Economics
Per-seat AI tax
One 25,000-student campus can spend $7.8M/year across four AI-enabled tools. The demo reframes those tools as thin clients to one institutional model.
What reviewers see
The private demo makes the public claim inspectable.
The public page explains the economic problem. This page shows the operational shape: endpoint, records, identity, audit packet, six-week pilot path, and the exact links an educator can forward internally.
Architecture
Local endpoint
Vendors point at a local HTTP inference endpoint. Queries stay inside the institutional boundary and are written into signed Memory Block rows.
Audit
Identity-bound output
AttestoBind records who signed each source row and who requested each AI output. The result is replayable later, even if a vendor tool disappears.
Blockie Talkie pilot packet
The private reviewer demo shows the operator console, proof packet, and talk track.
This is the approved-reviewer version of the public Blockie demo. It keeps the evidence public-safe while showing how the actual Blockie Talkie app surfaces become an institution pilot: local AI, BTCore, tools, approvals, ledger, and export.
Reviewer spread
Blockie Talkie turns the Sovereign AI pitch into a live operating workflow.
Choose a scenario, run the verification spread, copy the proof JSON, and use the presenter script to walk educators, CIOs, compliance leads, and vendor teams through the same story.
OperatorBlockie + Elliott
ShellTauri desktop
Terminal, AI chat, utility dock, settings, and review workspace.
BridgeBTCore 3006
Wallet, mesh mail, TLS, stack status, and PQ identity tools.
AI groupElliott + agents
Local model lane, mention routing, and agent wall review trace.
VoiceLocal TTS + STT
Operator speech output and input status for staff-facing sessions.
ConsentApproval tools
Payments, mail sends, reward sync, and X402 spend remain gated.
LedgerAttesto export
Instruction, build, review, sign, Memory Block, and export states.
Advisor answer: student-success plan
Local inference over approved Memory Block rows.
local AI
PQ signed
exportable
AI group@elliott + @reviewer
VoiceLocal TTS / STT
WalletqBTC + X402 approval
Mesh mailapproved reviewer inbox
LedgerMemory Block attesto
ExportJSON + brief
Presenter script
Seven-minute walkthrough for educators and CIOs.
Use this talk track during a screen share. It is written to move from budget pain to operational proof without overclaiming live production deployment.
00:00
Frame the pain
Every vendor is reselling AI inference. The institution pays repeatedly and data crosses multiple vendor boundaries.
01:15
Show the console
Open Blockie Talkie. Point out local AI, BTCore, mesh mail, voice, wallet approvals, and the ledger.
02:30
Run advisor flow
Show a student-success answer generated only from allowed Memory Block rows and held for advisor acceptance.
03:45
Open proof JSON
Explain source rows, identity binding, SHA3-512 digest, ML-DSA-87 signature, and Memory Block-first storage.
04:45
Switch scenarios
Show student earning, compliance export, vendor endpoint, and agent build attestos from the same framework.
05:45
Name the boundary
No model weights, private keys, student records, or production credentials are in the packet. Production uses client-owned data.
06:30
Close the offer
Ask for one tool, one records domain, one identity policy, and one six-week pilot sponsor.
07:00
Next step
Send the pilot brief and schedule the technical scope call with IT, compliance, and the tool owner.
# Blockie Talkie Sovereign AI Demo Script Company: Blockie Talkie LLC Products: Proofnet BTC, Proofnet AI, Blockie Talkie, Memory Blocks, AttestoBind Audience: educators, CIOs, provosts, compliance leads, procurement, and vendor integration teams 1. Start with the economics. The institution is not buying AI once. It is buying the same AI capability repeatedly through SaaS tools. The writing tutor, advising bot, research assistant, and proctoring reviewer each carry an AI line item. 2. Show the operating model. Blockie Talkie is the staff-facing console. It speaks to local Proofnet AI and BTCore. Staff can use AI Group, voice, mesh mail, wallet/payment approvals, and the ledger without sending student records to a hosted model. 3. Run the advisor scenario. The AI answer is generated from approved Memory Block rows. The output stays draft until a human advisor accepts it. That accepted output becomes a signed record. 4. Open the proof JSON. The packet shows source rows, requester identity, adapter policy, SHA3-512 digest, ML-DSA-87 signature, Memory Block status, and optional Bitcoin anchor field. 5. Switch scenarios. Student earns shows portable proof-of-knowledge and payment approval. Compliance shows audit export. Vendor endpoint shows how SaaS tools become thin clients. Agent build shows build/review attestos. 6. State the safety boundary. This packet contains public-safe fixtures. Production replaces demo identities and rows with institution-owned data, institution identity policy, and institution hardware. 7. Ask for the pilot. Pick one tool, one records domain, one identity source, and one pilot sponsor. The goal is to prove local inference, cost comparison, and replayable audit in six weeks.
Demo run
Run the private Sovereign AI verification spread.
This is the review-script version of the demo: it walks an institution from SaaS cost problem to signed, replayable AI output.
Institution scenario
Institution25,000 students
Current tools4 AI SaaS products
AI line item$7,800,000/year
01
Writing tutor
$2.0M/year, hosted AI endpoint.
02
Advising bot
$2.5M/year, hosted AI endpoint.
03
Research assistant
$1.5M/year, hosted AI endpoint.
04
Proctor reviewer
$1.8M/year, hosted AI endpoint.
Verifier output
Private demo check
Ready. Click "Run private demo check" to show the full review spread: - cost case - local endpoint - identity-bound rows - Memory Block audit packet - vendor integration boundary - pilot readiness
Technical packet
Endpoint, record, identity, and audit packet.
Local model endpoint
Vendor calls the institution, not a hosted model.
POST /v1/chat/completions
Host: proofnet-ai.institution.local
Authorization: Bearer institution-issued-service-token
{
"model": "proofnet-ai-local",
"messages": [
{"role":"system","content":"Use approved Memory Block rows only."},
{"role":"user","content":"Draft advising plan for student record MB-2026-04-1138."}
],
"audit": {
"record_scope": "student_success",
"identity_policy": "attestobind_required"
}
}
Memory Block row
Every AI output becomes replayable evidence.
{
"type": "proofnet_ai_output_v0",
"institution": "approved-reviewer-demo",
"requester_identity": "faculty-advisor-demo-id",
"source_rows": ["MB-2026-04-1138", "MB-2026-04-1142"],
"model_endpoint": "local_openai_compatible",
"state_digest": "sha3-512:4d3e...b9a1",
"signature": "ML-DSA-87",
"record_layer": "proofnet_memory_block_first",
"external_anchor": "optional"
}
Security boundary: this private demo is a review packet. A production pilot replaces fixture identities, local endpoint examples, and sample rows with institution-owned hardware, institution identity policy, and institution data. The Proofnet binding contract stays the same.
Pilot path
Six-week institutional pilot checklist.
✓
Week 1: scope
Ready: pick one tool, define records, identity system, hardware target, success metric, and compliance owner.
✓
Week 2: deploy
Ready: install Proofnet AI, Memory Blocks, AttestoBind, and Blockie Talkie on institution-controlled hardware.
✓
Week 3: connect
Ready: point one SaaS or internal tool at the HTTP inference local endpoint.
✓
Week 4: verify
Ready: show signed input rows, signed AI output rows, replay, export, and tamper checks.
✓
Week 5: compare
Ready: run hosted vendor AI and local Proofnet AI side by side for cost, latency, privacy, and audit.
✓
Week 6: expand
Ready: add the next tool without adding another per-seat inference contract.
Forwardable brief
Sovereign AI pilot brief.
Copy or download this text when forwarding the private demo internally.
# Proofnet BTC Sovereign AI Pilot Brief Company: Blockie Talkie LLC Product family: Proofnet BTC, Proofnet AI, Memory Blocks, AttestoBind, Blockie Talkie Audience: institutions, educators, public agencies, healthcare systems, and enterprise compliance teams The problem: Large institutions are buying AI repeatedly through per-seat SaaS tools. A 25,000-student institution can pay about $7.8M per year across four common AI-enabled tools while sending student or institutional data to multiple hosted inference endpoints. The Proofnet BTC answer: Deploy one sovereign reasoning model on institution-owned hardware. Vendors become thin clients that call the institution's local HTTP inference endpoint. Queries, source records, and AI outputs stay inside the institution. The record layer: Proofnet Memory Blocks preserve every accepted source row and AI output as a deterministic, identity-bound record. AttestoBind binds records to the signer's existing identity system. ML-DSA-87 signatures and SHA3-512 digests make the proof packet durable and replayable. The pilot: 1. Scope one tool and one records domain. 2. Deploy Proofnet AI, Memory Blocks, AttestoBind, and Blockie Talkie on institution-controlled hardware. 3. Point one vendor or internal tool at the local endpoint. 4. Compare cost, latency, privacy boundary, and audit output against the hosted vendor path. 5. Expand to additional tools without buying another per-seat inference contract. The claim: Vendors should sell the tool. The institution should own the AI boundary, the records, the audit trail, and the cryptographic proof of what happened.