Legal
Knowing where every contract is, without asking three people
The question that eats a legal team's week is "where is that one now?". A worker keeps the answer current and ages every stalled agreement.
What the worker does
An AI worker in Polaris tracks contract review by reading the negotiation threads in a connected Gmail account and the document versions in Google Drive, then maintaining one task per agreement with its current state, who is holding it, and how many days it has been there. It posts an ageing report as a comment. Lawyers do the reviewing, the negotiating and the signing.
- Connections
- Gmail · Google Drive
- One task
- Per agreement, from request to signature
- Reported
- State, holder, days waiting
Contract state is knowledge that lives in one person's head
A deal is waiting on redlines. Is it with us or with them? Did the customer's counsel respond last Tuesday? Did anyone tell sales that the indemnity question was answered a week ago?
In most companies the answer to all three is that one person knows, and finding out costs a message and an hour of latency. When that person is on holiday, the answer costs a day.
The information is not hidden. It is in a mail thread and in a folder of files named v3, v3-final and v3-final-ACTUAL. Reading both and writing down the state is a job with no judgment in it, and it is the job that makes a legal queue visible.
The states a worker tracks
Use your own names for these. What matters is that every agreement is in exactly one of them and the clock is running.
| State | What the worker records | The question it makes answerable |
|---|---|---|
| Requested | Who asked, what for, what the commercial terms are | How many requests came in this month, and from where |
| With us | Which reviewer, since when | Are we the bottleneck right now |
| With them | Which version was sent, on what date | How long has the other side been sitting on it |
| Open point | The point, quoted from the thread, and who owes an answer | What is actually blocking this one |
| Ready to sign | Which file is final and where it is | Is the version about to be signed the version that was agreed |
| Signed | The executed file and its location | Can anyone find the signed copy in ten seconds |
Tracking versus reviewing
The worker
- Reads threads and files to determine current state
- Ages every agreement and flags anything past your threshold
- Quotes open points verbatim from the correspondence
- Chases the internal owner who owes an answer
- Assembles the file before a review session
The lawyer
- Reads the document
- Decides what is acceptable
- Writes the redlines
- Advises the business on the trade
- Approves execution
Two habits that make the tracking trustworthy
Make the worker cite the evidence for the state it recorded
"With them since 12 Aug" is a claim. "With them since 12 Aug, per the mail sent at 16:42 attaching v4" is a fact someone can check.
Never let it infer that a point is closed
An unanswered question in a thread is open until a human says otherwise. Brief it to list ambiguity rather than resolve it, because a wrongly-closed open point is the one error here that costs real money.
Questions people ask
+Does this replace a contract lifecycle management system?
It is not a CLM and does not pretend to be one. There is no clause library, no template automation and no e-signature. What it gives a small team is the part they are missing: current state, ageing and a chase, in the same workspace as everything else.
+Can it draft a first-pass redline?
Drafting text is something the worker can do, and whether you want that is a decision for your counsel, not for a marketing page. What Polaris guarantees is that nothing it drafts leaves the workspace without a person reading and sending it.
+What if negotiations happen in Slack rather than email?
Give it the Slack connection as well as Gmail. Slack messages can also arrive in the Polaris Inbox as prefilled task suggestions, so a commitment made in a channel becomes a tracked item after one click rather than being lost in scrollback.
+How far back can it reconstruct?
As far back as the connected mailbox and Drive folders go. The first run on an existing pile is the expensive one and usually the most revealing, because it surfaces the agreements nobody had realized were still open.
Related
Legal operations work an AI worker can carry
Tracking, chasing, assembling and watching. Everything a legal team spends time on that is not actually practicing law.
A register of what you actually signed
Finance tracks what a vendor costs. Legal needs to know what you promised them, what they promised you, and which of those promises has a date on it.
Compliance checklists with evidence attached and gaps named
A checklist where every line is ticked and nothing is evidenced is not a control. It is a document that will fail an audit slowly.
Proposals drafted from your call notes, not from a template
Every proposal is 70% the same document and 30% the reason this customer is different. The 30% is the part worth your evening.
Hire an AI paralegal
It organises the paperwork and surfaces the dates and clauses, and it stops well before the point where anything becomes advice.
Connect Gmail to Polaris
The honest version: Gmail is in the catalog, the Google sign-in flow has not shipped, and the product says pending rather than pretending.
Connect Slack to Polaris
Six scopes, no access to direct messages, and a bot that only reads the channels somebody invited it into.
Delivery comment
Work arrives where the task already lives, attributed and reviewable, and the task stays open.