July 2026 DD - Sourcing Sweep

Fort is the strength-training wearable the fitness-tracker industry never built - a screenless band from three ex-Tesla engineers that uses automotive-grade motion sensing to auto-log reps, sets, form and recovery with no phone tapping. Backed by Y Combinator (W2026), Patron and angels from OpenAI and Tesla, it's in pre-order with real early press (WIRED, Fitt Insider) ahead of a Q3 2026 ship. Whoop and Oura ignored lifters; this team is applying Cybercab-grade sensor rigor to a large, underserved category.

Shadowfax is rebuilding analytics for the AI era - an auditable, reproducible platform where business analysts orchestrate AI agents to go from raw data to shipped apps without touching Python. The founders are Palantir, Snowflake and Rubrik veterans who've scaled data products past $1B ARR, and they're backed by Khosla Ventures and the Snowflake accelerator. As enterprises struggle to trust AI-generated analysis, a transparency-first, analyst-in-the-loop platform from this pedigree is a credible wedge into a massive market.

Avala is physical-AI training-data infrastructure built by Emal Alwis, the third engineer on Tesla's Autopilot team - a genuinely elite founder-market fit for the embodied-AI data problem. On $4.7M seed it claims real scale (>1B labels/day) across global training centers, though the dataset's 150 headcount appears to count that distributed labeling workforce, not ~31 core FTEs. The category (data labeling) is crowded and margin-thin. Keep for the founder and the physical-AI positioning, but reconcile the headcount and interrogate defensibility.

MP Space applies deep ex-Tesla battery and energy engineering to satellite power systems, shipping integrated Li-ion packs and full EPS for LEO constellations. The 13-hit Tesla pedigree is real and unusually deep for pre-seed, and $2.85M from MFV/Speedinvest keeps it well under cap. The risk is a semi-commoditized hardware category and a thin balance sheet for a manufacturing ambition. Keep as a credible picks-and-shovels space play with elite battery talent.

Atomic is a Tesla-supply-chain-alumni startup (Michael Rossiter, Neal Suidan) building agentic AI planning software, born from the Model 3 spreadsheet chaos. It has a confirmed $3M seed (DVx / Madrona, under cap), a live product, and early customers citing 20-50% inventory-cost cuts. Strong founder-market fit in a crowded incumbent-heavy market. Keep - watch for the next raise Axios flagged and whether it can out-execute o9/Kinaxis.

Astrolab builds the rovers that will carry Artemis astronauts and cargo across the lunar surface - full-scale, tested FLEX/FLIP vehicles now under a NASA Lunar Terrain Vehicle contract worth up to $4.6B. Founder Jaret Matthews (ex-JPL, ex-SpaceX) has assembled a 150-person, SpaceX-heavy hardware team that has already put operational rovers on the ground. It's less a stealth seed bet than a de-risked, government-anchored moon-mobility leader - belongs on a serious space list, if slightly outside a pure early-stage cap.

Chestnut Robotics (formerly TetherIA) builds human-like robotic hands - its open-source, $314, 16-joint Aero Hand won real attention as a cheap alternative to proprietary dexterous hands. The founding team is precisely right: Tesla Optimus mechanical plus a Waymo foundation-model researcher. The open-source, ultra-low-cost strategy makes monetization the central question, and there's no disclosed funding yet. Keep for the on-point pedigree and momentum; press hard on the business model before conviction.

Argo Space is building water-propelled, refuellable orbital tugs and a lunar-water propellant supply chain for in-space logistics - an ambitious ISRU bet from a deep, verified SpaceX propulsion team. At $9.9M across 7 rounds (Crosslink-led Oct-2024 round plus SBIR) it remains just under cap, with no newer raise found - flag it to re-check, since one more round tips it over. The thesis is capital-ferocious and speculative (lunar water is unproven, and a crowded race), so it's the slowest-return profile here. Keep for the elite team and mission; underwrite the very long timeline.

Hex (Hextronics) builds autonomous drone-in-a-box docking stations with rapid battery swapping, seeded with $2M from Founders Fund - verified, well under cap. Founder Curt Lary (Georgia Tech) bootstrapped the first prototype from his family's garage in 2020 and has since enabled 100k+ flights across 25 countries, with an early SpaceX launch engineer on the team. Genuinely engineering-dense and deep-tech, but capital-intensive hardware with a thin post-seed raise history - diligence current traction and burn. Keep for elite validation plus real deep-tech signal.

Durin builds sensor-instrumented, increasingly automated drill rigs for critical-minerals exploration, raising a $3.4M pre-seed led by 8090 Industries with Andreessen Horowitz participating - a16z verified, under cap. Founder Ted Feldmann is a third-generation mining professional (Georgia Tech) marrying robotics to an industry where drilling eats ~70% of exploration capital, backed by a strong deep-tech syndicate (a16z, Lux, 8090, Bedrock) and an ex-SpaceX engineer. Very early and capital-intensive, and deep-tech rather than pure software, but the elite validation and genuine builder DNA are real. Keep.

Wondercraft is a YC- and Will Ventures-backed AI audio studio from ex-Palantir engineers (Dimi Nikolaou, Youssef Rizk) plus a Spotify/Acast media operator, turning text into podcasts and audiobooks at scale. It has real traction - 30k+ users and marquee podcast customers - on a confirmed $3M seed (under cap). The main risk is heavy reliance on ElevenLabs TTS in a commoditizing category. Solid keep, funding fully verified.

Satlyt is building decentralized 'compute in orbit' - LEO edge data centers that let satellites process data locally and sell spare cycles. The team is genuinely pedigreed (ex-SpaceX Starlink PM, a repeat founder) and has NASA STTR + Antler backing on ~$1M raised. It's a hype-adjacent category with brutal physics and a long revenue runway, so treat it as a high-variance moonshot. Keep for the pedigree and external validation, but underwrite the thermal/economics risk.

Belli builds end-to-end air-cargo management software for complex airlines, founded by ex-SpaceX/ex-Teleport operators who know the cargo-IT pain firsthand. It has real paying customers (Etihad, SpaceX), a $2.3M seed led by Iterative, and a Palantir Startup Fellowship slot. The auto-scorer over-credited the Palantir fellowship as a $2.5B founder 'exit' - that's noise; the real signal is genuine traction in a sleepy, digitizable vertical. Keep as a solid seed-stage operator play.

Exlens is a quality/inspection operating system for factories built by two ex-Tesla factory-software directors - precisely the people who lived this problem at scale. The product framing (an AI reasoning engine that clusters defects to root cause and drafts the CAPA) is sharp, and the founding pedigree is director-level and verified. The catch: its ~$3M round isn't publicly confirmed and the space is crowded. Keep for the elite domain authority; confirm the round before deep diligence.

T51 is a 2025-vintage, two-person Palantir-mafia startup (Jonah Adler, MIT; Vipul Shekhawat, Harvard, ex-Palantir CTO) building forward-deployed AI decision systems on top of manufacturing/logistics ERPs. The pedigree and lean FDE ethos are exactly the profile that wins in industrial AI, but it is pre-funding, pre-traction, and in a crowded lane. Keep and watch - conviction depends on a real deployment and a named lead investor.

Motives (YC S25) replaces the $20k, two-month consumer focus group with a team of AI agents that runs and analyzes video interviews in a day or two, and it already has real logos - big US/European consumer brands including Unilever's Wild. The founding pair is credible: ex-Palantir CTO Sam Stoll and ex-Google/TrueLayer CEO Sean Conley, both Oxford. Funding is a small, honest pre-seed (Rebel Fund + YC), well under cap. The risk is a crowded AI-research category and an unproven moat, but the early commercial pull and lean team make it a solid watch.

Headroom is a lean, engineering-heavy (4 of 7) team of ex-Palantir operators building an 'AI operating system' that turns the SMB inbox into a CRM/task system. Pedigree is genuine - co-founder Jishai Evers and colleagues are verified ex-Palantir - and the team is refreshingly small and unfunded (do not confuse it with the unrelated ex-Palantir '$12M' raises at Perceptic/Conduct). The concern is a brutally crowded AI-email/SMB category with zero disclosed traction yet. Strong team, unproven wedge: a KEEP to watch for early customer signal.

Bastion is an AI-native compliance and security platform built by an ex-Palantir cybersecurity team (7 years protecting sensitive orgs), with a shipped, purchasable product (AWS Marketplace) and a clean, in-cap $2.8M seed. The forward-deployed 'embedded engineers + AI agents' wedge is a credible differentiator in an otherwise brutal compliance-automation market dominated by Vanta and Drata. Pedigree and product fit are strong; the concerns are heavy competition, a sales-weighted org, and no marquee logos yet. A solid KEEP.

Ryvn (YC) lets B2B SaaS companies deploy into customers' own clouds and on-prem/air-gapped environments with a Vercel-like push-from-GitHub flow - and its two ex-Palantir founders built exactly this infrastructure at scale (10k+ deploys/day across hundreds of environments). The team is razor-lean (4 people), YC-backed with an in-cap ~$500K pre-seed, and the pedigree-to-product fit is as tight as it gets. The risks are scope-vs-headcount and a contested BYOC-deployment market with no public logos yet. Strong founder fit; KEEP pending traction.

Webstudio is a mature open-source Webflow/Framer alternative built by a genuinely credible founder - Oleg Isonen, ex-Webflow engineer and creator of the JSS library - and backed by a notable angel roster (Naval, Masad, Balaji) plus Seedcamp/Tiny.vc on a small, in-cap ~$1.2M pre-seed. It has a real open-source user base and paid hosted tiers, but faces the classic open-core monetization challenge against deep-pocketed incumbents, with ARR undisclosed. Strong technical/community pedigree; the open question is durable revenue. A KEEP.

Teller offers a screen-scraping-free banking-data API (7,000+ institutions) positioned directly against Plaid, seeded with $4M from Founders Fund, Lightspeed and Max Levchin's SciFi VC - an elite fintech syndicate, verified and under cap. Founded 2014 by Stevie Graham, it's run remarkably lean (18 of 25 are engineers) and quiet since its 2020 seed, which reads as either disciplined bootstrapping or a stalled trajectory - diligence current revenue. Off-core-thesis regulated fintech, but genuine engineering density and elite validation. Keep.

anchr builds an AI-native back-office/logistics OS for food distributors, raising $5.8M led by a16z Speedrun with OpenAI-ecosystem angels - the elite-investor claim is verified (via Speedrun, a16z's accelerator, rather than a partner-led fund round). Founded in 2025 by childhood friends Tzar Taraporvala and Smayan Mehra; note the Tesla 'Senior PM' pedigree badge actually belongs to a senior software engineer (an ex-intern), not a founder. Very young and off-core-thesis vertical SaaS, but real engineering density and a credible AI-ecosystem syndicate. Keep, with the pedigree caveat flagged.

AL Advanced Industries is a lean, two-person German team of ex-Tesla security engineers building ruggedized autonomous surveillance (video + radar) for critical infrastructure that works where power and connectivity don't. The Tesla physical-security pedigree is a genuine fit for the problem, and the company is refreshingly unfunded rather than hiding an over-cap round. The concern is pure earliness: no disclosed customers, no capital, and a hardware-heavy category that usually demands both. A watch-and-verify candidate, not a conviction bet.

Viscera (viscerahealth.com) is a 2025 Palantir-alumni startup automating documentation, coding and billing for urology practices - a 'digital revenue operator' with a clean real-time-to-EHR wedge and fast hiring (+90% in 90 days). Co-founder David Starnino has a verified ~4-year Palantir tenure; the 'Jeremy Myers' pedigree carries a name-collision caveat (Palantir FDE vs a same-named urologist MD) worth confirming. Note the separate, unrelated $8M IBD 'Viscera Health' - don't conflate. Narrow but credible; keep and diligence founder identities and funding.

Endeavor is a Craft Ventures-backed ($7M seed, under cap) AI platform automating sales/production/supply-chain workflows for American manufacturers, founded by 22-year-old Wharton grad Sami Senapathy. The Palantir/AWS pedigree is real but college-era, not senior FDE - don't over-index on it. Confirmed funding and a live product are positives; a ~22% headcount drop and a very crowded lane are cautions. Keep, with diligence on burn and founder depth.

Overstand Labs (YC W25) turns a company's scattered customer communications - email, Slack, calls, CRM - into real-time AI insights for expansion and issue-spotting, built by an ex-Palantir BD + PM/eng founding pair with genuine customer-facing pedigree. It's a lean, in-cap YC company (~$1M incl. YC), but with only one engineer, a crowded revenue-intelligence field, some positioning drift across YC launches, and no disclosed logos, execution risk is high. Credible founders in a hard, contested category - a borderline KEEP.

Kyten is bringing aerospace-grade battery packs to the whole aerospace and defense industry, promising qualification and delivery in weeks instead of months. Founder Cooper McBride went from a Tesla intern to building and leading 30-plus person engineering teams at SpaceX (Starlink) and Shield AI, and the team has shipped thousands of packs to low earth orbit. A lean YC W26 company on roughly $500K, with elite hardware pedigree against a real supply-chain bottleneck. Strong.

Beyond Reach Labs makes deployable solar arrays that unfold to football-field scale in orbit, targeting the power shortage facing orbital data centers and stations. CTO Pele Collins ran parachute engineering across 30-plus SpaceX Dragon missions, and CEO Mitchell Fogelson is a CMU/NASA deployable-structures PhD who invented the core truss tech. It just closed a $10M seed led by Interlagos, which sits it right at the top of our thesis band, so a solid keep rather than a low-raise gem.

BloomPath AI is an operational-intelligence layer that watches engineering work in real time and flags bottlenecks and risks without status meetings. Co-founder Mohammad Kotb brings a verified staff engineering role at Waymo plus senior time at Google, and the company has a clean $1.3M pre-seed led by Raed Ventures. The pedigree and low raise fit the thesis; the dev-productivity category is the main knock. Keep.

Klavis AI sells the live, verifiable environments and agentic data that frontier labs need to post-train long-horizon agents, plus Strata, an MCP server that beats the official GitHub and Notion servers on the MCPMark benchmark. Founder Xiangkai Zeng was a senior engineer on Google Gemini at DeepMind and co-authored the Gemini paper. On roughly $500K with HSG and YC behind it, a high-pedigree, low-raise infrastructure bet squarely on thesis. Strong.

Minerva Humanoids is taking on humanoid actuators and controls, one of the genuinely hard problems in embodied AI, from a near-zero raise. Founder Sandor Felber is a former MIT CSAIL humanoid-robotics researcher who interned on Tesla's Optimus program and has the Forbes AI25 nod to match, backed by high-signal scouts including Pareto and Sequoia Scout. Elite young-founder pedigree plus a pre-seed check size. Strong.

Gibran is an ex-Google DeepMind team building foundational AI that performs in low-data regimes, borrowing from complex adaptive systems and starting in drug discovery. The $2.6M Together Fund seed and the DeepMind lineage check out, though the pedigree is spread across the team rather than one marquee role, and it sits a bit off the core US-frontier-lab thesis. A credible keep on strength of team and thesis.