Kairos — Proposal DRAFT v1

ARCHIVED 2026-08-10 (compression session) from 10-projects/ root. Superseded 2026-07-23 by Kairos — Proposal v2 (2026-07-23) (full rewrite: monthly retainer, Kairos name, EN-only). Kept as the July 6 historical record of the bilingual SWAT proposal. Original deploy history recoverable in Git under ~/Claude/Projects/BRAND BOOK v2.0/deploy/swat/.

Draft v1 of the document Mysterium received — written for Šaras to take over, cut, and own (his voice message asked exactly for this: budget, timeline, goals, working mode). ⟦…⟧ = open call, decided at the Lee+Šaras sit-down. English default; flips to Lithuanian in an hour if Šaras prefers. This note is the canonical record of the July 6 bilingual SWAT proposal; the former SWAT address now redirects to current Kairos. Public immutable history starts with the later Kairos v1.0.0 release from July 24.

SWAT Unit — 30-Day Proposal (Mysterium Network)

What this is

A two-man special-operations unit — Šarūnas + Linas — hunting disproportionate ROI for Mysterium Network. Not consultants, not headcount: a self-contained team with its own tools, its own AI leverage, and one mandate — find the shortest credible path toward break-even and put visible motion on it within 30 days.

The mission needs this. A sovereign network that loses money ends up serving someone else's values — profit is what keeps the mission sovereign. And the money is demonstrably there: Tesonet's group runs the same two markets — VPN and proxies — from the same city at over €1B in annual revenue. The market pays. The question is only where Mysterium's shortest path into that money is.

The 30-day bet

Weeks 1–2 — the Commercial X-ray. One page: where money actually enters and leaks. Revenue and cost per product line (dVPN / B2B proxy / agent-access work), distance to break-even, and the top 3 levers ranked by effort-to-impact. If this page already exists inside Mysterium, we verify it and go straight to the levers — faster. If it doesn't, you have it in two weeks. Either way, day 14 ends with it on the table.

Weeks 3–4 — move the #1 lever. Chosen from the X-ray, not preassigned. We launch it and attach a number to it: sales meetings booked with serious buyers, € value of deals set in motion, or the revenue lift from a pricing change — whichever the lever produces.

Day 30 — the verdict, in writing: X-ray delivered, top lever in motion, early signal measured. Either the numbers argue for continuing, or we shake hands and you keep the X-ray.

How we work

Two complementary modes, deliberately paired. Linas goes deep — systematic research that finds where the asymmetry actually is. Šarūnas launches torpedoes — fast live tests that create an opportunity and scan the response. The research aims the torpedoes; the torpedoes validate the research.

We operate ⟦outside the team, with named data-access points⟧ — that's what keeps a SWAT unit fast. One written update to Robertas per week — decisions made, numbers moved, what's next — plus a shared chat as an open channel for anything live in between, if preferred. No meetings-as-progress, though. (Ro asked for the shared chat by text, 07-06.)

What we need from day 1

The X-ray is only as honest as its inputs. We need access to whoever holds: revenue by product line (trailing 12 months), infrastructure and team cost per line, the B2B client list with revenue concentration, pricing sheets and take-rate by geography, and current burn. One person who can answer money questions within 48 hours is the floor — necessary, not sufficient (Lee's correction, 2026-07-23: the org's distributed, undocumented knowledge is a primary research asset). Alongside the money line: listening access to the team — Slack presence and informal coffee conversations, opt-in, zero meetings added.

Budget & terms

• €10,000 for the 30 days, for the unit, pre-tax, all-inclusive of our own tools and AI infrastructure — each of us covers his own toolkit. The torpedoes are deliberately designed labor-heavy, cash-light; any experiment that needs real cash (paid traffic, bigger test accounts) is proposed case-by-case with a budget and the expected signal, and runs only if Mysterium approves and funds it.
• Invoiced in halves: 50% on signing, 50% at day 30 — standard for short engagements.
• Capacity: ⟦15–25⟧ hours/week each. Surgical, not resident.

After the trial

If the day-30 numbers argue for it: a 90-day engagement on the validated lever, with a target number agreed before it starts. And one sentence for the long game: if a lever we build produces sustained recurring revenue above ⟦€X⟧/month, we open the upside conversation — revenue share, equity, or token, whatever fits.

How this could fail — named upfront

1. Data access stalls. It's the precondition. If nothing moves by day 3 we escalate to Robertas; if nothing by day 7, the clock pauses — the trial doesn't burn while blocked.
2. The X-ray finds no fast lever. Then the written verdict says exactly that, you keep the X-ray, and we part at day 30. That's the trial working, not failing.
3. A lever works but nobody inside can carry it after us. The Monday Updates and a handover note make the motion repeatable; the 90-day phase names an internal owner from day one.
4. Scope creep eats focus. Hours are capped and scope is written. New asks become proposals, not silent additions.

Start: ⟦date⟧.

Šarūnas Savickas · Linas Matulis

Annex — first torpedo candidates (illustrative, X-ray decides)

T1 (rebuilt from evidence, 07-06) · Crypto-native wedge probe. The one space the market map found uncontested: protocol-level, no-KYC, crypto-paid access for decentralized and crypto-agent builders — a buyer no incumbent's compliance posture lets it serve, native to Mysterium's architecture. Thin self-serve storefront + outbound to ~20 named crypto-agent teams; an agents-pay-for-bandwidth demo doubles as proof. Cash ≤€100. (Generic "AI agent access" checked and rejected — kill-list, Annex G.)

T2 · Pricing & margin probe. Now armed: Mysterium's supply already retails publicly at $1.40–2.59/GB via GoProxies — the wholesale/retail gap is the first number the X-ray pulls. If premium geos under-priced: quiet repricing on a subset, measure uptake/churn. Cash €0.

T3 (half-done already) · Trust-surface sprint. Desk research found the diligence gap: competitors lead with certificates and named entities; the current B2B surface names no legal entity, publishes no list price. Week-one fix: published pricing, entity, trust page. Competitor price matrix already compiled, free. Cash €0.

(T4 — Šaras's quick-market test idea from his voice message: parked until he defines it.)

Annex B — the 30 days, day by day

• Day 1–3 · Wire the access. Money-person named, first data pull, one 45-minute kickoff call — the only standing meeting of the month.
• Day 4–10 · Build the X-ray. Revenue and cost per product line, distance to break-even; competitor pricing gap analysis (already started — Annex E).
• Day 11–14 · X-ray on the table. One page + top-3 levers ranked. Lever choice locked with Robertas, async, within 24h.
• Day 15–25 · Torpedo runs. The #1 lever in motion; Monday Updates carry the numbers.
• Day 26–30 · The verdict. Signal measured, written recommendation: continue with a 90-day plan and target, or stop — X-ray stays either way.

Annex C — the X-ray skeleton (the page you get by day 14)

Annex D — data request, ready to forward

Written so Robertas can forward it as-is before Tuesday:

Subject: SWAT unit — data access [Name], Šarūnas and Linas start a 30-day commercial X-ray on ⟦date⟧. Please give them read access to (or exports of): 1) revenue by product line, trailing 12 months; 2) infrastructure + team cost per product line (approximate is fine); 3) the B2B client list with monthly revenue per client; 4) pricing sheets / take-rate by geography; 5) monthly burn. First call is 30 minutes, they drive. — R.

Annex E — pricing landscape (public list prices, desk-checked 2026-07-05)

Market splits enterprise ($8–12/GB) vs mid-market ($3–6/GB); per-success unblocking pricing sits above both. To verify against Mysterium's actual realized prices in week 1. Sources: brightdata.com pricing, use-apify.com pricing guide, aimultiple.com proxy-pricing, proxyhorizon.com, docs.mysterium.network/faq — all public list prices, 2026-07-05.

Annex F — the Monday Update (what Robertas receives, every week)

One page, five fixed lines, Monday morning, written: 1) decisions made · 2) numbers moved vs last week · 3) next 7 days · 4) blockers + who unblocks them · 5) spend. No meeting attached.

Annex G — work already done (before day one, €0; added 07-06)

Outside-in half of the X-ray, run on public sources before proposing anything (full sourced files: 30-resources/mysterium/ — outside-in-xray, market-map, token-reanimation-precedents, alpha-hunt): - The B2B door: mysterium.network has no buyer path; public records read GoProxies as a third-party partner retailing Mysterium supply at $1.40–2.59/GB (Lee's internal-brand prior didn't survive primary sources; Šaras arbitrates). Day-one question: who captures that margin — two prepared branches: own the retail layer, or deliberately multiply wholesale storefronts. - The narrative gap — CORRECTED 2026-07-18: not silence — fragmentation. Four content engines run unlinked (near-daily VPN news desk, GoProxies research arm with press pickup, thin MystNodes blog, dead legacy blog); the mothership links none of them. The lever is consolidation + routing, not resurrection; Sept 15, 2026 Cloudflare agent-block still gives it a dated clock. Historical note: the July 6 draft above carries the superseded "blog silent ~3 years" line; the later Kairos proposal does not (public-surface-sweep-2026-07-18). - Engineering signal: public GitHub all-dVPN; no visible B2B/agent-access work — verify internally day one. - Token, measured: revival from MYST's zone ≈ low-single-digit base rate; second-order amplifier after operational wins, never the plan. - Kill-list (checked, rejected): generic AI-agent outbound (vendor-dense, freshly funded) · head-on proxy sales vs Oxylabs (the 397-person Vilnius neighbor, ~$43.7M ARR) · token-first revival · "ethical supply" as legal moat (Bright Data court wins) · consumer dVPN growth race.

(Internal only, NOT on the deployed page: the commercial-director situation — exists, good, dimmed by grunt work; name TBC via Lee. Levers S1–S5 full form: 30-resources/mysterium/alpha-hunt.md.)

Open calls for the sit-down

⟦working mode⟧ outside vs embedded — Lee leans outside · ⟦hours⟧ 15–25/wk — Lee's number to fix · ⟦spend⟧ — own tools each; Lee's hard cap €300/mo total incl. Claude Max 20x; Šaras sets his own; real-cash experiments → Mysterium-funded on approval · ⟦jackpot threshold €X⟧ — Šaras calibrates · ⟦start date⟧ · ⟦language⟧ EN vs LT · ⟦advance framing⟧ — Šaras runs the money talk · whether Ieva receives it directly or via Ro.