Plug and Play University Pitch Day — Daxos Teardown
2026-09-03 · Internal draft
39
Rated
0
Hard-filtered
6.50
Median company rating
Glacian / Nirubi
Top pick (tied, FOR DAXOS 6.5)
Top picks
Glacian TechnologiesFOR DAXOS 6.5 · company 7.75 — US data-center cooling AI with DOE/DoD research lineage and a tiny verified raise. Take the meeting; first question is the $1.95M-vs-$150K funding gap.
Nirubi TechnologiesFOR DAXOS 6.5 · company 7.50 — Bootstrapped CMU SEI formal-verification spinout claiming USAF/Army/DOE contracts. Most Daxos-shaped in the field; verify awards went to Nirubi Inc. via SAM.gov, then meet.
Rockfish DataFOR DAXOS 6.0 · company 7.25 — CMU-professor synthetic-data platform with verified seed, investors, and USG customers. $5.5M already in, so a small check dilutes fast.
DentiteFOR DAXOS 5.5 · company 7.25 — Real operating dental-medical billing business on a small raise, revenue-aligned pricing. Headline metrics are founder-supplied; verify ARR before a check conversation.
LimFi LabsFOR DAXOS 5.5 · company 7.00 — Repeat-exit treasury founder (Speranza, acquired) at $0 raised. No product evidence yet; the pedigree is the whole bet. Take the meeting.
LyoWaveFOR DAXOS 5.5 · company 7.00 — Purdue lyophilization retrofit with a verified Millrock distribution channel. Picks-and-shovels pharma manufacturing; slow capex sales cycles cap it.
VistaXMFOR DAXOS 5.5 · company 6.75 — Thirty-year channel-CX operator; the entire thesis hangs on unverified Dell and HPE anchors. Worth one call to verify the Dell contract, nothing more without it.
Lavoro AIFOR DAXOS 5.0 · company 7.75 — A-tier CMU Robotics pedigree (Jean Oh) in pre-product, capital-intensive elder-care robotics. Thesis-stage.
LupitorFOR DAXOS 5.0 · company 7.00 — Best traction signals in the field if real ($514K+ MRR self-reported), but the funding story is internally inconsistent and a $30M priced pre-A blunts a small check.
Citrus OncologyFOR DAXOS 4.5 · company 7.50 — Clinician-founder quality and a tiny raise, but crowded oncology care-navigation with no gov angle is the wrong shape.
Full table
Click a header to sort. Funding shows the verified (Harmonic) figure where one exists; * marks a self-reported deck figure. Default sort: FOR DAXOS descending.
No company blocks are missing or malformed; all 39 rated blocks plus the Lone Star Lunar teardown are complete. Three companies were rated at low confidence because they could not be firmly identified, which is itself the finding:
Winnow (#9): no website, no Harmonic entry, no press; every public "Winnow" is a different company. Get the real domain from Plug and Play before any further look.
Zelena (#18): the deck blurb (AI workflow automation) does not match the only findable zelena.ai (data-center efficiency). Needs deck contact info to disambiguate.
Henry AI (#17): founders unidentifiable; severe name collision with the unrelated YC S24 CRE company of the same name.
WHATPhysical AI (physics-based digital twins + ML) that cuts data center cooling power so operators can run more compute on the same grid connection.
FUNDING$150K, 2 pre-seed rounds (Harmonic). Deck's $1.95M is NOT in Harmonic — flag the gap. The "$8M+ in grants" on their site is 10+ years of DOE/DoD/NSF research funding to the parent university labs pre-spinout, which does not count as the company's raise.
CLAIMS53%/74% cooling-savings figures trace to DOE-funded research projects (Penn State news), i.e. lab research, not paying-customer results. One commercial integration deal (Alerify data center, Harrisburg) verified via the same PSU press. No revenue disclosed.
FOUNDERSJie Zhao (CEO, CMU PhD — the deck's CMU tag) + Wangda Zuo (CTO, Penn State professor of architectural engineering, SBS Lab); professor-CTO part-time risk, 2 people, founded 2026.
COMPETITIONPhaidra ($60M+, ex-DeepMind cooling control) is the direct incumbent; also Etalytics, Vigilent, and Siemens/Schneider BMS suites.
COMPANY RATINGBase 5.75 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 7.75
FOR DAXOS6.5 - US picks-and-shovels on the data center buildout with DOE/DoD research lineage and a tiny raise; capped by 2-person academic-spinout stage and Phaidra overhead. Take the meeting.
FLAGSdeck raise ($1.95M) 13x Harmonic ($150K) — reconcile before any check; grant-vs-raise conflation on their own site.
WHATNKlave — formally verified, mathematically proven trusted-path security framework for mission-critical software, autonomous/aviation platforms, and agentic AI workloads, spun from a decade of CMU SEI/CyLab research.
FUNDINGBootstrapped (deck; Harmonic: $0, EXISTS_BUT_UNDISCLOSED). CMU-era NSA SoS Lablet / NSF / DoD grants are pre-spinout parent-org R&D and do NOT count as the raise.
CLAIMS2022 NSIN award, 2022-23 USAF contracts (close-combat aviation platforms), 2024 Army aviation/missile contract, DOE cybersecurity contract — all trace to the company's own about page; no SAM.gov/SBIR-database confirmation found in this sweep — SELF-REPORTED, UNVERIFIED at entity level (possible they ran through CMU SEI, not Nirubi Inc.). DEFCON 33/34 presence referenced on site.
FOUNDERSAmit Vasudevan (founder/CEO) — senior CMU SEI/CyLab researcher, creator of uberSpark/uberXMHF, the real formal-verification pedigree (linkedin, scholar). Solo-academic-founder commercialization risk.
COMPETITIONGalois, RunSafe Security, Kry10, and the cautionary comp BedRock Systems ($84M raised, still got absorbed) — formal methods historically monetize as services, not products.
COMPANY RATINGBase 5.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 7.50
FOR DAXOS6.5 - the most Daxos-shaped in the batch: bootstrapped, US, defense infrastructure, tiny cap likely, and claimed Army/USAF deployments; capped until contracts verify at the entity level. If SAM.gov confirms, this is the take-the-meeting pick and re-rates up.
FLAGSAll contract claims self-reported; verify awards went to Nirubi Inc. (not CMU SEI) before any outreach; 5-person team, no commercial revenue visible.
WHATGenerates high-fidelity domain-specific synthetic time-series/operational datasets so enterprises can eval, test, and train AI models without exposing production data.
FUNDING$5.49M total, 5 rounds, last $5.45M seed Oct 2024 led by Emergent Ventures w/ Foster, TEN13, Dallas VC, NewBuild (Harmonic; consistent with deck $6M)
CLAIMSSeed round and investors verified (businesswire, techcrunch). Conviva as customer and "US government departments" verified via press. Dell/HPE deck claim: no independent evidence found — UNVERIFIED.
FOUNDERSVyas Sekar + Giulia Fanti (both CMU professors, decade of generative-model research since 2016) + Muckai Girish (CEO, serial telecom exec) + Nathan Haugo. Strongest pedigree in this batch.
COMPETITIONGretel (acquired by NVIDIA), Mostly AI, Tonic.ai, Hazy (acquired by SAS) — the category's two best exits were absorptions, which caps outcomes but validates demand.
COMPANY RATINGBase 6.0 + EV +0.25 + Geo +1.00 + Chain 0 = FINAL 7.25
FOR DAXOS6.0 - real pedigree, USG customers, picks-and-shovels shape; but $5.5M in and 4 years old with modest visible commercial velocity, and the next round dilutes a small check fast.
FLAGSDell/HPE claim unverified; category consolidating into NVIDIA/SAS.
WHATB2B SaaS that lets dental practices bill patients' medical insurance for medically necessary dental procedures, automating claims end-to-end on a performance-based "we only get paid if you get paid" model.
FUNDING$1.75M in Harmonic (7 rounds, last Nov 2025, pre-seed stage; Launch NY, NY Ventures, University at Buffalo, GrowthX, MATTER, Bold Venture). Deck's $1.6M is roughly consistent.
CLAIMS"+12% new revenue" and "10-13% practice revenue increase" — the proof case is White Dental Studio, which is the CEO's own practice (self-referential, flag). 29 states / ~2000% 2024 customer growth appear in UB partnership material (https://www.buffalo.edu/partnerships/collaborate/case-studies/dentite.html) — founder-supplied to a friendly source, UNVERIFIED independently. "Elaine" the AI billing specialist appears in the deck only, not on the site — UNVERIFIED/new. No ARR disclosed.
FOUNDERSDr. Petar Tofovic (CEO, practicing dentist and practice owner — Pitt per deck), Jerod Sikorskyj (CTO), Dan and Xenia Richter. Founder-market fit real; no prior exits.
COMPETITIONToothy AI (YC, same dental-billing automation), Overjet ($130M+, moving into billing), Vyne Dental, DentalXChange, Zentist — crowded RCM space, though medical-insurance-at-the-dentist is a differentiated wedge.
COMPANY RATINGBase 5.5 + EV +0.75 + Geo +1.00 + Chain 0 = FINAL 7.25
FOR DAXOS5.5 - best of my five: real operating business, revenue-aligned pricing, small raise, US, 4 years of execution; capped because it's healthcare RCM (no gov/defense or infra angle) and the headline metrics trace back to founder-supplied sources.
FLAGSproof-case customer is the CEO's own practice; "Elaine" product deck-only; none hard-filter.
WHATControl layer above ERPs, TMS, and banks for multi-entity treasury — in-house banking, real-time intercompany netting, and policy governance embedded in payment decisions.
FUNDING$0 in Harmonic; $0 per deck (consistent).
CLAIMSSite metrics ("74% netting efficiency," "$1.1M typical annual savings," "$300M→$78M gross-to-net") are illustrative with no named customers — UNVERIFIED, likely modeled not observed. No deployments claimed anywhere.
FOUNDERSGlen Solimine, founder/CEO — 25 years treasury, ex-BofA/JPMorgan, founded Speranza Systems (acquired by Wallstreet Systems), Treasury & Risk "100 Most Influential"; CTO Joseph Roveda ex-JPMorgan (led 200+ engineers, wholesale payments); advisors include Ed Banas (ex-CFO/Treasurer PayPal, Kraken). Strongest team of the slate by a wide margin.
COMPETITIONKyriba, FIS, TIS, Coupa/Bellin on in-house banking; Nilus and Fennech on the modern-stack side — treasury orchestration is incumbent territory with long enterprise sales cycles.
COMPANY RATINGBase 5.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 7.00
FOR DAXOS5.5 - exactly the right check size ($0 raised) and a domain founder with a prior exit in this precise space, but rated on what EXISTS: no product evidence, no customers, 3 people. Take the meeting; the pedigree is the whole bet.
WHATModular microwave/qRF heating systems that retrofit pharmaceutical freeze-dryers to cut lyophilization cycle times and energy use.
FUNDINGHarmonic shows a $354K grant (5/2025) with equity "exists but undisclosed"; deck's $1.5M is self-reported — scored on the deck figure.
CLAIMSPurdue license and Millrock Technology distribution collaboration VERIFIED via Purdue newsroom. Caveat: the ~$1M NIIMBL/Gates grant and the Merck + IMA Life partnerships attach to the Purdue research project, not necessarily the operating entity — per the spinout rule that money doesn't count as the company's raise.
FOUNDERSDrew Strongrich (CEO, Purdue AAE PhD) + Alina Alexeenko (President, Purdue professor and LyoHUB co-director — arguably the leading lyophilization academic in the US). Note the deck's "Penn State" tag is wrong or unverified; this is a Purdue spinout.
COMPETITIONIncumbent lyo OEMs IMA Life, GEA, SP Scientific/ATS, Tofflon; Millrock is partner rather than rival, which is the smart channel move; EnWave does microwave drying in food.
COMPANY RATINGBase 5.25 + EV +0.75 + Geo +1.00 + Chain 0 = FINAL 7.00
FOR DAXOS5.5 - best fit of the four: picks-and-shovels manufacturing infrastructure, US, domain-defining founder, verified industry channel; capped by slow pharma capex sales cycles and no gov/defense pull.
FLAGSdeck university attribution wrong (Purdue, not Penn State); parent-project grants risk being conflated with company raise.
CLAIMSDell anchor OEM and HPE signed — appear NOWHERE outside the deck; the Dec 2025 PartnerPulse launch went out via paid EIN Presswire with no named customers. Both UNVERIFIED. Harmonic: 8 heads, Salt Lake City, founded Mar 2024, and describes the offering as CXaaS "utilizing platforms like Qualtrics" — i.e., built on Qualtrics rails, services-tinged.
FOUNDERSErik Vogel — CMU Tepper, ~30yr operator, built HPE GreenLake's CX program, led high-tech/telecom at Qualtrics, ex-Cisco (Crunchbase). Best pedigree in my block.
COMPETITIONQualtrics and Medallia themselves, Impartner, 360insights, Crossbeam — and the anchor-vendor dependency cuts both ways.
COMPANY RATINGBase 5.0 + EV +0.75 + Geo +1.00 + Chain 0 = FINAL 6.75
FOR DAXOS5.5 - right raise size, US, genuinely strong domain founder; but the whole thesis hangs on unverified Dell/HPE logos and it's a niche measurement layer on someone else's platform. Worth one call to verify the Dell contract; nothing more without it.
FLAGSnone hard; deck's central traction claims (Dell/HPE) unverified.
WHATPhysical AI robots for caregiving/elder care, built on CMU Robotics Institute research, currently pre-product with an open-source robot-control framework (RIO) as visible output.
FUNDING$0 announced per Harmonic (NONE_ANNOUNCED, 0 rounds). Deck's $1.5M is self-reported and unverifiable — scored on the deck figure.
CLAIMSFounder identities verified on the company site. No customers, deployments, or revenue claimed or found. Note: Harmonic describes the product as a robot-control framework while the site says caregiving robots — early-stage positioning is still moving.
FOUNDERSJean Oh (ex-CMU RI Associate Research Professor, director of Bot Intelligence Group, ICRA/IROS/HRI best-paper awards) + Mooyeon Oh-Park MD (30-year physiatrist, led Age-Friendly Care at Burke Rehabilitation). Top-decile academic pedigree.
COMPETITIONIntuition Robotics (ElliQ), Labrador Systems, Diligent Robotics, Hello Robot, plus 1X/Figure pushing humanoids into the home; elder-care robotics is a capital-intensive graveyard.
COMPANY RATINGBase 6.0 + EV +0.75 + Geo +1.00 + Chain 0 = FINAL 7.75
FOR DAXOS5.0 - A-tier founder pedigree, but pre-product capital-intensive robotics with zero execution evidence; thesis-stage caps it.
FLAGSdeck's $1.5M raise has no Harmonic or public trace; product story inconsistent between site and data providers.
WHATConversational/voice AI agents for customer service with full on-premise deployment as the wedge, targeting security-sensitive and international (non-cloud) markets.
FUNDINGHarmonic $0 raised; deck $4.3M self-reported ("4.25 of 5M seed at $30M priced with lead"). Web chatter separately mentions a $3.2M seed incl. Alex Hormozi's ACQ Ventures + YC — the two stories conflict and neither is in Harmonic. Treat the $30M priced round as UNVERIFIED.
CLAIMS"$514K+ MRR, 70%+ call automation, $5M+ sales enabled" comes from a former founding engineer's LinkedIn — self-reported, UNVERIFIED. Harmonic confirms 15 heads, SF, founded Jun 2025 — fast headcount if real.
FOUNDERSCMU CS circle — Derek Armfield, Alek Stefanov, Rhythm Vij (CMU SCS) all show Lupitor + YC affiliation; but Armfield/Stefanov also surface as founders of Acrely (YC, "Enterprise Voice Agents") and founding engineer Imren already left for Atlantic — either a rebrand or founder churn, unclear which.
COMPETITIONSierra ($4.5B), Decagon, PolyAI, Parloa, Retell, Vapi, Bland — the most capitalized agent category there is; on-prem is a real but narrow differentiator.
COMPANY RATINGBase 5.5 + EV +0.50 + Geo +1.00 + Chain 0 = FINAL 7.00
FOR DAXOS5.0 - best traction signals in my block if real, but a $30M priced pre-A makes a $100-500K check near-meaningless, the funding story is internally inconsistent, and the founder/Acrely ambiguity needs a direct answer before anything.
FLAGSfunding-narrative inconsistency (deck $4.25M@$30M vs web $3.2M vs Harmonic $0); possible entity split/rebrand with Acrely; founding-team churn.
WHATAI triage plus on-demand virtual specialty care for cancer treatment side effects (toxicities), sold to health systems as a coordination layer for oncology nursing teams.
FUNDING$220K (Harmonic, 5 accelerator/grant rounds: Techstars, CancerX, NC IDEA, CED; last 3/2026). Deck's $365K is higher, self-reported.
CLAIMSBacking by CancerX, Techstars, CareFirst/Healthworx confirmed on site and in Harmonic investor list. The "11% to 2% hospitalization reduction" figure traces to Healio coverage of the Duke-affiliated virtual clinic model, not an audited company metric — partially verified. No named health-system deployments or ARR verifiable anywhere. "Air traffic control" framing appears only in the deck.
FOUNDERSAabid Shariff (CEO, CMU computational biology PhD — the CMU hook) and Afreen Shariff MD (Duke endocrinologist specializing in immunotherapy toxicity). Real clinical credibility.
COMPETITIONMemora Health, Canopy, Jasper Health, Outcomes4Me, Hippocratic AI nursing agents, and Epic's own care-coordination modules; crowded oncology care-navigation field.
COMPANY RATINGBase 5.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 7.50
FOR DAXOS4.5 - clinician-founder quality and tiny raise are right, but healthcare-services AI with no gov angle and crowded incumbents is the wrong shape.
FLAGSnone hard; deck raised figure exceeds Harmonic by $145K.
WHAT"Deep-SKAI" decision layer that turns ERP/supplier/market signals into governed, auditable sourcing, risk, and pricing decisions for aerospace and defense supply chains.
FUNDING$935K deck-self-reported; Harmonic shows one undisclosed STRATEGIC round 2025-08-07 (funding exists-but-undisclosed), consistent with sub-$1M.
CLAIMSOwn site honestly labels demo scenarios "illustrative scenario — no named customer" — so zero verified customers. Deck's "founders with $100M+ exits and 35 patents" — UNVERIFIED, nothing public supports it. "Proprietary aerospace transaction data" — UNVERIFIED, source unclear for a 2024 startup.
COMPETITIONInteros, Resilinc, Exiger, Altana — all heavily funded A&D supply-chain-risk incumbents; NexStratus's wedge is decision execution vs their monitoring.
COMPANY RATINGBase 5.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 7.00
FOR DAXOS4.5 - real team and clean raise size, but no verified customer, no gov contract, and headline pedigree claims fail verification; take a meeting only if the exits/patents check out.
FLAGSpedigree claims unverifiable (the deck's central credibility asset); 4 co-founders + fractional CPO on <$1M = burn/dilution shape; consulting-firm origin.
WHATZero-trust, post-quantum-secure health data exchange using the KERI protocol (self-certifying key infrastructure) to sign and verify every transaction peer-to-peer without third-party certificate authorities.
FUNDING$620K in Harmonic (2 rounds, last a convertible note Mar 2025; investors Techstars, Key State Capital). Deck says $1.9M — DISCREPANCY, the extra ~$1.3M is self-reported and not on record.
FOUNDERSJared Jeffery (CEO, decade in health interoperability, ex-KLAS Research) + Dr. Samuel Smith (KERI inventor, Chief Architect) — strongest technical pedigree in this batch of five.
COMPETITIONTEFCA/QHINs and DirectTrust own healthcare exchange trust today; Health Gorilla, Particle Health on data exchange; generic PQC vendors (SandboxAQ) on the quantum angle. Protocol-evangelism businesses move at standards-body speed.
COMPANY RATINGBase 5.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 7.00
FOR DAXOS4.5 - real inventor-grade pedigree at a tiny raise is the right shape, but zero commercial evidence after 3 years and a sales motion that requires the healthcare industry to adopt a new trust protocol; if the deck's $1.9M is real, EV drops to +0.75 and Final to 6.75.
FLAGSfunding discrepancy deck $1.9M vs Harmonic $620K — ask for the cap table; none hard-filter.
WHATMeters what AI features cost to serve per model call (1,100+ models), matches cost to revenue, and produces gross margin per customer/feature plus auditable accounting closes (QuickBooks/NetSuite).
FUNDING$0 in Harmonic; $60K deck-self-reported.
CLAIMSSite makes no revenue or customer-count claims; only traction artifact is a testimonial from a fractional-CFO design partner — UNVERIFIED beyond that. Note: a separate "UnitPay, Inc." exists in the same niche and one index suggests a rebrand link; marginfront.com is live with no mention, treated as a name collision.
COMPETITIONMetronome ($100M+ raised), Orb, Stripe (native AI profit/usage billing), m3ter, Lago, OpenMeter — the best-funded corner of billing infra.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS4.5 - right picks-and-shovels instinct on a real problem (AI gross margin opacity), but two people with $60K walking into Metronome/Stripe's lane with no distribution wedge shown.
WHATAI "design intelligence" that mines a firm's past building projects to auto-generate schematic-to-BIM designs for repeat-build owners (this is the Atlanta AEC startup, not SKEMA Business School).
FUNDINGHarmonic shows SEED, 4 rounds, but only $300K of disclosed amounts — clearly incomplete given the investor list (STO Building Group, Thornton Tomasetti, SHoP Architects, Syska Hennessy, AEC Angels, Taylor family office). Deck's $10M is self-reported and plausible; EV tier taken conservatively from the deck figure.
CLAIMS"Automates up to 75% of building design" is company marketing — UNVERIFIED, no customer-published numbers. Real product shipping since ~2023-24 with AIA conference presence (Architosh coverage); strategic AEC investors are genuine validation. No ARR disclosed.
FOUNDERSMarty Rozmanith (co-founder/CTO, original Revit product team, ex-Autodesk/Dassault), Charlie Cichetti (CEO), Arol Wolford (construction-data veteran, early Revit backer) — top-decile AEC-software pedigree (AEC Business).
COMPETITIONAutodesk Forma (the incumbent selling to the same seats), Higharc ($53M+), TestFit, Hypar, Augmenta.
COMPANY RATINGBase 5.50 + EV +0.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS4.0 - real pedigree and product, but a ~$10M-in company with 14 people and no disclosed ARR is past the Daxos entry shape; a $250K check is noise on this cap table.
WHATAI-native PLM/QMS system of record for aerospace and defense hardware (parts, BOMs, compliance, change orders) claiming to replace 10-15 disconnected tools.
FUNDING$0 (Harmonic: no rounds announced, founded 2025, 1 headcount, Denver CO)
CLAIMS"26 modules / 31 AI capabilities" appears only on own site — UNVERIFIED; no customers, deployments, or revenue claimed or found anywhere.
FOUNDERSGeorge Bondarev, solo — Penn State ME, propulsion engineer (Quantum Space) per LinkedIn/RocketReach; real domain user, no exits.
COMPETITIONFirst Resonance (ION), Duro, Epsilon3, plus Siemens Teamcenter/PTC Windchill incumbents — well-funded and directly on this thesis.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS4.0 - right picks-and-shovels A&D shape, but solo, pre-revenue, pre-team; thesis only, nothing exists to underwrite.
FLAGSsolo founder / 1 headcount (borderline solo-project); zero external validation.
WHATAI-native hiring platform that builds deep-researched candidate profiles, runs AI voice conversations, and puts candidates through virtual work trials before interviews.
CLAIMSCustomers OpenObserve, Concentrate AI, Timelaps, Infraspec appear only in the founder's own LinkedIn post — UNVERIFIED. Harmonic confirms 3 heads, SF, founded 2026.
FOUNDERSShubham Srivastava (CMU MS, ex-Head of Product/Growth at Zenduty, $200K→$3M ARR to acquisition) + Yash Thakkar (CMU) — LinkedIn.
COMPETITIONMercor, micro1, HireVue, WorkTrial AI, Paraform — one of the most crowded AI categories alive.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS4.0 - crowded HR tech, no gov/infra angle, modest pedigree, all traction self-reported; thesis-stage.
WHATAI-native CPQ / quote-to-cash platform that unifies CRM, email, and document data into one grid to run complex enterprise deals from quote to booked order.
FUNDING$0 in Harmonic; $220K deck-self-reported.
CLAIMSSite makes no traction claims at all (no ARR, no named customers) — nothing to verify, nothing verified. Harmonic shows 9 heads (largest team of this slate), West Palm Beach, founded 2025.
FOUNDERSCEO Matthew Jeanty (thin public record, Penn State tie unverified); board chair Peter Horst, a real F500 CMO (Hershey, Capital One) per https://cmoinc.com/about/.
COMPETITIONSalesforce Revenue Cloud, DealHub, Conga, Tacton, plus a wave of AI-CPQ startups — deeply crowded incumbent category.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS4.0 - zero execution evidence to underwrite, unknown CEO, incumbent-owned category; the board chair is the only verifiable asset.
FLAGSnone hard; CEO's public record is unusually thin for a 9-person company.
WHATAI business-intelligence layer for marketing/sales teams that turns fragmented campaign data into decision-ready analysis, with "Emet" as an anti-hallucination checking layer.
CLAIMS"Zero hallucinations" appears only on the company site (fuse.is) — UNVERIFIED and indefensible as an absolute; no named customers or ARR independently visible. Company pivoted from DeckFusion (deck-generation) to BI (crunchbase).
FOUNDERSJeff Cherkassky (CEO, 4x founder, one 2018 exit of undisclosed size) + Jill Randell (COO, ex-VP Client Strategy at Found Search Marketing) (emarketingassociation.com). Agency-world pedigree, not AI research.
COMPETITIONDatabricks Genie, Snowflake Cortex Analyst, ThoughtSpot, Zenlytic on AI-BI; Improvado/Supermetrics on marketing data — crowded from above and below.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS4.0 - raise size fits a Daxos check but it's a wrapper in a category the warehouses own; no infra or defense angle; thesis-only caps it.
WHATAI energy management system for DC fast-charging hubs that shapes power in real time so charge point operators sell more energy and cut demand charges.
FUNDING$0 recorded in Harmonic (0 rounds). Deck's $800K is self-reported.
CLAIMS"Commercially deployed with Kempower/Zerova/Alpitronic/ABB/Siemens/Delta" on their site reads as hardware compatibility, not named customers — no CPO deployment, site count, or revenue is published anywhere. The "+40% maximum power" case study is anonymous. All UNVERIFIED.
FOUNDERSCEO listed only as "Oori S." on LinkedIn; Amit Cohen (Kellogg MBA) on the commercial side. The deck's CMU tie is unverified. 7 people, Haifa, founded 2023.
COMPETITIONDriivz (Vontier), AmpUp, Electric Era, Monta, and Wevo Energy (just acquired by SolarEdge) — crowded CPO-software field.
COMPANY RATINGBase 4.75 + EV +1.00 + Geo +0.25 + Chain 0 = FINAL 6.00
FOR DAXOS4.0 - no verified traction, no gov angle, crowded category, Israel ok but nothing to underwrite beyond thesis.
FLAGSnone hard; zero named customers after 3 years is the soft flag.
WHATPre-PLM "product intelligence" platform that turns a consumer brand's reviews, POS, and manufacturing data into a knowledge graph producing continuously updated product briefs.
FUNDINGHarmonic: $0-1M tier — grants/accelerator money (NSF SBIR, Techstars Boston, AJI Capital; last recorded a $275K grant Jun 2024; totals internally inconsistent). Deck's $650K self-reported; company elsewhere claims $1.25M non-dilutive. All in the same EV tier.
CLAIMSNo named customers found anywhere; the "secure per-brand knowledge graph" and lifecycle claims are deck/site-only — UNVERIFIED. 10+ papers and 4 patents claimed from the academic lineage (Crunchbase/company materials, unaudited). Company is a rebrand of Ada IQ (Techstars '24), second naming change.
FOUNDERSTucker J. Marion (Northeastern professor of innovation/product development — part-time academic risk) + Stuart Nixdorff (sales exec); hired outside CEO Liz Graham in 2024 (PR Newswire) — founder-led it is not.
COMPETITIONCentric Software owns consumer/retail PLM; Backbone PLM, Bamboo Rose upstream; Zappi/quantilope on the AI consumer-insights side.
COMPANY RATINGBase 4.50 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS3.5 - four years, two names, hired CEO, zero named customers; professor-founder SaaS in an incumbent-held vertical. Pass.
FLAGSnone hard; rebrand churn + no visible customers after 4 years.
WHATAI-native cyber deception platform (AMaze) that deploys synthetic decoy assets across IT/OT/identity/AI environments to lure attackers, detect lateral movement, and produce forensic intelligence.
FUNDING$0 announced in Harmonic (no rounds); deck says $0 — consistent.
CLAIMSNo customers, deployments, or revenue claimed anywhere — site is product-and-hiring only. "Synthetic Cognitive Agents / Neural Echoes" is branding, not verified tech. Advisor Cassie Crossley (Schneider Electric supply-chain-security VP) is a real credibility point.
FOUNDERSNaveen Zalpuri (Founder/CEO/CTO — CMU + NIT Rourkela, ex-Oracle, Foster City CA; the deck's CMU link, https://www.linkedin.com/in/naveen-zalpuri-250ab3/) + Rajesh Nanda (CPO); ~9-person team heavily in India ops.
COMPETITIONDeception tech already consolidated — Attivo sold to SentinelOne for $616M, Illusive to Proofpoint, plus Acalvio, CounterCraft, Zscaler Deception; buyers now expect it as a platform feature.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS3.5 - experienced founder and real team, but $0 raised, zero customers, and the category's exits already happened; OT/deception gov angle is theoretical only.
FLAGSIndia-heavy ops with US shell shape — clarify entity/IP domicile before any real look; none hard-filter.
WHAT"Vista," an AI-native demand forecasting and supply planning platform producing governed, auditable, executable plans; launching into product companies and healthcare medication ops.
FUNDING$0 (Harmonic: no rounds, founded 2026, 1 headcount, Washington DC)
CLAIMS"Private early access" with unnamed design partners — UNVERIFIED; no customers, ARR, or pilots named anywhere.
FOUNDERSNot publicly identifiable — no founder named on the site, LinkedIn, or press; deck says Penn State alumnus.
COMPETITIONo9 Solutions, Kinaxis, Blue Yonder, Anaplan plus a wave of AI-native planning startups — one of the most crowded categories in enterprise software.
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.00
FOR DAXOS3.5 - one anonymous person, founded this year, zero traction, brutal category; nothing exists to rate above thesis.
FLAGSfounder anonymity (cannot diligence the single most important asset); 2026 founding, 1 headcount.
WHATAutonomous forklifts plus "Warehouse AI" execution software and digital-twin infrastructure to automate warehouses (deck adds low-light vision).
FUNDING$80K deck-self-reported; Harmonic shows $0 announced (SF, founded 2023, 24 headcount)
CLAIMSDeck $1.4M ARR — UNVERIFIED, no public trace anywhere. Harmonic's 24 headcount against $80K raised is internally inconsistent unless revenue-funded or offshore. Website is a template WordPress with SEO-filler posts for invented "2027 summits" and contradictory positioning (homepage sells driver-assistance ADAS tuning, products page sells warehouse automation) — a serious credibility mismatch for a claimed $1.4M-ARR company. No Dell/HPE or any named customer found.
FOUNDERSNot publicly identifiable — no team page, no press; deck says CMU, unverifiable.
COMPETITIONFox Robotics, Third Wave Automation, Cyngn, VisionNav on autonomous forklifts; Dematic/Toyota on the incumbent side — capital-intensive category with $100M+ funded leaders.
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.00
FOR DAXOS3.5 - the entire pitch rests on an ARR number with zero corroboration and a web presence that undercuts it; rate what exists, and what exists is unverifiable.
WHATAI teammate living in Slack, Teams, iMessage/SMS and its own apps that executes tasks across 3,200+ connected tools (email, calendar, docs, CRM) with enterprise SSO/audit/self-hosting.
FUNDING$0 — Harmonic $0 and deck $0 agree.
CLAIMSProduct is live with published pricing ($50/mo team tier) — verified by site. Enterprise-safe claims (SSO, DB-layer workspace isolation) are site-only, UNVERIFIED. No customers named anywhere.
FOUNDERSNOT publicly identifiable — no team page, no LinkedIn trail; every search collides with the unrelated YC S24 CRE company "Henry AI" (Greenwall/Pratt, $21M raised — that is NOT this company). Deck says Penn State; unverifiable.
WHATTrains one dedicated, consented, isolated AI image model per person or product so individuals and brands control training data, likeness, and commercial use.
FUNDING$0 in Harmonic; $0 per deck (consistent).
CLAIMSNo concrete traction claims anywhere — use cases described, no customers, no numbers. Nothing to verify. Incorporated Austin TX May 2026 per bizapedia.com; 3 heads per Harmonic.
FOUNDERSEmi Kolawole — ex-Google X ("Factory Whisperer"), Stanford d.school editor-in-residence, founding editor of the Washington Post Innovations section; co-founder "Sarah" not publicly identifiable. Notable but non-technical pedigree; the CMU tie is unverified.
COMPETITIONPersonal AI, Official AI, Metaphysic on likeness rights; the site itself concedes it competes with ByteDance Seedance and Google Nano Banana, which are commoditizing personalized generation for free.
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.00
FOR DAXOS3.5 - thesis excitement only (consent layer for likeness is a real future issue), which caps fit at 5 on its own; pre-product, no technical co-founder visible, foundation models eating the substrate.
FLAGSnone hard; second founder unidentifiable publicly.
11. Gemba Robotics
CO 5.50DX 3.5
SITEgembarobotics.com (parked Squarespace "under construction" page; no live site)
WHATCaptures real factory-floor data to feed simulations so manufacturers can de-risk robot deployments before buying.
FUNDING$0 (deck self-reported; not in Harmonic, enrichment scheduled)
WHATKnowledge-verification platform (FactBlock repository + DeepVerify fact-checking layer + LiveArticle) that validates AI-generated and web content against a knowledge graph.
CLAIMS"299-attorney beta, 24% time reduction, 95% continue intent" appears only in the company's own PR (prnewswire.com) — UNVERIFIED. Deck says USA; Harmonic shows Seoul HQ, 9 heads, US entity only since March 2025. Founding date conflicts: press says 2011, Harmonic says 2022 — a 15-year-old company at $1.7M would be a zombie signal; UNRESOLVED.
FOUNDERSRandy Baek — 14 yrs NLP/ML at Samsung, HBO, FactSet, 3M; solid operator, no exit.
COMPETITIONPatronus AI, Cleanlab, Vectara groundedness, plus frontier labs building verification natively; DeepStamp side competes with Cloudflare/Spawning.ai on AI-training blocking.
COMPANY RATINGBase 4.5 + EV +0.75 + Geo +0.25 (Korea, deck-USA claim rejected) + Chain 0 = FINAL 5.50
FOR DAXOS3.5 - Korean ops with a thin US shell, all traction self-reported, and LLM vendors are absorbing the verification layer.
FLAGSDeck geo misrepresents a Seoul company as USA; founding-date conflict (2011 vs 2022); zero independently verified traction.
WHATThe Aspirometer, a neck-worn triaxial accelerometer + contact microphone that noninvasively screens swallowing safety to catch silent aspiration at bedside.
FUNDING$50K (Harmonic: prizes/grants, last a $10K grant 4/2026). Deck's $275K is self-reported.
CLAIMSScience lineage is real and verified — Pitt SHRS Coyle/Sejdic NIH R01 accelerometry research, a registered NIH clinical trial (NCT06637774), and the $50K 2024 Pitt Innovation Challenge win. No FDA clearance, no customers, no revenue. Penn State tie unverified.
FOUNDERSMahoney (CEO) + Lucatorto, commercializing the Pitt lab's work; neither shows prior exit or device-company operating history.
COMPETITIONThe free nurse-administered Yale Swallow Protocol, videofluoroscopy/FEES as gold standard, and hospital budget inertia; closest device comps are small dysphagia-screening plays that have not scaled.
COMPANY RATINGBase 4.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.50
FOR DAXOS3.0 - two-person pre-FDA medtech in a narrow reimbursement-dependent niche; years and millions from revenue, no defense or infra angle.
FLAGSnone hard; solo-project risk (2 people, $50K verified) borders the student-project filter but NIH trial + incorporation keeps it scoreable.
WHATQuickChain scans GitHub/Bitbucket repos to auto-generate SBOMs, vulnerability reachability analysis, and audit-evidence packages for CMMC L2 / FedRAMP / SOC 2.
FUNDING$0 in Harmonic (0 rounds); deck says $20K, self-reported.
CLAIMSNo customer, revenue, or deployment claims anywhere; site sells a $3,000 90-day pilot. Site's own disclaimer: "not a certification or an authorization." CMU tie verified only as a Heinz IT Lab student fellowship writing a white paper for them (https://www.heinz.cmu.edu/programs/itlab). Everything else UNVERIFIED.
FOUNDERSNot publicly identified — no team page, no LinkedIn hits, no press. Harmonic: founded 2026, 2 people.
COMPETITIONEndor Labs (reachability analysis is literally their pitch, $93M+ raised), Snyk, Chainguard, Anchore; CMMC-evidence niche also served by GRC platforms (Secureframe, Vanta).
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.00
FOR DAXOS3.0 - a 2026-founded landing page with invisible founders in a category where the incumbent ships the identical feature; DIB-compliance wedge is the only interesting thread.
WHATAutonomous AI agents run multi-channel (email/SMS/voice/social) social-engineering attack simulations against employees, plus training, as a human-risk-management platform.
FUNDING$0 in Harmonic (Vyvern Corporation, founded Sep 2025, 3 people, Philadelphia); deck says $0 — consistent.
CLAIMSSite displays sector examples naming American Transmission Company, Metagenomi, Forvis Mazars — appear only on their own site, UNVERIFIED as actual customers. "3x more effective than training" and "nation-state level" are marketing, no third-party evidence. No revenue.
FOUNDERSPranav Ramesh (CEO, "avid CTF player" — first-time, Penn State per deck), Jiaming Wang (GTM), Mike Johnson (security). No exits, no notable prior operating roles surfaced.
COMPETITIONKnowBe4 (category giant), Hoxhunt, Doppel ($35M+ doing AI social-engineering defense), Jericho Security — same AI-phishing-simulation pitch, far better funded.
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 6.00
FOR DAXOS3.0 - pre-everything student team in KnowBe4's shadow; customer logos that can't be verified cut credibility rather than add it.
FLAGSnamed-customer logos unverified (possible logo inflation); none hard-filter.
9. Winnow
CO 5.50DX 3.0
SITEnone found — company could not be identified
WHATPer deck only: cleans and structures agricultural data so growers and retailers can use ChatGPT/Claude/Copilot agents against their actual operations.
FUNDING$0 (deck-self-reported; not in Harmonic — keyword search returns nothing)
CLAIMSNothing verifiable. Every public "Winnow" is a different company (UK food-waste Winnow Solutions, WinnowPro auto-dealer AI, Winnow Health, winnowdata.com real-estate CRM, winnow.so publishing). No website, no LinkedIn, no press for a CMU ag-data Winnow.
FOUNDERSUnknown — CMU per deck, unverifiable.
COMPETITIONLeaf Agriculture (ag-data API standardization is exactly this), Bushel, John Deere Operations Center APIs.
COMPANY RATINGBase 3.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 5.50 (deck-only, low confidence)
FOR DAXOS3.0 - cannot verify the company exists; nothing to underwrite until organizers provide the actual entity/site.
FLAGSunidentifiable company — no web footprint, not in Harmonic; likely pre-incorporation or renamed. Get the real name/domain from Plug and Play before any further look.
WHATAI networking layer: upload your contacts/LinkedIn graph and agents continuously score mutual fit across 6 dimensions to surface intro and deal opportunities.
CLAIMSProduct is genuinely live (web app + Chrome extension, free tier). NVIDIA Inception badge is a self-serve program, not validation. No customer or revenue claims verifiable. The "plumbing for private networks of people, companies and products" framing is bigger than what the shipped product (a networking assistant) shows.
FOUNDERSThomas Zhao (SF Bay Area) and Keno Vigil appear on LinkedIn as the Meshi AI team; backgrounds thin publicly, Penn State tie deck-only.
COMPETITIONLinkedIn itself, Clay, Boardy, Happenstance, and the swarm of AI warm-intro tools; near-zero switching costs.
COMPANY RATINGBase 3.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 5.50
FOR DAXOS3.0 - shipped product and US team, but a consumer-ish networking tool with no moat, dressed in infrastructure language; thesis over evidence.
FLAGSHarmonic shows $0 vs deck $800K (unconfirmed raise); grandiose framing vs modest shipped product.
WHAT3-person Pittsburgh "frontier AI lab" claiming a proprietary compute-cutting pretraining method (CAT) behind image (Lunara), multilingual (Selenix), and voice (MoonWave) models.
FUNDING$0 in Harmonic; bootstrapped per deck (consistent).
CLAIMSDeck's headline is CONTRADICTED by primary sources. https://huggingface.co/moonworks shows NO public models ("None public yet") — so "open releases rank among giants on Hugging Face" fails on its face. Only 2 datasets exist (~2.85K + ~2K downloads, vs the site's "25K+ downloads" claim). "Outperforms larger models like Qwen" rests on a self-published benchmark chart with no third-party evaluation. DeepSeek/Black Forest Labs analogies are aspiration, not evidence.
FOUNDERSSabit Hassan — CMU BS CS, University of Pittsburgh PhD (NLP), per https://sabithsn.github.io/; plus Yan Wang, PhD. Real research credentials, no frontier-lab or exit pedigree.
COMPETITIONDeepSeek, Black Forest Labs, Alibaba Qwen, Mistral — the exact players it invokes, each with 100-10,000x its capital and compute.
COMPANY RATINGBase 3.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 5.50
FOR DAXOS3.0 - the Adatom pattern: interesting-if-real deep tech, credentialed founder, but a capital-intensity mismatch a seed check cannot fix, and the contradicted Hugging Face claim is a direct credibility hit.
FLAGSmaterial overclaim — deck/site HF ranking and download claims contradicted by the live Hugging Face org page.
WHATGPU-based combinatorial optimization solver ("quantum-ready" branding) for strategic open-pit mine planning, sold as the Cognomine cloud product, now in beta.
FUNDINGNot in Harmonic (all three domains queued for enrichment). Deck's $400K is self-reported.
CLAIMSThe 200k-variable-in-minutes claim appears only in the deck/own materials — UNVERIFIED. "Quantum-ready" is marketing: everything runs on ordinary GPUs today. No named customers or case studies; NVIDIA Inception membership is the only third-party validation. Note the name collides with the unrelated Polkadot "Hyperbridge" crypto project — this is Hyperbridge Pty Ltd, Perth (ABN 85 695 768 828).
FOUNDERSPublicly identified only as "Donny" (built the solver) and "Ray" (commercial) on cognomine.com — no full names, resumes, or CMU link findable. That opacity is itself a finding.
COMPETITIONDatamine, Deswik, Hexagon MinePlan, Minemax, MiningMath — entrenched incumbents with decades of site relationships.
COMPANY RATINGBase 4.00 + EV +1.00 + Geo +0.25 + Chain 0 = FINAL 5.25
FOR DAXOS3.0 - anonymous founders, Australia, beta product, no traction, incumbent-owned vertical. Pass.
FLAGSfounders not publicly identifiable; every technical claim single-sourced to own site.
WHATNo-code, hardware-agnostic computer vision platform: industrial users train detection models in the cloud and deploy to existing cameras/edge devices without an ML team.
FUNDING$654K total, last round $538K seed May 2024 led by BSV Ventures (Harmonic; legal entity Trashify Tech, Tallinn, 4 people, founded 2021). Deck's EUR 260K understates the Harmonic figure.
CLAIMSTartu waste-facility pilot appears real (covered on their own channels only). Platform breadth claims are site-only. UNVERIFIED beyond company materials. Note: MAKRR is a rebrand/pivot of Trashify Tech (waste-tracking CV) into generic no-code CV.
WHATAgent optimization/eval platform that sweeps configurations (models, prompts, params) against accuracy and cost metrics so teams ship agents with proof of testing.
CLAIMSSite is a thin single page ("Optimize your AI agent and prove it"); no customers, benchmarks, or revenue visible anywhere — all deck claims UNVERIFIED.
FOUNDERSLikely Gil Taran (LinkedIn shows "Founder"; ex-CEO of iCarnegie, COO of CMU School of Computer Science) — genuine CMU tie but education-executive pedigree, not AI product; company is publicly hiring a CTO, i.e. no technical co-founder (linkedin job). Founder identity not fully confirmed.
COMPETITIONBraintrust, LangSmith, Arize, Patronus, Galileo, Vellum, HoneyHive — one of the most crowded and best-funded categories in AI tooling.
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +0.25 + Chain 0 = FINAL 5.25
FOR DAXOS3.0 - 5-person Tel Aviv team with no technical co-founder entering a space where incumbents have raised $50M+; nothing that exists to underwrite.
WHATDeck says applied-AI workflow automation and custom software; the only findable Zelena (zelena.ai) says AI-driven data-center efficiency/cooling — the two descriptions do not match.
CLAIMSNo customers, no product evidence, no CMU connection findable anywhere. Everything is UNVERIFIED because the company itself can't be firmly identified — no Pittsburgh/CMU "Zelena" exists in Harmonic, LinkedIn, or search.
FOUNDERSIf zelena.ai: Lionel Selwood Jr., ex-CEO Romeo Power (NYSE via SPAC; stock collapsed, sold to Nikola, which went bankrupt) — real operator experience, mixed outcome record, no visible CMU tie.
COMPETITIONIf automation-agency shape: every AI consultancy plus Automat, Zapier; if data-center: Verrus, Phaidra, incumbents' own DCIM AI.
COMPANY RATINGBase 3.5 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 5.50
FOR DAXOS2.5 - cannot confirm what the company even is; 2-person services-shaped entity with positioning that changes by source.
FLAGSidentity mismatch between deck blurb and only findable website — needs deck contact info to disambiguate before any further look.
WHATLightweight edge-optimized vision-AI models (defect detection, OCR, tracking) deployed fully on-premise for manufacturing, utilities, retail, and defence customers in India.
FUNDING$200K pre-seed Jul 2024 (Harmonic; Bangalore, 12 people, founded 2023). Deck's $550K conflicts with Harmonic's $200K; treat the deck figure as self-reported and unconfirmed.
CLAIMS"30+ enterprise deployments" appears ONLY in the deck; site shows customer logos (Polaris, Genus, IPEC, Paras) with no case studies. Nasscom Deeptech recognition and NVIDIA/AWS affiliations are program memberships, not traction. UNVERIFIED.
FOUNDERSNot publicly identifiable; no founder named on the site or in press. CMU tie is deck-only. That alone is a diligence problem.
COMPETITIONMatroid, viso.ai, Landing AI globally; in India, Detect Technologies and a deep bench of CV services shops; on-prem/edge privacy angle is not defensible.
COMPANY RATINGBase 4.0 + EV +1.00 + Geo +0.00 + Chain 0 = FINAL 5.00
FOR DAXOS2.5 - India-based, services-shaped, headline traction and raise both unverifiable, no US/government pathway.
FLAGSDeck raise ($550K) vs Harmonic ($200K) discrepancy; founders anonymous in all public materials; headline deployment claim unverified.
WHAT"Secure operating system for AI agents" (Matrix architecture) claiming modular agent-to-enterprise connectivity and up to 90% token-usage reduction.
CLAIMSThe 90% token-reduction number and "world's first agentic stack" are site-only, UNVERIFIED. The drones/robotics/hardware angle in the deck appears NOWHERE on the actual site, which is a generic enterprise-agent landing page. No customers, no pilots named anywhere.
FOUNDERSNot publicly identifiable; no founder named on site, LinkedIn, or press. Penn State tie is deck-only.
COMPETITIONPalo Alto Networks agentic-security push, Entro, Arcade.dev, Cloudflare Mesh, plus every funded agent-infra startup; the most crowded category in the batch.
COMPANY RATINGBase 3.0 + EV +1.00 + Geo +1.00 + Chain 0 = FINAL 5.00
FOR DAXOS2.5 - one anonymous person with a landing page in the most contested space in software; nothing to underwrite.
FLAGSSolo, anonymous founder; deck positioning (drones/robotics) contradicts the live product page; borderline solo-project profile.
Featured: Lone Star Lunar
Identity first: this is NOT a separate company. lonestarlunar.com IS Lonestar Data Holdings Inc. Harmonic resolves the domain straight to Lonestar Data Holdings (lonestar.space), lists lonestarlunar.com as a domain alias, and the exec emails are [email protected] / [email protected]. The site itself is titled "Lonestar Space" and shows the known Lonestar team. There is no second "Lone Star Lunar" entity. Everything below is Lonestar Data Holdings.
What they do and sell (per https://www.lonestarlunar.com/): "StarVaults," sovereign space-based data storage plus AI inference platforms, solar powered and radiatively cooled, launching "to all Earth orbits and to the Moon." Pitch is resilience-as-a-service and data sovereignty via space law for governments and regulated industries. They claim four test flights (two ISS, two lunar, the lunar ones as payloads on Intuitive Machines IM-1 Feb 2024 and IM-2 Feb 2025), next mission April 2027, "actively booking capacity." Recent PR: "Sovereign Language Models from orbital data centers," a "Data Embassy," and a July 2026 release about building "the first Dyson Swarm" with new investor 012T Capital. That last one is hype drift.
Founders and pedigree. Real gray-hair team, 19 heads, St. Petersburg FL, founded 2018. Chris Stott (Executive Chair, founder) built ManSat, which he claims was the largest commercial satellite-spectrum provider; Steve Eisele (CEO) led Virgin Orbit global sales through IPO; Matt Friters (CTO/COO) US Army/Space Force cyber; Jim Burns-Montante (Chief Engineer) ex-Draper/Honeywell space electronics; Carol Goldstein (CFO) built satellite finance units at Morgan Stanley/ING/ABN AMRO. Strong domain pedigree, no prior venture-scale exit among them.
Funding (Harmonic): $14.87M total across 4 rounds. Last round: $2.4M, tagged PRE_SEED, closed 2026-08-18; Harmonic-estimated valuation ~$20M. Investors: Scout Ventures, Seldor, 2Future, Atypical, IronGate, The Veteran Fund, Gaingels, Kitty Hawk, angels. That trajectory is the tell: an 8-year-old company in a capex-heavy category still raising $2.4M pre-seed-shaped rounds, with roughly a 1.3x total-raise-to-valuation ratio. Note the total sits $130K under the >$15M hard filter.
Government/NASA/defense traction. Site claims work for "National, State, and City Governments" and the ISS/lunar flights. The lunar flights are real (widely covered; they carried test data including for the State of Florida and Isle of Man), but they rode as customers on NASA CLPS landers, not as NASA contractors — and IM-2 tipped over on landing, cutting the mission short. No SBIR/DoD contract surfaced. All revenue, gov-customer, and "booking capacity" claims appear only on their own site: UNVERIFIED.
Competition. Brutal and better funded: Starcloud (NVIDIA-backed orbital data centers), Axiom Space's orbital data center line, Ramon.Space, plus Sophia Space (already in the Daxos pipeline at 7.5), and terrestrial sovereign-cloud offerings from AWS/Azure that solve the compliance problem without launch costs. The physics ("free cooling," bandwidth, radiation) is contested for the whole category.
Rating. Anchors: Sophia Space 7.5 (7-exit founder, NVIDIA-led SAFE, MIT Lincoln Labs customer), BESXAR 7.5 (SpaceX deal + DoD contract), Satlyt 6.75 (1 confirmed on-orbit deployment, NASA STTR), Cosmic Shielding 6.71 ($2M real revenue). Lonestar has more flight heritage than Satlyt but no verified revenue or gov contract like BESXAR/Cosmic, plus the 8-years-to-pre-seed velocity problem and marketing overreach. Base 6.0.
FOR DAXOS. The $2.4M round size means a $100-250K check is meaningful and accessible, and Daxos loves gov + space picks-and-shovels. But gov traction is unverified, four rounds over eight years implies a tired cap table, the category leaders are 10x better capitalized, and Sophia Space is a cleaner shape already in-pipeline. Meeting-only: 5.0.
WHAT: Sovereign space-based data storage + AI ("StarVaults") in Earth and lunar orbit for governments and regulated industries; 4 test flights flown, next mission April 2027.
FUNDING: $14.87M total, last $2.4M pre-seed 8/18/26, ~$20M est. val (Harmonic)
CLAIMS: Gov customers, revenue, and 2027 capacity bookings are site-only UNVERIFIED; lunar flights real but as rideshare payloads (IM-2 lander tipped over); "Dyson Swarm" PR is hype.
FOUNDERS: Chris Stott (Exec Chair, ManSat founder), Steve Eisele (CEO, ex-Virgin Orbit sales lead), Jim Burns-Montante (Chief Eng, ex-Draper/Honeywell), Carol Goldstein (CFO, ex-Morgan Stanley/ING)
COMPETITION: Starcloud (NVIDIA-backed), Axiom orbital data centers, Sophia Space (Daxos pipeline 7.5), Ramon.Space, terrestrial sovereign cloud.
COMPANY RATING: Base 6.0 + EV +0.00 + Geo +1.00 + Chain 0 = FINAL 7.00
FOR DAXOS: 5.0 - accessible check size and real flight heritage, but unverified gov/revenue claims, 8-year pre-seed velocity, tired cap table, and better-capitalized rivals cap it at take-the-meeting.
FLAGS: Same company as Lonestar Data Holdings (not a distinct startup); $14.87M raised is $130K under the >$15M hard filter; 2018 vintage still at pre-seed; Dyson Swarm hype drift.