AI that reads every clause, not just the headline.
Linkt deploys intelligence across review, diligence, and drafting without moving a document outside the privilege boundary. Your matters, your DMS, your ethical walls.
Privilege, the bar rules, and your client's guidelines.
Legal AI does not fail on drafting quality. It fails when nobody can answer what happened to the client's documents. We arrive with those answers.
Professional responsibility
Built to be supervised: every output is a draft for a lawyer to review, with a record of what the system read, so Model Rules 1.1 and 5.3 are satisfied by the workflow rather than by policy.
Privilege and work product
Privilege turns on disclosure to a third party. Nothing is disclosed: inference runs inside your tenancy, no vendor receives matter content, and nothing trains an outside model.
Your clients' outside counsel guidelines
OCGs increasingly restrict AI or require notice. We deploy per-firm or per-client, and hand your client's security reviewers an architecture diagram instead of a vendor questionnaire.
- ABA Model Rule 1.6
- Model Rules 1.1 & 5.3
- Outside counsel guidelines
- Attorney work product
- ISO 27001
- SOC 2 Type II
- GDPR
Where Linkt goes to work in a law firm.
The document-heavy work that consumes associate hours, done inside iManage or NetDocuments, cited back to the source page.
Contract review against the playbook
Redline against your playbook or your client's, with every deviation cited to the clause that triggered it and the fallback position you already approved.
Diligence and data rooms
Read the full data room, build the issue list, and cite each finding to document and page, so the associate verifies rather than reads.
Privilege review and logging
Draft the privilege log with a stated basis for each withholding, ready for a lawyer to sign, from the same review pass.
Matter research and memos
Pull authority and trace precedent grounded in your prior work product, not a general web index.
Deposition and transcript summaries
Turn transcripts into issue-tagged summaries with citations to line and page.
Conflicts and intake
Screen new matters against your existing book and flag the relationships a human should look at.
Billing narratives
Draft compliant time narratives from matter activity in the UTBMS and LEDES formats your clients require.
Inside the DMS, inside the wall.
Ethical walls are respected
Access mirrors your matter-level permissions and screens in iManage or NetDocuments. The system cannot read what the lawyer cannot read.
Nothing leaves the boundary
Deployed in your tenancy. No matter content reaches a vendor, and none of it trains anyone's model.
Every output is traceable
Each answer cites the document it came from, so review is verification rather than a second full read.
Straight answers for your reviewers.
- Does running AI over client documents waive privilege?
- Privilege is waived by disclosure to a third party. In a sovereign deployment there is no third party: the models run inside your tenancy, matter content never reaches Linkt's infrastructure, and the commitment is contractual as well as architectural. Your GC gets the architecture to review rather than a vendor's assurance.
- How does this sit with our clients' outside counsel guidelines?
- Many OCGs now require notice before AI touches a client's matter, and some prohibit third-party processing outright. Because the deployment sits inside your environment, the answer to "who else processed this" is nobody. We can also scope a deployment per client where their guidelines demand it.
- Does it work inside iManage or NetDocuments?
- Yes. We integrate with the DMS you already run, and access inherits your matter permissions and ethical walls. Read-only by default, with writes back to the workspace gated behind a human.
- What satisfies Model Rule 5.3 supervision?
- Nothing is filed or sent by the system. Every output is a draft with citations to the source, and the audit log shows what was read and when, which is the record a supervising lawyer needs to demonstrate reasonable supervision.
Put the document work where it belongs.
We get AI workflows through compliance, legal, and security and into production.