The uncertainty principle of Heisenberg is usually written as dp.dx~hbar and it says that the uncertainty in momentum of a quantum object (dp) times its uncertainty in position (dx) is always a constant (hbar). If a quantum object knows well where it is (dx=small), then it loses the ability to know its speed (dp=big). Conversely, if it knows its speed very well (dp=small), it'll be lost in space (dx=big). This relation from quantum mechanics, and special relativity also, are two clues that physics is due to be reworked around the concept of information. This is what quantised inertia does, joining these two pillars of physics (QM and relativity) on the large scale.

Imagine a red mass (see diagram, top part, red circle). Suddenly you put another mass on the left of it (the black circle). The uncertainty of position of the red mass is shown by the black quadrilateral around it. The red mass can see a large amount of empty space up, down and rightwards (forgetting directions perpendicular to the page for now) so its uncertainty in position (dx) is large in those directions because it cannot position itself well in empty space. However, it can see less far into space to the left because the other mass blocks its view, so its uncertainty of position that way (dx) is lower. The quadrilateral represents dx in each direction. It is skewed outwards to the up, down and right where dx is large, and skewed in to the left where dx is small. Therefore, according to Heisenberg, the quadrilateral showing the uncertainty in momentum has to be the opposite: skewed out to the left and skewed in for the other directions (see the blue envelope). Since momentum involves speed, this predicts that it is statistically or quantum mechanically more likely that the object will move to the left. In a formal derivation I have shown this not only looks like gravity but predicts it (see reference below).

Now, as the red object approaches the black one (see lower panel) its uncertainty in position (dx) to the left gets ever smaller, so dp must increase and the red object must accelerate. "Aha!" Says the other great fundamental pillar of physics: relativity, "I now become relevant!". Since the red object is now accelerating away from the space to the right, information from far to the right cannot get to old Red, and a horizon forms (the black line) beyond which is unknowable space for Red. This Rindler horizon is like the black mass. It blocks Red's view and so Red's uncertainty in position to the right reduces (dx, see the black quadrilateral contract from the right) and so the uncertainty in momentum to the right increases (see the blue quadrilateral now extends further to the right). Red now has a chance of moving both left and right and this has the effect of cancelling some of its initial acceleration towards the black mass. This looks like inertia, and indeed it predicts quantised inertia (see reference below).

In this way, you can derive something that looks like quantised inertia (if you consider also the cosmic horizon) and gravity, just by allowing quantum mechanics and relativity to mix at large scales. The whole package could be called horizon mechanics. The word 'horizon' from relativity, the 'mechanics' from the quantum side. As a happy side effect, quantised inertia or horizon mechanics solves a lot of problems in physics that you may have heard of: it explains cosmic acceleration, predicts galaxy rotation without dark matter, and its redshift dependence, and predicts the emdrive. These successes should not be sneezed at, representing 96% of the cosmos, and with the emdrive practically offering a new kind of propulsion. Oddly enough, for a theory intended to replace general relativity, the behaviour I have just described looks quite tensor-ish..

References

McCulloch, M.E., 2016. Quantised inertia from relativity & the uncertainty principle, EPL, 115, 69001. ResearchGate preprint, arXiv preprint