Imagine you run a mail room where every package must be sorted into two chutes — red and blue — but the sorter machine only works if the packages arrive on two separate conveyor belts. Ferromagnets give you one belt (easy to sort, but the belt's own magnetic field shoves nearby packages off course). Ordinary antiferromagnets cancel the interfering field but dump both colors onto one belt, so you can't sort. What you want is two belts, zero stray field, and the sorter working at room temperature. That's what Luttinger-compensated semiconductors promise for spintronics: zero net magnetism, a usable band gap, and spin-sorted band edges wide enough to shrug off thermal noise. The committed claim here: an AI-agent-assisted computational screening has identified two candidate Luttinger-compensated semiconductors with spin windows far exceeding thermal energy at 300 K. One is a newly designed oxide (YBaMnFeO₅) that may be unsynthesizable in its useful form. The other — and this is the real find — is KV[Cr(CN)₆], a Prussian blue analogue first synthesized in 1999, whose spin-sorting properties were apparently hiding in plain sight in a 2008 hybrid-functional study that never remarked on them. The ladder matters. A 2025 study predicted two other Luttinger-compensated semiconductors, but both lost magnetic order below room temperature. KV[Cr(CN)₆] was measured at 376 K ordering temperature in 1999 — above the 300 K threshold that the field has named as the open goal. The spin windows predicted here (2.6 eV for holes, 1.6 eV for electrons against a ~26 meV thermal floor) are enormous if the HSE06 calculations hold. But 'if' is doing heavy lifting: neither the band gap nor the spin sorting has been experimentally measured yet. Architecturally, this is density functional theory at two rungs: PBE+U for fast screening and HSE06 (hybrid functional) for the headline numbers. The AI agents handled candidate generation and simulation orchestration, but the physics is standard DFT — no novel computational method is claimed. The interesting architectural choice is using agent-driven search to trawl existing materials databases, which is how a 1999 compound got flagged as a candidate nobody had recognized. The integrity picture is mixed in the way computational materials science often is. The HSE06 numbers are the more trustworthy tier, and the authors are admirably transparent that PBE+U and HSE06 disagree on whether hydrated KV[Cr(CN)₆] retains its spin sorting. The 376 K ordering temperature comes from a 1999 experimental measurement — that's real, independent data, not a simulation artifact. But no one has measured band gap or spin polarization experimentally. The YBaMnFeO₅ candidate's own thermodynamic instability (checkerboard ordering collapses around 950 K, below typical synthesis temperatures) is honestly disclosed, which is a good sign. The milestone is concrete: someone needs to grow a dry, crystalline KV[Cr(CN)₆] sample and measure its band gap and spin polarization with spin-resolved photoemission or tunneling spectroscopy. The 1999 sample was a hydrated powder with a residual moment of 0.125 μB per formula unit. A clean single crystal with measured zero net moment and a confirmed ~2.1 eV gap would validate the entire Luttinger-compensated semiconductor concept at room temperature — not just this material, but the design principle. The obvious experiment not run: actual synthesis and characterization. This is computational prediction, not experimental confirmation. The honest read is (a) — the authors are theorists and AI-tooling people, not synthetic chemists. Growing Prussian blue analogues as dry single crystals is a known but nontrivial materials chemistry problem. The paper is essentially a signal flare to experimentalists: this compound exists, it's been made before, and our calculations say it does something nobody noticed.
