Your contract process is an inbox

A reply chain, or an agreement record

Email is excellent at delivering messages and structurally incapable of holding a negotiation. These are not usability complaints; they are consequences of the medium.

• Email and attachments: Every revision is a new file, so the current version is a matter of opinion
• Email and attachments: Reasoning lives in prose around the attachment and is gone the moment the thread is archived
• Email and attachments: Approvals are sentences in replies, unbound to any particular version of the document
• Email and attachments: Each party keeps a private copy, and nobody can be sure the copies match
• Email and attachments: Follow-up depends on someone remembering to chase
• Email and attachments: Obligations and renewal dates exist only in the document, so they are found again by accident
• Paqt: One live agreement with clause-level version history and a single current state
• Paqt: The conversation sits beside the clause it is about, permanently attached to it
• Paqt: Approvals bind to a specific version and an identity, so acceptance is evidenced
• Paqt: Both parties work in the same room with per-party visibility — no private forks
• Paqt: An AI assistant handles the cadence of follow-up and the ordinary turns under a mandate
• Paqt: Obligations, notice periods and renewal windows are extracted at execution

The five failures everyone recognises

None of these are anybody's fault. They are what happens when a negotiation is stored as messages rather than as an agreement.

• Version drift — Two people editing two copies of the same attachment produces two agreements. Reconciling them is manual, late and the point at which real mistakes get signed.
• Lost reasoning — The explanation for a concession was in a reply that is now three forwards deep. A year later the clause is there and the reason it is there is not.
• Unbound approval — “Approved” in an email refers to whatever the sender was looking at. If the document moved afterwards, the approval silently covers something nobody agreed to.
• Uncontrolled distribution — An attachment is a copy. Once sent, it exists outside any access control you have, in mailboxes you will never see again.
• Commitments that resurface late — Notice periods, auto-renewals and service obligations are in the file. Unless someone extracted them deliberately, the first reminder is the consequence.
• Throughput tied to attention — Every turn needs a person to notice it. Agreements stall for days not because anyone disagreed but because the reply sat below the fold.

What the same agreement looks like here

The familiar parts stay familiar — it still starts as a conversation and still ends with a signature. What changes is that each step leaves something behind.

• Start in a room, not a thread — Invite the counterparty by link. The discussion happens in chat, in the same place the document will live, so the context never has to be forwarded anywhere.
• Draft in place — The agreement is generated from your template inside the room. There is no attachment, so there is no second copy and no question about which file is current.
• Negotiate clause by clause — Positions are proposed, countered and accepted at the level of the clause, each one attributed to a party and a time, with the discussion attached to the clause itself.
• Let the assistant take the ordinary turns — Under a mandate you set, it responds to standard deviations, chases the counterparty and brings you only the terms you said you wanted to decide.
• Sign where you negotiated — Signature happens in the room with identity capture, and the executed agreement keeps its full negotiation history rather than becoming a flat file.
• Keep the answers — Obligations, notice windows and renewal dates are extracted from the executed agreement, and amendments later resolve against the clauses they change.

The habits this makes unnecessary

• No more: Naming files v4-final-FINAL-2
• No more: Searching your inbox for the latest draft
• No more: Forwarding a thread so someone has context
• No more: Asking which copy the counterparty is reading
• Instead: One current state per clause, always
• Instead: The discussion attached to the clause it concerns
• Instead: Per-party visibility inside one shared record
• Instead: A diff between any two versions on demand
• And afterwards: A timestamped, tamper-evident history
• And afterwards: Approvals bound to a version and an approver
• And afterwards: Agent actions attributed with their mandate
• And afterwards: The whole record exportable as one package

What email still does better

Email is universal, and that is a real advantage no product can match. The sensible move is not to abolish it but to stop using it as the place the agreement lives.

Counterparties join a room by link, so nobody has to buy anything or install anything to negotiate with you. The first agreement is the whole evaluation — there is no migration to do first.

• Email keeps: Universal reach — everyone has it and nobody has to be invited
• Email keeps: The informal conversation that happens before there is an agreement to negotiate
• Email keeps: Correspondence that is not about a specific clause and does not belong in the record
• Email keeps: A one-off document going to someone you will never transact with again
• Move into a room when: There is a document both sides are going to change
• Move into a room when: More than two people need to agree before it can be signed
• Move into a room when: You will need to answer questions about this agreement after it is signed
• Move into a room when: The same agreement type comes up often enough that an assistant could run it

The practical questions

• Will the counterparty actually use this? They join by link into a chat room with the document in it, which is a lower ask than installing anything or learning a repository. In practice the thing that persuades the other side is that they stop receiving attachments and start seeing one current version.
• What if they insist on email? Then you lose the clause-level record for that negotiation, which is a real cost and worth naming. The usual middle path is to run the agreements you control in a room — your own paper, your renewals, your NDAs — and keep email for the counterparties who will not move.
• Can we bring in a draft that already exists? Yes. Import the document into a room and negotiate from there. It gains clause-level history from that point forward; it will not retroactively acquire the history of the email thread that produced it.
• Is the chat part of the legal record? The room's history is timestamped and tamper-evident, and the discussion attached to a clause stays attached to it. That is usually the point people most want and least expect — the reasoning survives alongside the text.