Your client’s positions, applied the same way on every draft.
A client’s preferred positions usually live in old markups, shared folders and a partner’s memory. Lexnus puts them in a versioned playbook. Everyone on the matter checks each new draft against the same rules and sees exactly where it departs from the client’s instructions.
Most clients have clear views on liability, termination, data protection and governing law. Few of those views are written down in one place. The team pieces them together at the start of every review.
In old markupsThe last negotiation shows what the client accepted. It doesn’t show why, or whether that position still holds.
In shared foldersA guidelines memo from two years ago, a precedent bank, and an email from the client’s GC that changed one of them.
In a partner’s memoryThe associate learns the client’s red lines when the partner marks up their review.
The result is inconsistency. Two associates give two answers on the same point. Review time goes on rediscovering positions instead of negotiating them. And the client notices.
How it works
01
Import the client’s guidelines
Upload the client’s guidelines or a playbook memo as a Word document. Lexnus proposes rules from it: liability caps, required terms, notice periods, governing law. A lawyer reviews each rule before it is published. Nothing goes live on the AI’s reading alone.
02
Add the approved wording and fallbacks
Keep the client’s preferred clauses in the clause library. Add variations for the positions the client accepts under set conditions, such as a higher cap above a certain contract value.
03
Check each draft
Upload the counterparty’s draft as a PDF or DOCX. Lexnus extracts the terms and applies the client’s rules. Each finding quotes the text, names the rule and says what kind of problem it is: a violation, a missing term or a point for a lawyer to decide.
04
Escalate what matters
Each rule has an escalation level: flag, require approval or reject. A liability cap above the client’s limit can require a named approver’s sign-off. A routine deviation is simply flagged.
Example. The client accepts liability up to 1.5× annual fees. The draft says 3×. Lexnus flags the deviation, quotes clause 9.1 and shows the client rule behind it.
Counterparty draft
Master Services Agreement
Halden Logistics Ltd (the Customer) and Vireo Systems Ltd (the Supplier)
9. Limitation of liability
9.1 The Supplier’s total aggregate liability under this Agreement shall not exceed three times (3×) the fees paid in the preceding twelve months.
11. Suspension
11.2 The Supplier may suspend the Services on reasonable notice if any undisputed invoice remains unpaid.
18. Governing law
18.1 This Agreement is governed by the laws of England and Wales.
Halden Logistics · MSA, customer side
Checked against 9 client rules
Violation · Requires approvalLimitation of liabilityCap is 3× annual fees. The client accepts up to 1.5×.
MissingData processingNo data processing terms. The client requires them when the supplier handles personal data.
Needs reviewSuspension noticeThe notice period isn’t stated. The client requires 30 days. A lawyer should confirm.
PassGoverning lawEngland and Wales. Matches the client’s position.
Client playbook v3 · published 12 Aug 2026
Illustrative. Fictional client, counterparty and figures.
Keep the reasoning when instructions change
Clients revise their positions. A new GC arrives, a dispute changes their appetite for risk, a regulator moves. In Lexnus, each change is a new published version of the playbook.
Every result records the version it was checked against. Months later, anyone on the team can see which rule applied to a draft and why the finding read the way it did, even if the client has since moved its position.
AI reads the draft. The client’s rules decide.
AI extracts the terms from the document: the cap, the notice period, the governing law. The playbook then decides whether each one complies. The outcome is a pass or a fail against a rule a lawyer approved, not a confidence score.
When a term can’t be established from the text, Lexnus marks it for review instead of guessing. What to accept, and what to push back on, stays with your lawyers.
The client’s playbook goes where your lawyers work
Lexnus delivers the playbook itself, not just a review screen. Over MCP, an AI assistant such as Claude can read the client’s rules and approved clauses, and check a draft against them. A lawyer who asks Claude to rework a clause gets the client’s position, not a generic one.
Give each client its own workspace, with the members and playbooks for that client. The rules your lawyers approved are the same ones every surface applies.
What changes for the team
PartnersSet the client’s positions once. See where a draft departs from them without re-reading every markup.
AssociatesStart each review knowing the client’s red lines. Spend the time on the points that need judgement.
Knowledge lawyersMaintain one versioned source for each client’s positions, instead of memos and precedent folders.
The clientGets the same positions applied on every draft, whoever on the team does the review.
What a pilot looks like
Start small enough to judge the result properly: one client and one contract type.
01Pick the client and contract typeChoose a client with settled positions and a steady flow of one kind of agreement.
02Build the playbook togetherWe import the client’s guidelines and your precedent clauses. Your lawyers review and publish the rules.
03Run it on reviewed draftsCheck agreements your team has already marked up. Compare Lexnus’s findings with your lawyers’ markups.
04Decide on the evidenceSee where the findings matched your team, where they didn’t, and why. Then decide whether to add more clients.
Client documents stay in Europe
A pilot means uploading client material. Here is what happens to it.
Processed in the EEAThe platform runs on Clever Cloud in France. Contract files are replicated to OVHcloud in Germany for disaster recovery, as sealed ciphertext.
Never used to train modelsLexnus does not train on your contract text, and does not let its AI sub-processors do so. That is a commitment in the DPA, not a setting.
Named sub-processorsEvery system that touches contract text is listed by name on the Security page.
You stay the controllerLexnus processes personal data as your processor under the GDPR and acts only on your instructions.
Questions from law firms
Does Lexnus give legal advice?
No. It applies rules your lawyers approved for the client, and shows where a draft departs from them. What to accept, what to counter and how to advise the client stays with the lawyer.
Can a client have different positions by contract type or deal size?
Yes. Use a separate playbook for each contract type. Within a playbook, rules and clause variations can carry conditions, such as a different cap above a certain contract value.
What happens when the client changes its instructions?
Edit the rule and publish a new version of the playbook. New reviews use it. Earlier results keep the version they were checked against, so the file still makes sense.
What if Lexnus can’t tell from the text?
It marks the point for review instead of guessing. You see which rule it relates to and what was missing or unclear, and a lawyer decides.
Do we have to build each client’s playbook from scratch?
No. Import the client’s existing guidelines from a Word document. Lexnus proposes rules from it, and your lawyers review and edit each one before publishing.
Where is client data processed?
Inside the EEA, on European infrastructure. Contract text is never used to train models. Both are commitments in our DPA. The Security page names every sub-processor.
Start with one client and one contract type
Bring the client’s current guidelines and a few representative agreements. We’ll test whether Lexnus applies the positions your team would apply in a real review.