☿ Kairos War Room
← Back to master state
Gated source note · private

Kairos — Research Dossier

Deep research for the Mysterium meeting prep. Two agents: one on the company, one on the market terrain. Confidence tags and source URLs preserved verbatim. Treat any company self-reported figure as marketing until independently corroborated. Key gaps flagged at the end of each section.

Headline (the honest reframe)

My earlier one-pager was too rosy. The research says: Mysterium is a real, still-technically-active 9-year-old project (GitHub commits through today) undergoing a genuine but small B2B proxy pivot (GoProxies is real) — sitting on a micro-cap token in acute decline (~$0.09, market cap under $3M, −60% yr, −31% in 30 days, ranked outside the top 1,500), with node count flat since 2021, zero disclosed revenue or VC funding since the 2017 ICO (~€14.4M raised), a publicly-named team of two people, and a for-profit Swiss holding company (BlockDev AG) behind non-profit "foundation" marketing. Meanwhile a younger competitor, Grass, has out-executed it on exactly the AI-data play (Grass market cap ~$304M — ~100× Mysterium's — 2.5M claimed nodes vs Mysterium's ~22–31k).

Name to verify with Lee: Mysterium's co-founder & CEO is Robertas Visinskis. Is that Lee's "Ro," or a different Robertas? Changes the entire read of the meeting.

COMPANY DOSSIER (agent 1)

1. Founding & Origin

Conceived Oct–Dec 2016 by Lithuanian developer Robertas Visinskis (proposed on Bitcointalk), 6,000 ETH seed on 2016-12-16. Three founders consistently named: Robertas Visinskis (BA CS Kaunas Univ. of Technology; ex-CEO/CTO Trustribe, ex-CTO Meta Group), Valdas Petrulis (Lead Software Dev), Paulius Mozuras (Lead Ethereum Dev; also co-founded the separate consumer "Mysterium VPN" entity in 2021). Country of origin is mixed, not a clean Lithuania story: dev team Lithuanian; operating legal entity is BlockDev AG, a Swiss for-profit AG in Zug (CHE-464.393.106). Material discrepancy to flag: markets itself with foundation/non-profit language while legally a for-profit holding company; no "Mysterium Foundation" entity independently verified, no Cayman entity found.

2. ICO / Funding

Token sale closed 2017-05-30, reportedly sold out in 45 min. Raised ~14M CHF / €14,358,975 (~$13.3–14.4M; noise-band). ICO price ~$0.77/MYST. KR1 named as ICO-era backer. No further disclosed institutional funding in nine years — funded almost entirely by the 2017 ICO. PitchBook lists "privately held, no institutional backing."

3. Pivot Timeline

2016–17 founding/ICO/whitepaper (consumer dVPN) · 2017–18 consumer apps · 2019–20 B2B partnerships, Hermes L2/Polygon payment rail, Raspberry Pi nodes · 2021-06 "Tokenomics 2.0" (PARSIQ's IQ Protocol, subscription/rental over staking) · 2021-11 mainnet (Ethereum + Polygon) after ~4yr testnet · 2022 node count crossed 20k · 2022-11 Delegation Pool + MEXC trade-mining · 2023 "Mysterium 2.0" + Necto Labs ($3M seed fund/incubator, up to $50k/team) · main blog goes quiet after ~April 2023 · 2024-08 Hivello partnership (DePIN aggregator integrates MystNodes) · 2025-01 consumer VPN entity → MN Intelligence UAB (Vilnius) · GoProxies partnership (residential-proxy B2B channel — CONFIRMED real, terms/date unknown, 403 on direct fetch) · 2026-03 MystNodes points-based Rewards System · 2026 GitHub daily commits (engineering alive).

4. Business Model

Buyers pay crypto → Hermes smart-contract channel (Polygon L2) → node runners earn MYST → Mysterium/BlockDev AG takes a 20% service fee on settlement (CONFIRMED, official docs + MystNodes help center). B2B/GoProxies pricing not public. No disclosed revenue, gross volume, or profitability anywhere, 2024–2026 — a real, material unknown.

5. Tokenomics

MYST = payment currency + staking/rental collateral + Delegation-Pool passive yield; no on-chain governance found. "Tokenomics 2.0" (2021 post cosmetically "updated Dec 2025" — staleness red flag) via IQ Protocol: stake MYST (~1 MYST ≈ 0.01 DAI/day earning cap), non-runners deposit into Delegation Pool for yield, runners short on MYST rent from the pool. Price 2026-07-03 (CoinGecko): $0.0903, mkt cap ~$2.93M, −15.5% 24h / −31.3% 30d / −60.7% yr; ~97–98% below Jan-2018 ATH. Supply ≈ 32.43M (circulating≈max). Listed on Coinbase + MEXC/HitBTC/Uniswap/QuickSwap; not on Binance. Volume ~$24–94k/24h — illiquid micro-cap.

6. People & Org

CEO: Robertas Visinskis (LIKELY still, no fresh 2026 primary confirmation). Team page names only two people ("RobertasVis — Founder", "Gitas — Director"). Team size ~25 (PitchBook estimate). Operating co: BlockDev AG (Zug, CH). Consumer VPN: MN Intelligence UAB (Vilnius). Only 2 Glassdoor reviews.

7. Current State

Alive: GitHub commits through 2026-07-03 (node repo 1,209★, same-day pushes) — strongest positive signal. Self-reported: ~31,547 IPs / 135 countries / $568k cumulative "network earnings" / 41k Twitter / 10k Discord. Rewards System shipped Mar 2026. Struggling: MYST −60.7% yr / −31.3% in 30d (acute, not just crypto-winter); no mainstream crypto press coverage found for 2025–2026; main blog stale since ~Apr 2023; no disclosed revenue/VC in 9 years; node count varies 7.5k–31.5k across the company's own properties with no timestamps (surface the inconsistency, don't cherry-pick).

8. What only an insider knows

(1) Actual revenue/profitability — is it living on ICO-treasury runway? (2) Real current node count (four conflicting owned figures). (3) Which "Tokenomics 2.0" mechanics actually run today. (4) GoProxies deal terms (date, rev-share, exclusivity, volume). (5) Whether MYST holders have any claim on BlockDev AG equity/profit, or the 20% fee is pure corporate revenue. (6) Whether Visinskis is still operationally CEO. (7) Why crypto trade press went silent.

TERRAIN DOSSIER (agent 2)

1. Residential-proxy / web-data industry

Market-sizing is mush — estimates range from ~$1B to mid-teens $B for the same year depending on firm/definition (disagree >10×). Frame as a range, never a single figure. Best non-vendor source: Proxyway's annual Proxy Market Research report (fetch directly). Commercial model: tiered GB subscriptions + pay-per-GB + enterprise custom. Bright Data ~$5–8/GB; Oxylabs Micro $99/mo @ $7.75/GB. Lithuanian concentration CONFIRMED and strong: Vilnius hosts IPRoyal, Oxylabs, Decodo (ex-Smartproxy) — "three of five top positions"; much of it traces to Tesonet (Vilnius venture builder behind Oxylabs, Surfshark, Nord Security, Hostinger, Decodo; group revenue reportedly >€1B). You are standing in the global capital of this industry. - Bright Data (Israeli, category leader, ex-Luminati): ARR crossed $300M → targeting $400M, >50% YoY, explicitly AI-data-driven; serves 14 of top-20 LLM labs. CEO: "Fresh data is the oxygen of AI... decays within hours." - Oxylabs (Vilnius, bootstrapped, ~397 staff, CEO Julius Černiauskas; acquired ScrapingBee Jun 2025; 175M+ IPs claimed). Smartproxy→Decodo (rebrand Apr 2025 after a trademark dispute with Bright Data over the word "Smart"). IPRoyal (Vilnius, 2020, CEO Karolis Toleikis, revenue ~3× YoY). NetNut (Israeli), SOAX (London).

2. How scraping / anti-bot actually works

Modern anti-bot (Cloudflare, DataDome, Akamai, HUMAN/PerimeterX, Kasada) stacks independent layers: IP reputation (datacenter ranges pre-blacklisted via ASN) · HTTP/TLS fingerprinting (JA3) · behavioral heuristics (timing/navigation) · JS/browser fingerprinting (canvas/WebGL) · CAPTCHA escalation. Why residential beats datacenter: residential IPs are real ISP household allocations, indistinguishable at the network layer from a genuine consumer; datacenter ranges are enumerable and pre-blocked. Residential ~92–98% success vs 65–80% datacenter (vendor-influenced, directionally true). Key nuance for the deck: "residential proxies alone won't save you" — modern anti-bot is defense-in-depth; IP rotation is necessary-but-not-sufficient. CAPTCHA-solving now bundled with proxy networks (2Captcha: €1/1000 solves). Use cases: geo-unblocking (streaming licensing), sneaker bots, price/data intelligence.

3. "Ethically sourced" controversies

Hola VPN → Luminati (2015): Hola secretly resold 46M+ users' bandwidth via sibling Luminati (→ Bright Data) with no opt-out; surfaced when a spammer used the 9M-IP botnet to DDoS 8chan. The market leader's origin story is an undisclosed botnet-resale op. Bright Data legal wins (2024): Meta (dropped Feb 2024) and X (dismissed May 2024) both lost scraping suits — emerging precedent: scraping public, non-logged-in data is generally defensible in US courts; behind-login is where liability attaches. Proxyware malware (Trend Micro 2023, primary research): Honeygain/PacketStream/IPRoyal Pawns exit nodes facilitated fraud, phishing, click-fraud, SQLi, PII scraping through unwitting home IPs — the "consenting" sharer bears legal exposure. Botnets becoming residential proxies (2024–26): FBI 911 S5 takedown (2024); Aisuru botnet pivoted 700k devices to residential proxies (Krebs, 2025); BADBOX 2.0; DSLRoot ("legal botnet" consent model — direct comparable to Mysterium/Grass). GDPR: proxy buyers can inherit Art. 28 liability if IP-owner consent was inadequate (theory/exposure, no named enforcement case found).

4. AI training-data demand wave

Bright Data attributes its growth wholly to AI data. Mechanism connecting AI → proxy demand: as more publishers block AI crawlers (GPTBot in ~5.5% of Cloudflare robots.txt blocks; ClaudeBot ~4.9%), AI companies route around blocks via residential proxies, not naked datacenter IPs. Counter-current = licensing: Reddit↔Google $60M/yr, Reddit↔OpenAI ~$70M/yr, Reddit disclosed $203M aggregate licensing value (CONFIRMED, TechCrunch). Needs verification: claims that Reddit sued Anthropic (Jun 2025) & Perplexity (Oct 2025) and a Cloudflare "pay-per-crawl" launch came from a single non-primary blog.

5. Decentralized/crypto competitors

Grass (Wynd Network) — the central comparable: GRASS mkt cap ~$304M (2026-07-02); 2.5M nodes / 8.5M users claimed; funding pre-seed $1M (2023) → $3.5M seed (Polychain/Tribe, Dec 2023) → Series A (Hack VC, Sept 2024, implied ~$1B valuation); token launched Oct 2024 across 8 major exchanges simultaneously; sells real-time residential-IP-routed web data explicitly to AI labs, pitched as the "ethical" alternative to proxyware. Co-founder Radonjic: "your bandwidth has been stolen by companies that sneak software into your free apps." Doesn't disclose downstream buyers. Others: Orchid (stagnant since 2022), Sentinel (rebranded $DVPN→$P2P, added "Scout" AI-data layer), Deeper Network (hardware router model), Nodepay (2M+ users, AI-bandwidth thesis), Gradient ($10M seed, 1.6M nodes, no token yet).

Scale comparison: Grass ~$304M mkt cap / 2.5M nodes · Gradient 1.6M nodes / $10M seed · Mysterium ~$3M mkt cap / ~22–31k nodes / €14.4M ICO (2017), no VC since. Mysterium's node count is ~2 orders of magnitude below Grass; market cap ~1% of Grass. The single most important comparative data point.

6. White space / strategic implications

Browser rendering of the War Room source library. The vault remains canonical.