TCO & ROI

Stop Paying the Desktop Chat Seat Tax.

Calculate TCO: Per-Seat SaaS Chat vs. Private AI Workforce Architecture.

Per-seat chat bills you for every human who might open a window. A private AI workforce acts as direct labor substitution. ArcFact runs autonomous workers 24/7 inside your perimeter — solving structural labor shortages in manufacturing, trading houses, and back-office operations. Structured Blueprints and pre-indexed retrieval cut token waste by roughly 75% against ad-hoc desktop prompting, and the platform itself lands on a fixed $25,000 enterprise license funded from operating and labor lines instead of IT headcount multipliers.

24/7 autonomous

Workers plan, execute, and report around the clock on your own servers — capacity that has no seat equivalent.

75% less token waste

SKILL.md instruction files and Biblio retrieval replace the re-pasted context every desktop chat pays for on every turn.

$25,000 fixed license

One enterprise license for the platform. Add headcount, add tasks, add hours — the license line doesn't move.

Interactive model

Run the numbers on your own headcount

Move the sliders. Every figure recalculates against your seat count, task volume, and the tool you're renting today.

Your current environment

10250 seats
10200 tasks
Billed monthly, per human, whether or not they log in

Net annual savings with ArcFact

$0

Based on 50 seats at $125/seat and 80 tasks per user per month.

  • Annual SaaS spend Seats × seat cost × 12 + task token drag @ $0.25 $0
  • ArcFact annual cost $25,000 license + $3,000 compute + tokens @ $0.03 $0
  • Recaptured hours per year 4 hours returned per seat, per week, 52 weeks 0

Token drag assumes $0.25 per task on ad-hoc desktop chat versus $0.03 per task under Blueprint-structured execution with pre-indexed retrieval — the ~75% waste reduction, priced. Compute assumes one commodity GPU server amortized annually. Figures are directional; a pilot proposal is quoted against your actual workload.

Architecture, not discounting

Where the seat tax actually comes from

The price gap isn't a negotiation outcome — it's four structural differences between a chat window rented per human and a worker platform installed once.

Dimension
Desktop chat SaaS
ArcFact
Context Window ScalingCost per turn
Every turn re-ships the same background documents into the window. Context grows linearly with the conversation, and you pay for it again on each message — the longer the analyst works, the more expensive each answer becomes.
Workers retrieve only the passages a step needs from a pre-indexed corpus, then release them. Context stays flat as work deepens, so a 40-step job doesn't cost forty times a one-step job.
Instruction OverheadSKILL.md
Process knowledge lives in whatever the user remembers to paste. The same standards, formats, and review rules get re-typed by every person, every session — unversioned, unauditable, and billed as fresh tokens each time.
SKILL.md instruction files are authored once, version-controlled, and loaded by workers on demand. One team codifies the procedure; every worker executes it identically, with no prompt-engineering tax per employee.
Model Lock-InBYOM
The vendor's model is the product. Price changes, deprecations, and rate limits arrive on the vendor's schedule, and your per-seat rate is set by the frontier tier whether or not the work needs it.
Bring your own model. Fine-tuned, open-source on your hardware, or a MaaS endpoint — mixed or swapped per workload. Route cheap steps to cheap models and keep the frontier model for the work that earns it.
Document SearchBiblio precision recall
Uploads are scoped to a conversation. Search quality degrades as the corpus grows, tables and scanned pages come back mangled, and there is no organizational memory between chats.
Archaea Biblio — sold in Japan by Krugle Inc. — indexes the whole corpus into a knowledge graph plus vector store with six retrieval modes, delivering precision recall over tens of thousands of documents instead of per-chat uploads.

Strategic ROI

Labor substitution & pipeline avoidance

Enterprise AI ROI extends far beyond IT budget optimization. In the face of structural labor shortages, Archaea shifts the conversation from standard software costs to true labor substitution and infrastructure efficiency.

ArcFact ROI Formula

Shift the Budget Narrative to HR & Labor

Because of structural labor shortages — especially in manufacturing, trading houses, and back-office operations — automation is no longer just an IT efficiency play. The budget for ArcFact often comes directly from the labor or HR line, not just the IT budget, because it acts as direct labor substitution.

The ArcFact Pitch: Replace person-years of labor + pass compliance audits + cut overtime by hundreds of hours — without adding headcount.

Biblio ROI Formula

Eliminate DIY RAG Pipeline Taxes

Building and maintaining a custom in-house RAG pipeline requires continuous engineering overhead, vector store tuning, and complex parsing maintenance that eats up valuable technical bandwidth.

The Biblio Pitch: Save enterprises the engineer-years required to build and maintain a DIY RAG pipeline, while simultaneously lifting retrieval accuracy across your entire document corpus.

Fast-track procurement

A $25,000 entry point clears without an RFP

The hardest part of enterprise AI is rarely the technology — it's the calendar. Six-figure platform deals trigger formal RFPs, committee review, and security questionnaires that push a start date two or three quarters out.

At $25,000, the ArcFact entry license sits under most VP-level discretionary procurement limits. A sponsoring VP can authorize it directly against an operating budget, so the pilot starts in weeks instead of waiting on a six-month cycle.

  • Fits discretionary sign-off authority — no capital request, no committee slot
  • Skips the 6-month RFP cycle and its vendor-comparison overhead
  • Self-hosted from day one, so security review covers your own perimeter
  • Proves value on one real workflow before any enterprise-wide commitment

$25,000

Fixed entry license
·
Under VP discretionary limit
·
No RFP required