Contract review
Can AI do the first review of supplier and customer contracts?
Last updated
Yes. AI can do the first review of supplier and customer contracts: it extracts key clauses, compares each one with your contract playbook, flags departures such as uncapped liability or auto-renewal, and records obligations and renewal dates in a register. A lawyer or contract manager reviews every flag and stays responsible for the advice and the signature. Run it privately and test accuracy on your own contracts first.
Who this is for
In-house counsel, contract managers, category managers and procurement leads in multi-entity groups across Australia and New Zealand, where supplier and customer contracts arrive faster than legal can review them and renewal dates live in people's inboxes. Contracts usually sit in SharePoint or a contract repository, with commercial terms mirrored in Business Central, NetSuite, Pronto Xi, SAP or Salesforce.
Solve8 delivers this in fixed-fee phases: discovery on one contract type in one business unit with a measured baseline, a shadow pilot on live contracts where your lawyers set the accuracy bar, a build on your infrastructure in Australia, then support or handover. A person approves every review. Workshops run on site and builds run remotely under your access controls.
How a first-pass review works
Intake
Contracts arrive from a legal inbox, SharePoint library or contract system, tagged with entity, counterparty and contract type.
Extract
AI reads the full document, including schedules, and pulls out each clause your playbook covers, with page references.
Compare
Each clause is compared with your standard position and approved fallbacks for that contract type.
Flag
Departures are rated against your risk levels, with the playbook's fallback wording suggested for a reviewer to accept or change.
Review
A lawyer or contract manager works through the flags beside the source text and decides what to accept, negotiate or escalate.
Record
Agreed terms, obligations and key dates go into the register once the contract is signed.
What AI checks against a playbook
| Clause | Typical playbook position | Flag raised |
|---|---|---|
| Limitation of liability | Capped at a multiple of contract value, with standard carve-outs | No cap, a cap below your minimum, or one-sided carve-outs |
| Indemnities | Mutual and limited to third-party claims | Broad or one-way indemnities |
| Term and renewal | Fixed term, renewal by agreement | Automatic renewal with a short notice window |
| Termination | Rights for convenience and breach on both sides | Only the counterparty can terminate |
| Payment terms | Your standard days and invoicing rules | Shorter terms, interest or price escalation |
| Data and privacy | Data stays in Australia, breach notice within your timeframe | Offshore processing or no breach notification |
| Insurance | Cover types and minimum limits from your policy | Missing cover or lower limits |
Obligations and renewal register
What is captured
Start and end dates, renewal and notice dates, price review dates, reporting duties, service levels, insurance requirements and the internal owner.
Where it lives
A SharePoint list, Power BI report or your contract system, linked to the source clause so anyone can check the wording.
Reminders that reach the owner
Notice deadlines go to the named owner ahead of the window, so every renewal is a decision someone makes.
Existing contracts
The backfile can run through the same extraction, with a person checking each register entry before it is relied on.
Privilege and confidentiality
Keep contracts out of public tools
The OAIC recommends that organisations do not enter personal information, particularly sensitive information, into publicly available generative AI tools. Contracts also carry pricing and confidential terms.
Privilege rules come from your lawyers
Reviewer notes and advice on a contract may be privileged. In-house counsel should decide how AI output and comments are stored, labelled and shared so confidentiality is kept.
Access follows the matter
The assistant only reads contracts the user is already allowed to see, and every query and output is logged.
Private AI, with legal still responsible
Contract review suits private AI: the application, document store and logs sit in your own cloud environment in an Australian region, model services are chosen for Australian processing, and the provider commits in writing that your contracts are not used to train any model. Legal stays responsible for the advice and the signature. The AI produces a first-pass review and a draft register entry, the reviewer accepts, edits or rejects each flag, and the record shows who decided.
Testing accuracy before anyone relies on it
Use contracts already reviewed
Build a test set from signed contracts your lawyers have marked up, across your main contract types and counterparties.
Measure misses and false flags
Count clauses the AI missed and flags that were wrong, clause by clause. Missed risks matter more than extra flags.
Set the bar first
Legal decides the accuracy needed per clause before the shadow pilot. Contract types that miss it stay with full manual review.
Retest after changes
Rerun the test set when the playbook, the model or the contract mix changes.
Related work
Private AI infrastructureAI that runs in your own environment in Australia, under your security policies.
AI for tender and RFP responsesFirst drafts from your past submissions, with a compliance matrix against the RFT.
Staff AI use policyWhat staff can put into which AI tools, and who approves new ones.
Data sovereignty in AustraliaWhere AI tools store and process data, and how to keep it onshore.
Sources
OAIC, Guidance on privacy and the use of commercially available AI products: https://www.oaic.gov.au/privacy/privacy-guidance-for-organisations-and-government-agencies/guidance-on-privacy-and-the-use-of-commercially-available-ai-products
Questions
Can AI review a contract against our standard terms and playbook?
Yes. Given a playbook with your standard positions and approved fallbacks for each contract type, AI can extract each clause, compare it with the playbook and flag departures with page references. Results depend on how clearly the playbook is written, so most projects tighten it first. A lawyer or contract manager reviews every flag.
Is it safe to upload contracts into ChatGPT or Copilot?
Public AI tools are the wrong place for contracts. The OAIC recommends against entering personal information into publicly available generative AI tools, and contracts also carry pricing and confidential terms. Use an enterprise tool your IT team has assessed, or a private deployment in your own environment in Australia, under a staff AI policy.
Does using AI on contracts affect legal professional privilege?
It can, depending on how outputs and reviewer comments are created, stored and shared. In-house counsel should set the rules before rollout, including where AI output is saved, who can see it and how it is labelled. Keeping the tool inside your environment with matter-level access helps keep communications confidential.
How accurate is AI contract review?
Accuracy varies by contract type, document quality and playbook clarity, so measure it on your own contracts. Run a test set of contracts your lawyers have already reviewed, count missed clauses and false flags, then shadow live reviews until the results meet the bar legal sets.
Who can set up AI contract review in Sydney, Melbourne or Brisbane?
Contract lifecycle management vendors suit organisations ready to move contracts onto a new platform. Where contracts stay in SharePoint or an existing system and must be reviewed privately, an AI consultancy builds the review and register around them. Solve8 does this across Australia and New Zealand, with workshops on site and legal owning the playbook.
Related
Custom AI developmentAI that reads, checks and drafts the first pass.
Our AI pilot worked in testing but nobody uses it six months later. What went wrong, and should we fund the next phase?Why an AI pilot that worked in testing goes unused months later, how to find the cause, and how to decide whether to fund the next phase, fix it or stop.
Managed AI servicesMonitored and improved after go-live.
We bought Microsoft 365 Copilot licences. How do we make sure it doesn't show staff files they shouldn't see?Microsoft 365 Copilot only shows what staff can already open, so oversharing is a permissions fix. Reports, labels and Purview steps to run before rollout.
Contracts waiting on legal review?
A 30-minute review with a senior consultant. You leave with a written one-page summary you can share internally.
Fixed-fee phases, agreed before we start. Stop after any phase.