[00 / STRUCPACT BY LAUNCHSTRUC LLC]
Procurement and legal teams upload an incoming MSA, DPA or SaaS order form. StrucPact checks every clause against your internal risk tolerances and returns tracked redlines, each citing the playbook rule it enforces, in seconds instead of days of back-and-forth.
[ Specification / sample values ]
| Triage latency (Haiku, typical) | 0.4 to 0.9 s |
| Redline drafting (Sonnet, per clause) | 0.6 to 1.8 s |
| Context capacity per document | 200K tokens |
| Cached playbook prefix, input cost | up to 90% lower |
| Cached playbook prefix, time to first token | up to 75% lower |
| Retrieval chunking applied to contracts | None |
Latency ranges are typical and vary with document length. Savings figures are upper bounds that apply to the cached portion of the prompt.
[01 / WORKBENCH]
Switch playbook guardrails on and off, pick a contract, and watch the redlines change. Then inspect the structured payload the Claude API returns.
[ A / Document ]
[ B / Playbook Guardrails ]
4/4 active
Cloud Infrastructure Master Services Agreement
Sample contract text for demonstration. Clauses are fictional.
Provider shall make the Services available at a Monthly Uptime Percentage of at least 99.95%. Service Credits are calculated under Schedule B and are Customer's sole monetary remedy for availability failures below that threshold.
In no event shall Provider be liable for any indirect, incidental, special, exemplary or consequential damages arising out of or relating to this Agreement, including loss of profits or data, provided that Customer shall remain liable for all consequential damages arising from any breach of Section 6 (Acceptable Use).and the foregoing exclusion applies equally to each party. Neither party shall be liable to the other for any such damages, whether in contract, tort or otherwise.
This Agreement renews automatically for successive twelve (12) month terms. Upon each renewal, the Fees shall increase by the greater of seven percent (7%) or the percentage change in CPI-U plus two percent (2%)the lesser of four percent (4%) or the percentage change in CPI-U over the preceding twelve (12) months, applied to the Fees in effect immediately prior to renewal.
Customer may terminate this Agreement for convenience upon one hundred eighty (180) days'thirty (30) days' prior written notice, and shall pay all Fees for the remainder of the then-current Term as an early termination charge, and shall pay only Fees accrued through the effective date of termination.
Provider may use Usage Data and Customer Content to develop, train and improve Provider's machine learning models and productssolely to provide and secure the Services for Customer, and shall not use Customer Content to train, fine-tune or evaluate any machine learning model.
[02 / ARCHITECTURE]
StrucPact is built directly on the first-party Anthropic Claude API. Three platform features carry the load.
01
The complete MSA, with schedules, exhibits and order forms attached, is passed to the model in one 200K-token context. No retrieval chunks, so cross-references between a definition on page 3 and a carve-out on page 41 stay intact.
02
Your risk tolerances and fallback clause library sit in a cached system prefix marked with ephemeral cache control. After the first request in a window, each review reads the playbook from cache instead of paying for it again.
03
Claude Haiku extracts metadata and classifies each clause. Only clauses that breach a rule, such as bilateral indemnification or liability caps, escalate to Claude Sonnet for redline drafting.
04
Findings come back through a strict tool schema: clause identifier, playbook rule citation, original text, proposed text. Your review tooling parses the result without scraping prose.
[03 / ECONOMICS]
Flat subscription tiers. No per-page fees and no charge for re-running a document after you edit the playbook.
Annual billing: two months included at no charge
[ Tier 01 ]
$249/mo
$2,990 billed annually
One legal or procurement team running a single playbook.
Start with Growth[ Tier 02 ]
$666/mo
$7,990 billed annually
Multiple business units with separate risk tolerances.
Start with Scale[ Tier 03 ]
Custom
Quoted per engagement
Negotiated volume, security review and deployment terms.
Request a quote| Feature | Growth | Scale | Enterprise |
|---|---|---|---|
| Contracts reviewed per month | 40 | 150 | Custom volume |
| Active playbooks | 1 | 5 | Unlimited |
| User seats | 5 | 25 | Custom |
| Context window per document | 200K tokens | 200K tokens | 200K+ tokens |
| Ephemeral prompt caching of playbooks | Included | Included | Included |
| Haiku to Sonnet model routing | Included | Included | Included |
| Redline export (DOCX, JSON) | Included | Included | Included |
| Role-based access control | Included | Included | Included |
| Audit log retention | 90 days | 1 year | Custom |
| SAML single sign-on | Not included | Included | Included |
| Data Processing Addendum | Standard | Standard | Negotiable |
| Support | Email, next business day | Priority email, 4 business hours | Named architect |
Prices in US dollars, exclusive of applicable taxes. Subscriptions renew under the Terms of Service.
[04 / FAQ]
Vendor-side commercial paper: master services agreements, data processing addenda, SaaS order forms, statements of work and their schedules. Documents up to roughly 200K tokens are read in a single pass, which covers most MSAs with exhibits attached.
A playbook is a set of rules, each with an identifier, a risk position and an approved fallback clause. Your team writes them in plain language, and StrucPact cites the rule identifier on every redline it produces.
No. StrucPact uses the Anthropic Claude API under commercial terms, which exclude API inputs and outputs from model training. LaunchStruc LLC does not train or fine-tune any model on customer documents, playbooks or findings.
No. StrucPact is software, not a law firm, and does not provide legal advice. It drafts first-pass redlines against your own policy so counsel can approve, adjust or reject them faster.
Every review is available as tracked changes in DOCX and as structured JSON with clause identifier, rule citation, original text and proposed text, so it can feed contract lifecycle or ticketing systems.
Yes. Use the workbench on this page to see how guardrails change the output, then book an architecture review and we will run StrucPact against a sample agreement with your playbook rules.
[ Next step ]
Thirty minutes with an engineer: we load a sample of your playbook, redline a real contract and walk through the JSON your tools would receive.
