Best tracks to revenue, per-project plays, data-transfer upgrades · content/money_and_uses.md
Money & Use-Case Strategy — where Qals actually pays off for qalarc
Date: 2026-09-06 · Question: "What is the best use case for me here and best track to making money? How should I use it across my projects? Could elements be useful for data transfer itself — SSH, Tailscale, etc?"
1. The money tracks, ranked by (speed to revenue × fit × risk)
🥇 Track 1 — Verified AI work product ("provenance premium") — weeks, no token, no licence
Every AI buyer is starting to ask for proof of how outputs were made; almost nobody can provide it. You own the full machinery as of today: ComputeReceipt NFTs + DataAnchors + Qal ID signatures = a verifiable provenance certificate per deliverable (model hash, inputs, agent, timestamp, unmodified-since proof).
- Productise: "Every qalarc deliverable ships with a tamper-evident provenance certificate" — a premium tier on existing qalarc.ai contracts.
- Flagship fit — endispute.com.au: evidence integrity is the product. Anchor dispute-evidence hashes + chain-of-custody; settle via escrowed
Holds (the dispute platform and the credit system were made for each other: parties fund a Hold, winner receives settlement, receipt notarised). - Why first: sells into revenue you already have; zero regulatory surface (no token involvement); the chain runs privately.
🥈 Track 2 — The internal cost engine — immediate
One metering/credit ledger across all apps:
- Per-client, per-agent margin visibility (most multi-app shops can't see this today — it's where money quietly leaks).
- Chain-enforced agent caps make the "compromised/looping agent drains the API card" failure structurally impossible (loopd demonstrates this now; qal_credit enforces it on-chain).
- Fleet GPU time properly costed → priced for external sale later.
- Runs today as loopd/; graduates to Qalnet with zero redesign (its JSON ledger is the genesis import).
🥉 Track 3 — Qal Compute externally — months
Sell spare GPU-hours (superlocal's 8060S + minirig) at the 5–20× markup that hosted inference APIs charge over raw compute. Internal first (Track 2 proves the settlement rails), external providers later. Receivables vault (Qalx) differentiates: job escrows are on-chain, so invoice-financing against them is underwritable.
🏅 Track 4 — The token itself — years, deliberately gated
Honest framing: QALS/B-QALS is not the money-maker; it's the moat and the plumbing. Float/redemption fees matter only at scale. Public transferability waits for the AUSTRAC/VASP path (06_bank_exchange/). Anyone promising faster token profits is selling IOTA's 2017 — we researched how that ends (−99.1%).
The one-line strategy: sell the proof (Track 1), run on the rails (Track 2), rent the machines (Track 3), and let the token stay boring until lawyers say otherwise (Track 4).
2. Per-project playbook
| Project | Play | First move |
|---|---|---|
| endispute.com.au | Evidence anchoring + chain-of-custody + escrowed settlements | Anchor one real evidence file (the loop-spec anchor tx BthvVwJ4… is the template); pitch "notarised dispute files" to one client |
| qalarc.ai | Provenance premium tier; metered client billing; agent wallets | Add provenance certificate as a line item on the next proposal |
| tradez.au | Signal receipts: "generated by model X on data Y at Z, unmodified" — auditable signal history; invoice escrow | Anchor each daily signal digest; show clients the verifier |
| agent hub | Caps, receipts, revocation drills for every agent | Move hub agents onto loopd caps this week |
| doof.ing | Doofs NFTs + micro-tips (gasless via Qal Pass) | Wait for audience; brand kit ready |
| The Fleet | Internal compute settlement → external marketplace | Meter minirig jobs through loopd |
| qalarc.com itself | The QALS wiki = the sales asset | It's live — link it from the main nav |
3. Data transfer: SSH, Tailscale & friends — upgrade, not replace
Starfish is consensus, not networking. Keep WireGuard/Tailscale for transport, SSH for shells. The chain adds the layer tunnels can't: verifiability and accountability.
| Integration | What it does | Effort |
|---|---|---|
| Transfer receipts | every rsync/scp/sync between fleet machines emits a DataAnchor (sha256 + src + dst + time) → silent-corruption & tamper detection after the fact, wire-independent | days — wrap your sync scripts |
| Infra-state notarisation | anchor hashes of authorized_keys, Tailscale ACLs, host-key fingerprints → immutable "who changed which key, when" |
days — cron + anchor |
| Identity layering | Tailscale answers which node; Qal ID answers what it may do & spend — capability credentials bound to devices, revocable in ~2s even inside the mesh | Phase A of Qal ID |
| Streams-style channels | payloads stay on the Tailscale mesh; hash-anchors + per-recipient key grants on-chain → selective disclosure without a data lake | the qal_data pattern, extended |
| Backup integrity proofs | nightly Merkle root of backups anchored → provable intact-and-unmodified over time | days |
Rule of thumb: bytes ride Tailscale; hashes ride Qalnet; identity rides Qal ID.
4. What NOT to do (money-losing traps)
- Don't launch a public token or DEX liquidity before the legal path — it converts your best asset (the honest prepaid framing) into a securities problem.
- Don't build exchange features before Track 1 revenue exists — Qalx Phase B is gated on volume for a reason.
- Don't put bulk data on-chain (storage deposits make it expensive by design — that's the spam defence working).
- Don't sell "blockchain" — sell verifiable AI work, evidence integrity, compute receipts. The chain is the how, not the pitch.
5. Start-Monday list
- Move hub agents onto
loopdcaps (an hour, running code). - Anchor the next real client deliverable from qalarc.ai (one CLI call — see runbook).
- Add "provenance certificate" as a premium line item in the next proposal.
- Anchor nightly backup Merkle roots (cron, 20 lines).
- Demo the endispute evidence-anchor flow to one friendly lawyer/client.