I'm still betting that the first "assembler" will be an ultra-high vacuum chamber with a bunch of lasers and well-controlled electromagnetic fields.

You will not get better than few-nm resolution, and you will get basically incoherent deposition. If the guiding field is optical, the resolution will not be a factor of >1000 finer than the wavelength, which is what you would need in order to achieve atomic specificity. In the case of guidance by wires or IC traces as described in the news item, the resolution will always be poorer than that of the guiding structures. Perhaps a three-dimensional focusing/aperture structure fabricated to atomic specifications (e.g. by an assembler) might be able to deliver atoms with angstrom precision. However, if it were not cold, the atoms would not be, either. In that case, it would be hard to get them to stick in just the right places with 100% yield, and you would need some way of probing to determine when a good bond had been made. This doesn't sound like a way to get around the difficulties of making a first assembler.

"atomic ink-jet printer"

Throughput? Resolution? This might be another useful tool for the compleat nanotech lab, but it's not going to be a way of making large-scale products.

"atom-coupled device."

Probably intended for quantum computing; however, schemes of this sort have been plagued by heating and decoherence due to fluctuating fields in the electrodes.

Once folks start moving Bose-Einstein condensates this way, all kinds of exciting advanced fabrication techniques may become possible.

For the same reason (heating by the electrodes), it is probably not possible to move BECs around "a few microns above a microstructured surface" as described. BECs are formed and manipulated in optical traps, so again I don't see how you will get angstrom resolution.