I have been studying the use of impactors in retrograde solar orbit, trying to avoid ISRU. But actually, kinetic impact powered rockets are a good match for ISRU since unrefined substances can be used for impactor mass.
With that in mind, an ideal ISRU target is 2009 HC82. This asteroid is already in a retrograde orbit, with an extremely high relative velocity to Earth.
Impactor drones from 2009 HC82 only need 2.2km/s to reach Earth with a relative velocity of 67km/s. Impact velocities of 67km/s are enough for 1 ton of impactors to boost 10 tons of payload to C3. If the impactor mass is 99% ISRU rock and 1% guidance hardware, then only 1 ton of guidance hardware is needed to boost 1000 tons of payload to C3. This 1000x multiplier factor is vastly superior to anything I was getting without ISRU.
The bottom line is that a very small mass investment of drones can be used to boost large amounts of mass to Earth orbit or Earth escape.
Each drone might do its own mining using a pulsed electron beam. The main structure of the drone is a lightweight wire mesh dish which lands rim-down. The electron beam shoots straight down the centerline, sputtering material off the asteroid. Some of this refreezes onto the mesh, eventually closing the gaps and then gradually building
up a thick layer.
After mining is complete, the drone kicks off the asteroid with a small solid rocket charge. The electron beam is reaimed at the thick bowl of material, vaporizing small bits of material as needed to perform course corrections.
The drones home in on a virtual corridor based on GPS style navigation signals from Earth. This puts them in a linear formation, suitable for powering a kinetic impact powered rocket.
The client vehicle is a suborbital spacecraft which is also homing in on that corridor. Its main thruster is simply a large bowl shaped pusher nozzle. The spacecraft sprays a fine mist of water into this volume, where the impactor drones arrive. The extreme kinetic energies of the 60+km/s collison mean that both the impactor and the water spray are vaporized into hot plasma. The expanding plasma is directed rearward by the bowl shaped nozzle for thrust. This combines specific impulses of 10,000s with high thrust.
Of course, 2009 HC82 is a limited resource. It only has a mass on the order of a few billion tons. As such, it would "only" be suitable for boosting on the order of ten billion tons of payload to Earth escape. Hopefully, by the time this limited resource runs out, we will have charted out other retrograde NEAs or moved on to a more permanent and flexible launch system (such as microwave thermal SSTO RLVs powered by massive SBSP arrays).