Viska sjóðir · Visk-AI · 2026-05-26

Yfirlit fyrir hluteigendur

Staðan á Visk-AI

Hermes tekið upp sem sjálfstætt research runtime · xAI Developer reikningur opnaður fyrir tweet analysis.

boas.dev · Confidential

$183 Stage 1 floor · per month

Yfirlit

Tvær ákvarðanir, einn ársfjórðungur, einn reikningur.

Hermes (Nous Research, MIT, open-source) becomes the runtime that powers Visk-AI's autonomous research loop. xAI's developer API gives that same loop legal, programmatic access to Grok for tweet analysis from Slack.

$183 – $363Stage 1 light · per month
$50 – $115Tweet analysis · 100 / day
166.420GitHub stars · Hermes adoption
17 / 23Security controls on Hermes
zeroBroker credentials in stage 1

§1

Hvers vegna Hermes?

Hvers vegna ramminn vinnur á eigin verðleikum, og fimm skilyrði sem fylgja.

Two paths: hand-roll a minimal agent loop on Railway (4–8 months to parity) or adopt Hermes, Nous Research's open-source agentic framework. We adopt — hand-roll stays as fallback if any of five conditions fails.

AxisWhy it mattersHermes today
Code qualityA framework we can't patch fast becomes a liability.166,420 stars · 4,802 merged PRs · 165k LoC in last 9 days.
Release disciplineTrading-adjacent code needs auditable releases.Weekly tagged releases (vYYYY.M.D) · public release notes.
LicenseClean IP posture for client deployments.MIT verbatim · no CLA friction.
Security postureThe framework's threat model must match ours.Explicit "OS is the only boundary" doctrine.
Supply chainOne bad dependency wipes a quarter of work.Exact-pinned deps · documented quarantine response to the 2026-05-12 mistralai Mini Shai-Hulud incident.
Plugin surfaceOur security controls live as plugins.17 of 23 hardening requirements implementable as Hermes plugins, configs, or built-ins.
Community velocityBus-factor risk mitigated by inbound contributors.215–295 community contributors per release cycle.

Fimm skilyrði

If any fail, we revert to hand-roll.

1

Pin a tagged version

Deploy v2026.5.16 (Hermes v0.14.0). No auto-pull.

2

OS-level sandbox

Hardened Docker Compose stage 1. NVIDIA OpenShell stage 2.

3

Curated plugins only

No third-party marketplace. Egress-blocked at the network layer.

4

Mirror fork

Pantheon-controlled copy. Upstream disappearance does not strand us.

5

Hook half-life

Budget plugin maintenance into operating cost — framework moves fast.

What Hermes does not ship

Hermes is a framework, not a finished platform. Four things we add ourselves:

What's missingHow we fill it
Dual-LLM separationWe add it as a plugin · 1–2 days
Database access boundarySupabase provides it (Row-Level Security)
Network egress firewallEnvoy sidecar in the same Railway project — single highest-value control
Trading-grade audit trailSeparate Supabase project · append-only log

§2

Útfærsla fyrir Visk-AI

Hvernig útfærslan er löguð — og ákvörðunin sem afmarkar áhrifin.

Deployment shape

Deployment topology Deployment topology: operator → Slack → orchestrator → knowledge base, fed by Hermes loop, review sessions, and ingestion jobs

Hermes does research and writes into a staging area. A separate supervised review process is the only path that promotes work to approved. Slack and the dashboard only read from approved. The research engine cannot bypass review.

Two-tier memory split

Tier 1 · Hermes working memory. Inside the Hermes container. Transient reasoning state. Local SQLite-backed with trust scoring. If lost, rehydrates from staging.

Tier 2 · Synthesized research wiki. Separate service. Holds the curated client research. Hermes has no read access — writes raw to staging, separate synthesis promotes. Container compromise does not exfiltrate the wiki.

Lag 1 + sviðsetning fara í áfanga 1. Samsetningarþjónusta í áfanga 1.5. Rekstraraðili valdi þetta með orðunum „we can afford to take the time to ship this properly."

Runtime

RuntimeRailway CPU containers · Pro tier.
GPUNone. Railway does not offer GPU instances. All inference via API.
Primary inferenceNous Forge API (inference-api.nousresearch.com) · Hermes-4 family · OpenAI-compatible.
Non-Nous modelsOpenRouter · zero-retention filter · provider failover · model pinning.

Geta í Stage 1

Read raw market + thematic signals. Core research loop.

Write to staging tables. Designed boundary.

×

Write to approved tables. Only the review process can promote.

×

Read the curated wiki. Bounded blast radius (Tier 2).

×

Hold broker credentials. Stage 2 question · separately gated.

×

Execute trades. Stage 2 question · separately gated.

·

Reach the open internet. Envoy egress sidecar · default-deny allowlist.

§3

Grok / xAI — tweet analysis from Slack

Af hverju Developer API en ekki SuperGrok — og hvar reikningurinn liggur í verkflæði.

Why xAI Developer API and not "SuperGrok"

The xAI Developer API (api.x.ai) is the only Grok surface that legally permits programmatic access. Every consumer Grok variant is forbidden by xAI's terms for bot use.

VariantHas API?Use for bots?
SuperGrok consumer ($30/mo)NoForbidden
SuperGrok Heavy ($300/mo)NoForbidden
X Premium+ (includes Grok)NoForbidden
Grok Business ($30/seat)NoForbidden
xAI Developer API (api.x.ai)YesThis is the path

Hvað fáum við

Verkflæði — Slack /analyze

Tweet analysis flow Slack /analyze tweet-flow: user → Slack ACK → extract → Grok analysis → DB write → thread reply

Same Grok account is the analysis engine; same research database is long-term memory. Reddit, HN, YouTube, RSS plug into the same table going forward.

§4

Kostnaðaráætlun

Verðlagning á verkefni í tveimur sviðsmyndum. Light = Stage 1 grunnlína. Heavy = market-hours hot.

Verkefni 1 — Slack /analyze tweet analysis

ComponentPer call10 / day100 / day1000 / day
Tweet extraction (X API v2)$0.005 – $0.025$1.50 – $7.50$15 – $80$150 – $800
Analysis (xAI grok-4.3 + x_search)~$0.011~$3~$33~$330
Slack delivery$0$0$0$0
Storage (free tier)$0$0$0$0
Subtotal~$0.016 – $0.036~$5 – $11~$50 – $115~$480 – $1,130

Volume gate

$200/month soft cap. Crossing it pages the operator via a daily-cost probe.

Verkefni 2 — Autoresearch loop (Hermes á Nous Forge)

VolumeToken budgetMonthly cost
Light (stage-1 baseline)5M tokens / day~$23
Heavy (multi-room or market-hot)50M tokens / day~$230

Nous rate-limit tier: Default (180 req/min) at launch. Ultra only when telemetry justifies. Every request instrumented from day one.

Verkefni 3 — Yfirferð (Claude / GPT um OpenRouter)

VolumeCadenceMonthly cost
Light (event-driven, off-hours)On artifact-ready signal$50 – $150
Heavy (market-hours, 1–2 min cadence)Continuous$300 – $800

OpenRouter's 5.5% aggregation fee buys: single credential surface, per-request zero-retention filter, automatic provider failover, model-ID pinning that catches silent vendor swaps.

Verkefni 4 — Embeddings (Voyage-3-lite)

VolumeToken budgetMonthly cost
Light0.5M tokens / day~$10
Heavy4M tokens / day~$80

Finance-tuned, 1024-dim, cheapest in tier. Column dim pinned to 1024 → swap to self-hosted bge-large-en-v1.5 later without schema migration. OpenAI text-embedding-3-small held as DR fallback (1536-dim → re-embed required).

Samtals — Stage 1

LineLight volumeHeavy volume
Nous Forge (Hermes-4)$23$230
OpenRouter (non-Nous mix)$50 – $150$300 – $800
Review session tokens$50 – $100$150 – $300
Railway runtime (+ Envoy)$50 – $80$80 – $150
Embeddings (Voyage-3-lite)$10$80
Database (existing)$0$25
Stage-1 total$183 – $363 / mo$865 – $1,585 / mo

Tweet analysis adds $50 – $115 / mo at 100 analyses/day on top.

§5

Innleiðing — 10 skref

Staðfesting við hvert skref áður en næsta hefst.

  1. Credentials deposited — xAI key · X API bearer · Slack signing secret + bot token · Nous Forge · OpenRouter · Voyage · DB role JWTs.
  2. Database schema migration — new research schema, Row-Level Security forced on every table.
  3. Egress sidecar deployed — Envoy · allowlist · default-deny.
  4. Hermes loop deployed — Railway CPU service · staging-only write permissions.
  5. Review process wired — scheduled sessions read staging, write promotions.
  6. Dashboard read path — three curated views with realtime subscriptions on approved tables.
  7. Governance daemon — flag table is single source of truth for pauses + incidents.
  8. Audit pipeline — separate Supabase project · append-only action log.
  9. Soft launch — operator-only Slack channel · 7–14 days · calibrate the review process.
  10. General availability — end-client Slack access opens.

Tweet-analysis Slack workflow lands as a parallel sub-arc — five active phases · ~3.5 person-days · ready when step 1 credentials land.

§6

Opin atriði

Frestaðar ákvarðanir sem viðskiptavinurinn ætti að vita af — ekki óvissa.

EU data residency

Operating on working assumption of "no EU-residency requirement" pending legal confirmation. Architecture built so reversal is a configuration change, not a rebuild.

Zero-data-retention with Anthropic

Default routing for review prompts is OpenRouter with ZDR filter. Direct Anthropic ZDR is a separate contractual upgrade — confirmed before any prompt carries client-identifying tokens.

Self-hosted embeddings

Voyage-3-lite is the launch pick. Documented swap path to a self-hosted alternative that does not require corpus re-embed. Triggered by cost pressure or data-sovereignty needs.

Broker integration

Stage 2 work. Will not begin until stage 1 is stable and the review process has accumulated 30 days of operator-comparison data.