, one definition of what the board, the flight and the total may say, and one light kit under every effect in the portfolio.

Held by: no stylesheet may spell out a bezier, a duration or a rebound; the audit fails on both halves

Inherited · 4

The evidence, produced by the run

Certification, conformance against the published specification, the RNG battery, the PAR sheet, the replay, the market packs and the gates that stop any of it shipping stale. A title arrives with its dossier because the dossier is what the measurement writes.

Held by: a published figure that drifts from the maths that produced it fails the merge gate

Where this goes: 1,000 titles live, with a build team under 10.

That is the bet. The industry’s unit economics are headcount multiplied by months per title, so a provider with two to three hundred people ships between six and twenty titles a year and cannot ship the twenty-first at any price. Our answer is to make a title an assembly job instead of a project. Then the size of the team no longer sets the limit.

What is true today, and checkable on this site: 8 games built, 4 of them playable with no account, 4 measured against their own evidence with a complete pack per rung, 30 mechanic families in the vocabulary, and 63,276 lines of shared code that no future title pays for again. Against that: no operator live, no laboratory certificate in any environment, and no revenue.

If you are the one buying

The same arithmetic, read from the other side of the table. There is no percentage and no rate card here: with no live operator and no signed term sheet, any discount figure would be invented.

For the operator · 1

No headcount floor under the price

In this industry a game costs people multiplied by months, and every rate card inherits that floor. Ours does not carry it. The terms, the exclusivity window and the size of market worth serving are not tied to it the way they are for a studio with three hundred people making the same game.

For the operator · 2

The long tail becomes economic

A mechanic a large slate cannot justify, a market too small to be worth a certification, a concept for one operator's audience: each of those is a conversation here, not a budget round. What matters is whether anybody wants to play it.

For the operator · 3

And it arrives with its paperwork

Every game carries a complete evidence pack per rung of its RTP ladder, a spec sheet drawn from that pack, a market pack per jurisdiction and a round that replays from its own seed. We do not add any of it at the end. The build produces it, so producing it again costs nothing.

What we are not

Three of these are boundaries an operator's legal team checks for. The fourth is the certificate we do not hold. Under each one is what is there instead.

Not~~A casino~~
We hold no player accounts, take no deposits and pay out no withdrawals. The operator owns the player, the wallet and the licence the game is offered under; we own the game and the server it resolves on.
we hold no gambling licence of our own
Not~~A holder of player data~~
The only player identifier that reaches us is an opaque reference the operator chooses. There is no name, no email, no address, no date of birth and no IP address anywhere in the schema. Adding one would be a legal decision, not an engineering one.
invariant 2 · player_ref, and nothing else
Not~~A branch per operator~~
Nobody gets their own fork. A different bet ladder or a different skin is a row of configuration, an old build is a pinned version, and a real change to the maths is a variant with its own certification and its own price.
three tiers, one codebase
Not~~Certified~~
No laboratory has tested any build of anything here, in any environment, and the server refuses a real-money launch without a certificate, as it should. Every figure we publish is our own measurement, written by the run that produced it.
the licence register has the detail

Where we are

Counted from the repository, like every figure on this site.

Games: 4
Launch markets: 2
Checks that can stop a certification: 20
Known failures on the register: 20
Open defects in the maths log: 9
Schema migrations: 125

Two of those six count things that are still wrong. Read the full register

How a game gets made here

Four layers. Which one a decision belongs to depends on one question: how many games can it be wrong for at once? The stack below is read from the repository, and these are the layers the build is held to.

4
LOOKart, palette, sound, copy and motion curves, with no mathsCan be wrong for one skin · costs no certification
3
GAMEwhich families, which strips, which paytable, which leversCan be wrong for one game · costs a certification pack per rung
2
MECHANICSone module per mechanic family: how a mechanic is drawn, evaluated and pricedCan be wrong for every game using the family · costs the family, then a game
1
MOTORRNG, the round saga, measurement, calibration, the gatesCan be wrong for the whole portfolio, at once · costs re-certification of everything

A rule that can be wrong for every game belongs at the bottom, where one fix repairs the portfolio and one check can see it. We learned that from a routine that existed in five copies at three different strengths. One of them performed no check at all, and nothing noticed, because three correct copies and one missing copy agree perfectly.

A new look for an existing game costs no certification. A new return costs one evidence pack per rung. A new game costs a full pack per rung. A new mechanic family is the expensive one, and it is paid for once.

Every game, in its own colour

Each cabinet declares its background colour and the ink that is legible on it as a pair, and every pair is checked by an audit. The strip below is drawn from the registry and holds the games that have a spec sheet.

Farmyard Heist
Ultimate Coin Miner
Horizon Rescue
Sleep Tight

Who does what

Three seats and two roles outside them, described by the work they cover. One of them is a seat this studio cannot fill from inside, however small it stays.

Covered
Strategy: one seatWhich titles get built and in what order, which markets are worth a pack, what a licence is worth and what it costs, and where shared-code effort goes next. It is one person, on purpose. Our economics depend on a decision costing a conversation, and a layer of management between the decision and the build would be the first thing to take that away.
Covered
The build: two seatsThe generator, the paytables, the calibration engine, the RGS, the licensing chain, the cabinet, the motion layer and every game built on them. One codebase, four layers, and a rule about which layer a decision belongs to. That rule is what lets two people carry a portfolio and its evidence at the same time.
Covered
Sound: a collaborator, not a seatEach game's sound pack is written by someone outside these three, against the cue sheet the cabinet publishes, so the audio follows the round's own events instead of a timeline matched by ear. We do not count a collaborator as headcount.
Open
A clean-room author for the second engineThe independent implementation that re-measures a game must be written by somebody who has not read the first one. Nothing we have built makes that job cheaper, because its value comes from not having seen the code. It is why one of our games has no second opinion today: patching the existing adapter would have destroyed its independence and still let it report agreement. This needs a hire.

How we work

The lock

The maths cannot move unnoticed

Each game’s maths is hashed into a lock file, and every certified state is kept as an append-only snapshot. An edit without a deliberate re-lock fails the build.

A re-lock without a stated reason refuses to run.

Two samples

Two measurements are two samples

The tool that tunes a game and the tool that certifies it draw independent samples on purpose. Two scripts running the same arithmetic are not two checks. That is how a mode once shipped returning 119.6% while both scripts agreed it returned 96.1%.

Report the certification run, never the calibration run.

The register

What is wrong is written down

Every known failure is a row with a defect reference, an owner, what closes it and a date the excuse expires. An expired row fails the merge gate. So does a stale row that no longer matches a failing check.

Kept in the known-failure register.

If that sounds like what you need

Tell us which markets you hold

We will tell you which of our games can be served there today, which rung of the ladder the market’s own rules allow, and a realistic date. That includes the date for the certificate, which does not exist yet.

Email us Where the games can go How to check any of this

hello@getawaygaming.com · Operators and aggregators can also use the enquiry form on the contact page.

01Why the next game costs less than the last