JamesPolk.net

PDF Terms Explanation

Based on that PDF, here are the key terms / concepts (like SLA) explained in plain language, with why they matter for your HIPAA-compliant legal inference platform and how they tie to the numbers in the chat:

1. SLA – Service Level Agreement

A contractually backed promise about the performance and availability of your inference API. Typical SLA elements you’d offer to firms or platform integrators:

  • Uptime guarantee (e.g., 99.9% availability of inference endpoints)

  • Latency bounds for responses (important for interactive “secure chat” sessions)

  • Support response times (e.g., critical incident reply within 1 hour)

  • Data isolation assurances (dedicated or logically isolated compute per customer)

Why it matters: LegalTech/RegTech platforms and law firms rely on predictability. Having clear SLAs differentiates you from “best-effort” competitors and justifies premium pricing (e.g., the $15K+ platform tier).

2. LTV – Lifetime Value

Total revenue you expect to earn from a customer over their entire relationship, e.g.:

  • For a platform operator paying $15K/month, a 12-month relationship gives an LTV of $180K.

  • This drives your unit economics and signals how much you can spend to acquire them (CAC) while staying profitable.

3. CAC – Customer Acquisition Cost

What it costs you to win a paying customer (marketing + sales + onboarding).

  • Example: If you spend $5K to acquire a platform operator who delivers $180K in revenue over 12 months, your LTV/CAC ratio is extremely healthy (~36x).

  • For law firm clients at ~$3K/month, reasonable CAC might be $1K–$3K, expecting payback within a month or two.

4. HIPAA-compliant inference boundary & traceable evidence

This refers to how and where sensitive data (ePHI) is processed and the technical safeguards around it:

  • Inference boundary: Ensuring all document ingestion, model execution, and output generation happen within a controlled, isolated environment (your disposable container).

  • Traceable evidence: Every inference run logs what document fragments were used, prompt/model version, which hardware (with attestation), who invoked it, and when—producing an audit trail usable in audits (HIPAA/SOC II).

Why it matters: It’s core to the premium you charge—clients aren’t just buying answers, they’re buying trusted answers they can defend in compliance reviews.

5. Audit-ready provenance

Provenance is the lineage of a model’s answer:

  • What was asked (prompt versioning)

  • What sources (document snippets) were used

  • Which model and configuration generated it

  • On what hardware it ran (with your “certified clean” attestations)

  • When and by whom it was accessed

Packaging that into an “audit-ready” bundle means a lawyer or regulator can see the chain-of-custody of an inference, reducing ambiguity/risk.

6. Right-sized pricing / tokenized usage

  • Right-sized pricing means clients pay for what they actually need—e.g., a firm tier at $3K/mo bundles a practical volume of secure inference sessions rather than forcing overprovisioned flagship hardware purchases.

  • Tokenized usage underlies granular cost: each inference consumes tokens (input+output), which translate into compute cost (either explicit if on hosted APIs or implicit as GPU time if self-hosted).

  • You expose this via dashboards, caps, and burst packs so clients understand and control consumption.

7. Platform operator / integration tier

These are higher-volume, higher-touch customers (LegalTech or RegTech SaaS embedding your API). They:

  • Consume large numbers of underlying sessions (e.g., thousands of chats)

  • Get dedicated capacity, priority SLAs, and compliance bundles

  • Justify the $15K–$30K+/mo pricing because they drive high underlying usage and have multi-month engagements, giving you the strong LTV and ROI metrics cited.

8. Competitive positioning references

  • DISCO Cecilia: Embedded insight tool (often with opaque pricing and persistent context caching)

  • Westlaw / Westlaw Edge: Legal research, not inference automation; lacks built-in provenance/audit packaging

  • Your platform combines insight + auditability + composability (API) with clear, transparent tiers—hence the premium and stickiness.

9. “Roadmap” / Future code/contract interpolation

This is the planned extension from “insight” (what the model tells you about a contract or legal document) to automated action (translating clauses into executable logic or enforceable code). It’s a deeper value layer that enables upsells and cements integration.

Wrap-up

Everything in the doc (pricing tiers, $3K firm price, $15K+ platform tier, ROI/CAC, SLAs, compliance/moat language) ties back to:

  • Granular economics (sessions, tokens, compute cost)

  • Packaged value (predictable firm/monthly billing with audit evidence)

  • Scalable integrations (platform operators with high LTV)