Pricing

Hour-Based at $185/h · Substrate-First Tier Order · NIST Not Charged

Hour-based pricing at $185/h flat across roles. Devs orchestrate Claude Code (which is what makes this engagement deliverable at this cost); PMs and others run discovery, testing, and prompt engineering — both bill at the same rate. Per tier: estimated hours × $185 = build price, plus hard cost to us (AWS + AI API we incur, calculable), plus recommended retainer (cost-plus math: included support hours + hard cost absorbed + operating margin). All numbers are ranges reflecting scope envelope. Tier order updated: Substrate moved to Tier 1 (it's a foundation, not a finale). NIST 800-53 alignment work is Newfangled-side overhead, NOT charged to 4K.

At a glance

Cumulative tier-by-tier rollup

Six tiers, in dependency order. Hours, build price (one-time), cumulative, hard cost (monthly), retainer (monthly cumulative-by-tier-reached). Tier 0–4 is the core engagement; Tier 5 is per-item add-on when blockers resolve.

Tier Hours Build @ $185/h Cumulative build Hard cost (mo) Retainer (mo)
Tier 0 — Foundations
Setup, INSTRUCTIONS, vendor MCPs (AI-assisted authoring)		 20–40 $4–7k $4–7k $10–20 N/A standalone
Tier 1 — Substrate
CKA modules + prompts (foundation for everything below)		 190–340 $35–63k $39–70k $1,000–1,880 $3–5k
Tier 2 — Skills
12 SOP skills + 2 connectors (substrate-aware from day one)		 115–175 $21–32k $60–102k $1,020–1,920 $4–6k
Tier 3 — Automations
5 Managed Agents + Forecast pipe + ClickUp NF connector + Finance HQ discovery		 130–230 $24–43k $84–145k $1,130–2,200 $6–7k
Tier 4 — Intelligence
SiBorg + Client Health Agent (substrate-dependent)		 90–170 $17–32k $101–177k $1,250–2,470 $7–9k
Tier 5 — Extensions
Gated items: Finance HQ Phase 2, Harvest replacement, Zoom, shared-Slack		 70–170 (all 4) $13–32k (all 4) +$13–32k +$150–320 +$1–2k per ext.
Core engagement (Tier 0–4) ~545–955 $101–177k — $1,250–2,470/mo $7–9k/mo
Full deploy (Tier 0–5 all built) ~615–1,125 $114–209k — $1,400–2,790/mo $9–13k/mo
What changed from the previous pricing. Five things. (1) Hour-based at $185/h replaces value-anchored pricing — more transparent, defensible, easier for 4K to push back on specific estimates. (2) Substrate (formerly Tier 3) moved to Tier 1 — it's a foundation that everything else builds on; no longer the "scary big tier at the end." (3) Tier 0 hours dropped to 20–40 — AI-assisted INSTRUCTIONS authoring, templated per-engagement headers, MCP installs are mechanical. Most of what was previously hand-authored is now AI-drafted with leadership review. (4) AI + AWS hard costs doubled at every tier — explicit headroom for real-world growth, richer synthesis runs, and active query volume once skills + agents are in production. (5) NIST 800-53 alignment work is explicitly NOT charged to 4K — it's Newfangled-side overhead, covered separately on the Compliance page. Total build is meaningfully lower than the previous value-anchored numbers because the CKA's core abstraction work is being done outside this project.
NIST exclusion (important)

NIST 800-53 alignment is Newfangled-side, not 4K's bill

Worth being explicit because earlier drafts had this called out as a tier component and we want no confusion.

What's Newfangled-side
$33–48K Year 1 + $25–35K/yr

Newfangled's own NIST 800-53 Low alignment posture: SSP authoring, control implementation matrix, GRC platform adoption, initial pen test, ongoing compliance ops. Recalibrated with AI leverage (browser automation for audits, AI-assisted writing, evidence collection scripting). Year 1 includes a one-time vCISO readiness review. Newfangled pays for all of this. See the Compliance page for the full breakdown.

It's required to credibly deliver to 4K, but it enables Frontier to serve any regulated-adjacent client. We don't pass-through this investment to 4K.

What's folded into 4K's price
Per-build documentation overhead

Each build item carries ~10–15% engineering overhead for alignment documentation (data classification, control implementation matrix per build, evidence-collection plan). This is included in the hour estimates per tier — not a separate line item.

So when Tier 1 says "190–340 hours," that already includes ~20–40 hours of alignment documentation work.

Not on 4K's bill
Newfangled's compliance ops time

Newfangled's ~280–400 hrs/yr allocation for compliance ops (quarterly access reviews, audit responses, evidence collection, policy maintenance — much of which overlaps with normal eng ops) is part of Frontier's ongoing operational overhead, not 4K's retainer.

What IS in 4K's retainer
Evidence packaging for their audits

When 4K's auditors need evidence about the systems we built for them (control implementation per build, attestations, access logs), compiling and delivering those evidence packages is in-scope retainer work up to N hours per quarter. Beyond N hours, billed at $185/h.

01 — Core assumptions

Four load-bearing assumptions

Every number on this page rests on these four assumptions. If any of these break, the whole table moves.

Assumption 01
Single dev orchestrating Claude Code is the delivery model

A senior dev with Claude Code ships roughly 2–3× a traditional dev's output for code-heavy work (connectors, modules, scaffolding). If we go back to traditional team-based delivery, dev hours roughly double across the board.

$185/h works at this velocity because the hours are compressed but real — we're not selling phantom output, we're using better tools.

Assumption 02
Claude Code compresses code, not judgment

What compresses: connector adapters, module scaffolding, schema migrations, test plumbing, prompt boilerplate. What doesn't compress: prompt engineering against novel domains, validation cycles against real data, discovery interviews, business-rule extraction from people's heads, leadership review iterations.

That's why PM/discovery/testing/prompt-eng hours appear separately per tier and don't compress the same way.

Assumption 03
CKA core abstraction is built outside this project

Frontier is building the substrate's multi-tenant infrastructure, abstracted data-source ingestion, modular data-source framework, and prompt-override system as a separate Newfangled-side investment. 4K's pricing does not include this work.

For 4K, Tier 1 Substrate is modules + prompt tuning on top of the abstracted framework — much smaller scope than building substrate from scratch.

Assumption 04
~50 active engagements is the assumed 4K scale

Per-engagement work (INSTRUCTIONS headers, substrate ingestion, indexing) assumes this baseline. If 4K has 80 engagements, indexing time + Notion DB enumeration go up proportionally. If 20, down proportionally.

The retainer numbers also assume ~50 engagements. Scale-related adjustments are addressed in the open commercial questions below.

The connector-and-skill rule of thumb. An MCP connector (straightforward system) is ~4–6 dev hours, doubled for testing, prompt engineering, and client questions = 8–12 hours total. Same shape for a Skill. More for genuine complexity. This rule grounds the per-item estimates on every tier below.
02 — How pricing works

$185/h × estimated hours, plus operating cost

Three numbers per tier, derived from estimated hours and infrastructure rates rather than market-value judgment.

Number 01
Build price = hours × $185

For each tier we estimate dev hours (single dev orchestrating Claude Code) and PM/discovery/testing/prompt-eng hours separately, then multiply the total by $185. Range reflects scope envelope (smaller-end if simplifications land first; larger-end if discovery surfaces complexity).

Number 02
Hard cost to us (monthly)

What Newfangled pays each month to operate the tier. AWS infrastructure + Anthropic API usage we incur. Derived from known AWS rates and Anthropic API pricing × estimated workload.

NOT included: 4K's Claude Teams subscription, 4K's "extra usage" tokens, 4K's own AWS spend if Option A (their boundary), 4K's compliance audits, third-party tool subscriptions (ClickUp, Notion, Slack, Pipedrive, Harvest — those are 4K's existing costs).

Number 03
Retainer = hours @ $185 + hard cost + margin

Monthly retainer math is explicit: N included support hours × $185 + hard cost absorbed + operating margin. Cumulative across tiers — Tier 4 retainer covers Tier 0+1+2+3+4 running together, not stacked separately.

Assumes Option B operating model (Frontier-managed). Option A drops retainer ~30–40%; Option C effectively becomes advisory hours.

Why hour-based vs value-based
Transparency over capture

Earlier drafts used value-anchored pricing ("this is worth $X because the audit named pain Y worth $Z annually"). Defensible but opaque — 4K can't push back on the inputs because the inputs are their value.

Hour-based at $185/h makes the conversation about are these the right hours rather than is this worth it. The numbers tend to land lower than value-based for the same scope; the trade-off is honesty.

Tier 0 / Foundations

Tier 0 — Foundations

Items: #15 #12 #13 #20. Setup work, INSTRUCTIONS, vendor MCP installs. NIST item (#20) is Newfangled-side and NOT in this tier's billed price.

Build price
$4–7k = 20–40 hours × $185

Why so small. Most of Tier 0 is AI-assisted authoring on top of mechanical platform setup. Claude Teams + vendor MCP installs are checklist work. Org and per-engagement INSTRUCTIONS are AI-drafted from interviews + site crawl with leadership review. The previous estimate ran 70–120 hours; AI leverage knocks ~3x off the authoring components.

Platform setup: ~5–10 hours

  • Claude Teams workspace configuration: 2–4 hrs
  • Vendor MCP installations (ClickUp, Notion, Slack, Workspace): 3–6 hrs

INSTRUCTIONS authoring (AI-assisted): ~10–18 hours

  • Org INSTRUCTIONS file (AI-drafted from interviews + site crawl, leadership review iterations): 6–10 hrs
  • Per-engagement INSTRUCTIONS template + initial 3–5 priority engagements (rest can be added incrementally by 4K self-service): 4–8 hrs

Integration / config overhead: ~5–12 hours

Hard cost to us
$10–20/month

AWS infrastructure: $10–20/mo — Account setup overhead, CloudTrail, IAM, baseline monitoring. No compute yet. (Doubled headroom from original estimate.)

AI API: $0/mo — INSTRUCTIONS are static config that lives in 4K's Claude Teams subscription. No Newfangled API spend at this stage.

Retainer: N/A standalone. Tier 0 alone is uncommon as a retainer scope. Maintenance is light enough (~1 hr/quarter) to bundle into as-needed support rather than a monthly arrangement.

NIST alignment (#20) is in this tier's items list but the work is Newfangled-side — paid by Newfangled, not 4K. It enables 4K and every future regulated-adjacent client. The ~10–15% per-build alignment documentation overhead IS folded into the hour estimates above, but the org-level alignment work is not.
Tier 1 / Substrate

Tier 1 — Substrate

#28 CKA modules + per-engagement configuration + prompt tuning. Promoted from old Tier 3 because under the new framing the substrate isn't a heavy build — it's modules and prompts on top of Frontier's abstracted CKA framework. Foundation for every tier below.

Build price
$35–63k = 190–340 hours × $185

Module work: ~42–80 hours

  • 4 data-source modules (ClickUp, Notion, Slack-dual, GDrive-tagged) × 8–15 hrs: 32–60 hrs
  • Per-engagement configuration for 10 engagements: 10–20 hrs

Prompt tuning: ~108–179 hours

  • Extraction prompts (4 sources × 8–12 hrs each): 32–48 hrs
  • Summary prompts (3 cadences — daily/weekly/monthly — × 8–12 hrs): 24–36 hrs
  • Synthesis prompts (5 dimensions × 8–15 hrs — more complex because cross-source): 40–75 hrs
  • Strategic summary prompt (top-level engagement summary): 12–20 hrs

Initial indexing + validation against real 4K data: ~40–80 hours

  • Bulk-index orchestration + rate-limit-aware pacing: 10–20 hrs
  • Validation cycles against real queries: 30–60 hrs
Hard cost to us
$1,000–1,880/month additional

AWS infrastructure: $200–480/mo — CKA container (4–8GB memory, 2–4 vCPU on Fargate) + Meilisearch + Qdrant storage + DynamoDB orchestrator. (Doubled from previous estimate to account for real-world growth and headroom.)

AI API: $800–1,400/mo — Substrate synthesis pipelines run continuously. Daily strategy summaries, weekly activity summaries, Slack threading via Haiku, health signal extraction, SOW/decision extraction, per-source extraction pipelines, synthesis dimensions. (Doubled to account for higher query volume and richer synthesis runs once teams actively query the substrate.)

Cumulative (Tier 0 + Tier 1): $1,010–1,900/mo. This is where hard cost jumps — substrate is the operational-cost-heavy item.

Retainer (Tier 0+1 operational): $3–5k/month.
Math: 10 included support hours × $185 = $1,850 + hard cost ($1,010–1,900/mo absorbed) + operating margin (~15–20% on substrate ops) = ~$3,300–$4,500/mo. Rounded to $3–5k.

Includes: Substrate health monitoring (per-pipeline output review), prompt drift checks, embedding model evaluation, rate-limit monitoring, ~10 hours of support/tuning per month, quarterly substrate review.

Excludes: New synthesis dimensions, new data-source modules, scope expansion to additional engagements beyond initial ~50.

What this tier doesn't deliver to 4K users yet: No daily user-facing value during Phase 2 (when substrate is being built). The substrate is invisible until Tier 2 skills query it. Worth setting expectations explicitly with 4K.
Tier 2 / Skills

Tier 2 — Skills

12 SOP skills + 2 connectors. Substrate-aware from day one (no "lightweight first, deep later" rework). Skills: #01 #02 #03 #04 #05 #06 #07 #08 #09 #10 #11 #14. Connectors: #16 #17.

Build price
$21–32k = 115–175 hours × $185

Skills: ~96–144 hours

  • 12 skills × 8–12 hrs each (per the 4–6 dev + 4–6 testing/prompts rule)
  • Simpler workflow-as-skill items (#01, #04, #07) closer to 8 hrs
  • Skills with richer integration (#08, #11) closer to 12–15 hrs
  • Testing / prompt iteration is already included in the per-skill hours

Connectors: ~19–31 hours

  • #16 Pipedrive redeployment from WWP: 6–10 hrs (simpler than new — clone + credentials + custom-field remapping)
  • #17 Harvest pluggable-layer refactor: 12–20 hrs (internal refactor behind TimeTrackingAdapter protocol, not a new connector)
  • Service-account provisioning + Secrets Manager setup: 1 hr
Hard cost to us
$20–40/month additional

AWS infrastructure: $20–40/mo — 2 Lambda functions (Harvest + Pipedrive connectors), API Gateway, Secrets Manager. (Doubled to account for real-world headroom.)

AI API: $0–20/mo — Skills run in 4K's Claude Teams subscription (not our cost). Connector RunReport codegen on our API is minimal at typical query volumes. (Doubled headroom.)

Cumulative (Tier 0+1+2): $1,020–1,920/mo

Retainer (Tier 0+1+2 operational): $4–6k/month.
Math: 15 included support hours × $185 = $2,775 + hard cost ($1,020–1,920/mo absorbed) + operating margin (~15%) = ~$4,400–$5,400/mo. Rounded to $4–6k.

Includes: Everything in Tier 1 retainer + skill maintenance (prompt drift correction, voice-profile freshness checks), connector keep-alive (OAuth token rotation, API version monitoring), ~15 hours of support per month.

Excludes: New skills, additional per-engagement INSTRUCTIONS headers beyond initial 10 (4K self-serves these).
Quick-wins-in-parallel option. Roughly 7 of the 12 skills don't depend on the substrate for value (#02 sentiment scoring, #04 VTO, #08 SOW Q&A, #09 Joanna's aggregator, #10 Shanice's burndown, #11 brand voice writer, #14 late-invoice review). If 4K wants daily user-facing value before substrate fully lands, these 7 can be built during Tier 1 alongside substrate work. The 5 substrate-dependent skills ship later at full depth. Doesn't change total scope or price — just sequencing.
Tier 3 / Automations

Tier 3 — Automations

5 Managed Agents + 1 small Custom App + 1 Custom MCP Connector + Finance HQ discovery sprint. Items: #18 #19 #21 #22 #23 #24 #25.

Build price
$24–43k = 130–230 hours × $185

Managed Agents: ~85–157 hours

  • #18 Weekly briefing agent (simple, proven WWP pattern): 8–12 hrs
  • #19 AR automation (templates + escalation rules + audit logging): 12–18 hrs
  • #22 Pipeline upstream consolidation: 15–25 hrs dev
  • #23 Client Reports (3–4 per-PM templates as skills + agent wrapper): 32–60 hrs
  • #24 Scorecard Updater (agent + Beto's master-sheet formula migration is the real work): 23–42 hrs

Supporting infrastructure: ~25–45 hours

  • #21 Forecast watched-folder pipeline: 12–20 hrs
  • #25 ClickUp NF connector (RunReport pattern adds slight complexity): 8–15 hrs
  • Integration + cross-build plumbing: 5–10 hrs

Finance HQ discovery sprint (human consulting): ~20–40 hours

Hard cost to us
$110–280/month additional

AWS infrastructure: $40–90/mo — frontier-agents Lambda + EventBridge schedules (5 cron jobs); S3 + Lambda for #21; additional Lambda for #25. (Doubled to account for real-world headroom.)

AI API: $70–190/mo — Managed Agents sessions: AR daily, weekly briefing, scorecard pre-L10, pipeline weekly, client reports. Plus connector codegen overhead. (Doubled to account for richer real-world agent runs.)

Cumulative (Tier 0+1+2+3): $1,130–2,200/mo

Retainer (Tier 0+1+2+3 operational): $6–7k/month.
Math: 20 included support hours × $185 = $3,700 + hard cost ($1,130–2,200/mo absorbed) + operating margin (~15%) = ~$5,600–$6,800/mo. Rounded to $6–7k.

Includes: Everything in Tier 2 retainer + agent monitoring, threshold tuning, drift correction for agent prompts, Forecast watched-folder operations, ~20 hours of support per month.

Excludes: New agent builds, agent scope expansions, Forecast schema-drift work beyond hard-fail-and-flag, Finance HQ Phase 2 (separate when unblocked).
Tier 4 / Intelligence

Tier 4 — Intelligence

#27 SiBorg refactor + #26 Client Health Agent. Both substrate-dependent. Previously bundled with substrate in old Tier 3; now separated because under the new framing they're smaller items that consume the Tier 1 substrate.

Build price
$17–32k = 90–170 hours × $185

#27 SiBorg refactor: ~23–45 hours

  • SiBorg as a skill that calls CKA tools (workflow + comparable retrieval + rate-card lookup): 8–15 hrs
  • Win/loss memory migration with parallel-running validation: 5–10 hrs
  • Validation cycles with Si: 10–20 hrs

#26 Client Health Agent: ~66–127 hours

  • Composition skill (queries CKA dimensions, composes per-account view): 8–15 hrs
  • Managed Agent wrapper (scheduled + alert-driven): 8–12 hrs
  • Small leadership-only portal (drill-down UI): 25–50 hrs (this is real UI work)
  • Multi-channel delivery setup (SES domain verification + Slack webhook): 10–20 hrs
  • Threshold tuning + alert calibration: 15–30 hrs
Hard cost to us
$120–270/month additional

AWS infrastructure: $20–50/mo — Lambda for SiBorg + Client Health Agent; small portal hosting (Lambda + CloudFront if static, or Fargate if dynamic). (Doubled headroom.)

AI API: $100–220/mo — Client Health Agent composition runs (weekly per engagement). SiBorg invocations are infrequent (low cost) but query the substrate heavily. (Doubled to account for richer real-world composition runs.)

Cumulative (Tier 0+1+2+3+4): $1,250–2,470/mo

Retainer (Tier 0+1+2+3+4 operational): $7–9k/month.
Math: 25 included support hours × $185 = $4,625 + hard cost ($1,250–2,470/mo absorbed) + operating margin (~15–20%) = ~$6,800–$8,500/mo. Rounded to $7–9k.

Includes: Everything in Tier 3 retainer + Client Health Agent monitoring (alert sensitivity tuning, false-positive review), SiBorg validation when Si requests, leadership-only access permissioning maintenance, ~25 hours of support per month.

Excludes: Per-logo attrition forecasting (downstream extension, not v1), additional CHA dimensions, Tier 5 extensions.
Tier 5 / Extensions

Tier 5 — Specialized Extensions

4 items, all blocked. Items: #29 #30 #31 #32. Quoted per item when blockers resolve.

Build price (per item)
Hour estimates × $185
  • #29 Finance HQ Phase 2: 15–80 hrs = $3–15k. Low end (15–30 hrs) if Finance HQ has API. High end (40–80 hrs) if scraping or workflow automation needed against UI-only system.
  • #30 Harvest replacement integration: 8–15 hrs = $2–3k. New module/connector following our pattern; Productive/Runn are cleanest.
  • #31 Zoom transcript pipeline: 25–45 hrs = $5–8k. Webhook + polling fallback + module + sentiment prompt tuning.
  • #32 Shared-Slack-channel deep synthesis: 20–40 hrs = $4–7k. Extension of existing Slack module with additional synthesis prompts for client-side content.
Hard cost to us
$150–320/month additional if all built

AWS infrastructure: $10–40/mo — Webhook receiver Lambdas, additional Lambda for replacement-tool connector. Most extensions ride on existing infrastructure. (Doubled headroom.)

AI API: $140–280/mo — Zoom transcript sentiment is the dominant additional cost. Shared-Slack deep synthesis adds the rest. (Doubled to account for richer real-world synthesis volume.)

Retainer (Tier 5 add-on): +$1–2k/month per extension active.
Per extension: ~5 included support hours × $185 = $925 + incremental hard cost (doubled) + small operating margin = ~$1,150–$1,600/mo per extension.

Zoom transcript pipeline is the most operationally complex (webhook reliability monitoring); shared-Slack deep synthesis is mostly additional prompts on existing infrastructure.
03 — Explicitly excluded

What's NOT in any of these numbers

Worth being explicit. The pricing model has clear boundaries.

Excluded from hard cost
4K's existing tool spend stays 4K's
  • 4K's Claude Teams subscription — ~$30/user/month × team size
  • 4K's "extra usage" tokens — when Claude Teams users exceed included token allocation
  • Vendor subscriptions — ClickUp, Notion, Slack, Pipedrive, Harvest (and replacement), Zoom, Google Workspace
  • 4K's compliance audits, pen tests, certifications — their organizational overhead
  • 4K's AWS spend if Option A (4K-owned boundary)
Excluded from retainer
New work, not maintenance
  • New build items — quoted separately at $185/h
  • Major scope expansions — adding engagements past initial ~50, ingesting new data sources, building new synthesis dimensions
  • Beyond N support hours per month — billed at $185/h
  • Newfangled's compliance investment — $33–48K Year 1 + $25–35K/yr is Frontier business overhead, NOT 4K's bill
  • Newfangled's compliance ops time — ~280–400 hrs/yr for quarterly access reviews, audit responses, evidence collection
Operating model. All retainer math assumes Option B — Frontier-managed inside our boundary. If 4K chooses Option A (their boundary, we maintain via controlled access), retainer drops ~30–40% because we're not operating the AWS environment day-to-day. If they choose Option C (code-only delivery, they operate everything), retainer becomes "advisory hours" at $185/h — likely $1.5–3k/month regardless of tier.
04 — Phase timeline implications

What 4K experiences across the engagement

Tier ordering produces a specific user-experience arc. Worth being explicit so we set expectations right.

Phase 1 / weeks 1–3

Tier 0 — Foundations land

Claude Teams workspace configured. Vendor MCPs (ClickUp, Notion, Slack, Workspace) installed. Org INSTRUCTIONS file in place. Per-engagement INSTRUCTIONS headers for the priority ~10 engagements. 4K's team can use Claude Teams with brand voice and the official MCPs starting around week 3.

Phase 2 / weeks 3–10

Tier 1 — Substrate builds

CKA modules ingest 4K's tools. Per-engagement substrate config. Prompts tuned (extraction, summary, synthesis, strategic). Honest caveat: minimal daily user-facing value during this phase. The substrate is invisible until skills query it.

Optional quick-wins-in-parallel: 7 of 12 Tier 2 skills don't depend on substrate — could be built during this phase to give 4K daily value before Phase 3.

Phase 3 / weeks 8–14

Tier 2 — Skills land

Skills progressively go live. Each one is substrate-aware from day one — onboarding briefs are deep, Roundtable agendas use recency-weighted reconciliation, rate-card lookup uses indexed proposal context, SiBorg-lite queries comparable past engagements. No asterisks.

Phase 4 / weeks 12–18

Tier 3 — Automations

AR automation, weekly briefings, scorecard updater, pipeline consolidation, client reports. All wired to substrate context. Beto's, Peke's, Jade's, Joanna's named pains start retiring.

Phase 5 / weeks 16–22

Tier 4 — Intelligence

Full SiBorg and Client Health Agent live. Leadership gets the per-account composed health view. Si gets cross-engagement comparable retrieval and Harvest-actuals feedback loop.

Phase 6 / when unblocked

Tier 5 — Extensions

Per-item delivery when blockers resolve. Finance HQ Phase 2 after #22 discovery, Harvest replacement after 4K decides, Zoom transcripts after policy interpretation, shared-Slack deep synthesis after MSA review.

The "Phase 2 invisibility" trade-off is worth highlighting in the proposal. Substrate-first sequencing means 4K commits to ~7 weeks of foundation work before the deep skills land. The trade-off is: every skill that follows is at full depth from day one, no "lightweight first, deep later" rework. The quick-wins-in-parallel option mitigates this trade-off if 4K wants visible value earlier.
05 — Open commercial questions

Decisions before we quote

Pricing v0.3 · Hour-based at $185/h · Substrate-first ordering · NIST excluded from 4K's bill · Newfangled Frontier · 2026-05-15