WorkflowTrading & Execution

Buying · choosing the software

ISDA negotiation software lawyers will actually use

ISDA negotiation software earns its place when lawyers trust its reading enough to start from it, and they trust it when every value cites the dealer's page and every decision stays theirs. This guide sets out the requirements that decide adoption, what it must do that a comparison tool cannot, how to test it on your own files, and where ISDA Negotiation fits and where it does not.

Updated · 13 min read

See one case, screen by screen →
Watch with transcript →

ISDA negotiation software is software a fund's legal team uses to negotiate ISDA Schedules, credit support annexes and related agreements with dealers: it takes in each dealer's draft, extracts the negotiated terms, compares them with the fund's playbook and with what the dealer accepted before, drafts the redline, routes each term to its owner for approval, and keeps a trail of who decided what. The hard part is not building those features but getting derivatives lawyers to rely on them, because a lawyer who has to re-read the dealer's draft to check the software saves nothing. Lawyers use software that shows its reading, cites every value to the page, and leaves every decision with them. Botminds' ISDA Negotiation was built to those requirements: seven agents do the reading, placing, precedent search, drafting and checking; the lawyer decides; owners approve; the deputy general counsel releases.

The examples on this page come from a fictional sample firm, Quillfield Capital. Supervisors' statements and research are linked where they are used.

What ISDA negotiation software has to earn

33%

the highest hallucination rate measured in commercial legal research tools (lowest 17%), against providers’ claims of far less

J. Empirical Legal Studies, 2025

1,846

executed agreements across 212 funds and mandates and 38 dealers in the sample firm’s precedent library

fictional sample

120

executed agreements in the golden set the sample firm’s agents are evaluated on

fictional sample

41,880

append-only trail entries since go-live in the sample firm, kept 10 years

fictional sample

Magesh, Surani, Dahl, Suzgun, Manning and Ho, ‘Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools’, Journal of Empirical Legal Studies (2025). Sample figures from ISDA Negotiation.Source: Journal of Empirical Legal Studies, 2025

What is ISDA negotiation software?

ISDA negotiation software supports the negotiation itself, from the dealer's draft to the released turn, rather than storing the signed result. That separates it from three neighbours a fund probably already has. A document comparison tool shows what changed between two versions. A contract repository holds signed agreements and their data. The industry negotiation platform carries drafts between the parties. None of them tells a lawyer whether a dealer's 20% NAV trigger is acceptable for this fund, what the dealer gave a sister fund, or who must approve the answer.

Those three questions are the job. They need the fund's playbook as a scale for each term and fund type, its executed agreements as a precedent library, and its approval rules as routing. Software that brings the three together around a dealer's draft is negotiation software; software that does one of them is a component.

What are the requirements for ISDA negotiation software?

The requirements that decide whether lawyers use ISDA negotiation software are about trust, not features: every value cited to its source, every decision with a named person, nothing sent or booked without approval, and a trail an auditor can read. The table sets out ten, with the reason each matters.

Ten requirements for ISDA negotiation software, and why each decides adoption

RequirementWhat it means in practiceWhy it matters
Reads the whole packageSchedule, both credit support annexes, account control agreement, local-law masters, in the original languageA term in one document often depends on another
Cites every valueDocument, clause and page for each term; one click opens the passageLawyers check the source, not the software
Asks when unsureA confidence floor below which a person confirmsA guessed threshold is worse than a missing one
Playbook as a scalePreferred to walk-away for each term and fund type, versionedTurns a memo into decisions that can be measured
Precedent from executed agreements onlyWhat this dealer signed with the firm’s funds; superseded and old agreements flaggedThe strongest counter is the dealer’s own history
Checks the package for consistencyDefinitions, numbers and times across documentsA definition elsewhere can quietly undo a counter
Routes to owners, separates dutiesThresholds to credit, collateral to treasury; a maker never approvesCredit and collateral terms are not the lawyer’s alone
Never sends or books on its ownRelease by a named person; collateral updates after a treasury checkOne unreviewed turn can cost a term for years
Keeps a trailWho, role, what, why, evidence, for people and agents alikeSupervisors and auditors ask who conceded what and why
Puts signed terms into systems accuratelyField-by-field comparison with the collateral system before any updateSupervisors expect accurate legal terms in systems, not only in files
Our list, drawn from how ISDA Negotiation is built and from the supervisory expectations cited on this page: the Federal Reserve’s SR 11-10 on legal documentation in systems, CFTC 17 CFR 23.504 on documentation audits, and FINRA’s 2026 report on AI agents.Source: Federal Reserve SR 11-10; 17 CFR 23.504

Two of the ten are regulatory before they are commercial. US banking supervisors expect dealers to control legal documentation and incorporate "accurate legal terms and provisions" into their systems (Federal Reserve SR 11-10), and the CFTC requires swap dealers to have documentation policies audited periodically for "any material weakness" (17 CFR 23.504(c)). A fund's own collateral and risk systems face the mirror-image problem, and the auditors who test them will ask for the trail.

What should ISDA negotiation software do that a comparison tool does not?

It should say what a change means, not only that it happened. A comparison tool tells a lawyer that a dealer's third turn added a phrase; negotiation software tells her that the phrase sits outside the fund's playbook, that the dealer did not mention it in its cover note, and that it goes to treasury for approval. Both are useful; only one shortens the negotiation.

A comparison tool and negotiation software on the same dealer turn

Document comparisonISDA negotiation software
Input

Document comparison: Two versions of one document

ISDA negotiation software: The whole package, the playbook, the precedent library and the approval rules

Finds

Document comparison: Inserted and deleted text

ISDA negotiation software: Each negotiated term, its value, its page, and whether it is on, near or off the playbook

Unannounced changes

Document comparison: Shown as a redline if both versions are loaded

ISDA negotiation software: Flagged as a change the dealer’s cover note did not mention, with its playbook position

Across documents

Document comparison: One document at a time

ISDA negotiation software: Definitions and numbers checked across the Schedule and both CSAs

Precedent

Document comparison: None

ISDA negotiation software: What the dealer accepted with the firm’s other funds, cited

Decision

Document comparison: Left to the lawyer, unrecorded

ISDA negotiation software: Taken by the lawyer with a reason code, approved by the term’s owner

Illustrated by a fictional sample turn in ISDA Negotiation: a dealer’s third turn on a VM CSA added “unless the amount is less than USD 50,000” to Paragraph 13(g)(ii) without mentioning it.

The difference shows on precedent most of all. In the sample case, the agents searched 1,846 executed agreements and anchored six counters on one: the same dealer's 2024 agreement with another of the firm's funds, which had accepted a 30% one-month NAV trigger where the new draft asked 20%. The lawyer opened the source, saw the 2024 wording on page 7, and accepted the counter with the reason code "Matches the dealer's precedent with Quillfield".

Every counter in ISDA Negotiation opens its sources, from the dealer’s own executed agreement

Sources panel for the one-month NAV decline: six sources. The approved playbook entry; source 3, QF-ISDA-0918, Thornbury Bank N.A. with Quillfield Multi-Strategy Alternatives Fund, Part 1(h)(i), page 7, executed March 14, 2024, quoting that the Net Asset Value of Party B declines by 30% or more from the Net Asset Value as of the last Business Day of the immediately preceding calendar month; source 4, the draft redline striking 20% for 30%; source 5, a 30% median across 12 macro agreements in the library; source 6, Thornbury's Part 5(n) definition of NAV Trigger Event as first redlined, marked fixed. A footer notes that opening a source is logged.
Fictional sample in ISDA Negotiation. Each source carries a fingerprint of the stored passage, and opening it is logged, so a reviewer can see what the lawyer looked at before deciding.

Does ISDA negotiation software replace the industry negotiation platform?

No. The industry negotiation platform carries drafts between firms; negotiation software decides what the fund says in them. In ISDA Negotiation, drafts and new turns come in from legal email, dealer portals, the industry negotiation platform and document management, read-only, and are matched to a fund and a negotiation before anything is read. The released redline goes back the way the fund's lawyers choose.

The same applies to the fund's other systems. Signed agreements are filed in document management; legal entities and dealer limits come from the counterparty database and credit system; fund NAVs and holdings come from portfolio accounting, which is how the sample negotiation knew that a 10-year cap on eligible Treasuries would have stranded $610 million of the fund's $1.84 billion of Treasuries. Software that asks a fund to move its documents into a new repository before it can negotiate will wait a long time for adoption.

How do lawyers come to trust ISDA negotiation software?

Lawyers trust negotiation software when they have checked it against their own work and found it right, and when they can see that it never decides for them. The order matters: first prove the reading on past packages, then use it on live first turns with every decision still manual, then let the routing and the trail carry the approvals. The checklist sets out the steps and what each proves.

Adoption in six steps, and the trust each one earns

  1. Back-test on past packages

    Run dealer drafts the fund has already negotiated; compare the agents’ placements and counters with what its lawyers did. Earns: proof on the fund’s own files.

  2. Score extraction against the lawyers

    For each value, did the cited page say what the software said? Earns: a measured error rate, not a vendor’s claim.

  3. Live first turns, decisions manual

    The agents draft; the lawyer decides every term and edits freely. Earns: lawyers see the reading before relying on it.

  4. Turn on owner routing

    Thresholds to credit, collateral to treasury, exceptions with two names. Earns: approvers see the clause and precedent, not an email.

  5. Watch overrides

    Every person can override an agent with a reason code; read the reasons weekly. Earns: the fund knows where the agents are weak.

  6. Let the trail carry the audit

    Internal audit reads one append-only trail. Earns: the second line trusts it too.

Our adoption sequence, as used for ISDA Negotiation. It follows the research finding that providers’ accuracy claims for legal AI tools can be ‘overstated’ (Magesh et al., 2025): measure on your own files first.Source: Journal of Empirical Legal Studies, 2025

The first two steps are the ones vendors would rather skip. The researchers who tested commercial legal research tools found "the providers' claims are overstated", and that the tools differed substantially "in responsiveness and accuracy" (Magesh et al., 2025). A fund that measures the tool on its own executed agreements, with values its lawyers agree, has a number it can defend; one that relies on a demonstration does not. Broader benchmarks help frame the question but not answer it: LegalBench, whose tasks were written by legal professionals, tests general legal reasoning, not a fund's playbook (Guha et al., arXiv, not peer reviewed; how to measure a legal search tool on ISDA clauses builds the fund's own test).

Supervisors expect the same discipline. The SEC's examiners note that registrants already use generative tools of what the staff call "super intelligence" (SI) for drafting among other work, and say they will check that firms still meet the obligations they already had, which do not depend on the technology (SEC Division of Examinations, Fiscal Year 2027 Examination Priorities, October 2026). FINRA has said the same of its own rules (Regulatory Notice 24-09), and ESMA warns against relying on AI so far that human judgment is set aside (ESMA, 2024). Software that leaves the judgement with the lawyer makes those obligations easier to meet, not harder.

How long does it take to roll out ISDA negotiation software?

Rollout time depends on what the fund already has written down, and no regulator or independent study publishes a benchmark. The inputs that decide it are the fund's playbook memos for each fund type, its clause library, its approval rules and its executed agreements. A fund whose playbook is a set of memos needs them turned into scales, term by term, before any software can place a dealer's term on them.

ISDA Negotiation is built from exactly those inputs and goes live after its quality checks pass, including a back-test on the fund's past negotiations. The sample firm went live in July 2026; its reports on sample data since then show a median of 0.8 business days to a first turn, against 5.0 by hand in 2025. Those are sample-data figures, not a customer's result, and the right expectation for any fund is the one its own back-test produces. Where those days went before is shown in ISDA negotiation delays.

How do you test ISDA negotiation software before buying?

Test it on a past dealer package of your own, against your own playbook and executed agreements, and score it the way your lawyers would. A demonstration on the vendor's documents proves the vendor's documents. A test on a package your lawyers already negotiated shows whether the software finds the terms they found, places them where they would, cites the right pages, proposes the counters they made, and catches the inconsistencies they caught.

Five measures make the test comparable across vendors: the share of negotiated terms found; the share of values that match the cited page; agreement with the lawyers' on, near or off placement; whether the precedent cited was the one the lawyers used; and the inconsistencies the second read found that the lawyers missed, or missed that they found. Add the boundary tests from the ISDA AI tool guide: try to send, try to concede beyond the playbook, ask for the trail. The concessions the test surfaces are also the start of measuring the playbook.

After signature, ISDA Negotiation checks every collateral term before the system is updated

Check and book panel for QF-IMCSA-0212, Pemberton Global Markets with Quillfield Global Macro Fund, 2018 IM CSA and account control agreement executed September 25, 2026. The Booking agent compared seven collateral terms with the collateral system: method ISDA SIMM, threshold allocated to the fund $6,000,000, minimum transfer amount $250,000 shared with VM, eligible collateral cash USD and Treasuries of all maturities, custodian Calder Trust Company and notification time 11:00 a.m. New York all match; valuation percentage for Treasuries over 10 years differs, 96% in the signed CSA against 94% in the system from a template default. Buttons: return to legal, approve push with 96%.
Fictional sample in ISDA Negotiation. The difference came from a template default in the collateral system, not from the negotiation; nothing is written until treasury approves.

Where ISDA Negotiation fits, and where it does not

ISDA Negotiation fits a fund manager that negotiates ISDAs and credit support annexes with many dealers for several fund types, has or will write a playbook with preferred and walk-away positions, and wants credit risk, treasury and the general counsel's office to approve the terms they own with evidence beside them. It reads ISDA Schedules, variation and initial margin CSAs, account control agreements and local-law masters, and, as the sample firm's intake shows, repo and clearing drafts as well.

It is not a contract repository for every agreement the firm signs, and it does not decide questions of law or business: whether a trade is in scope of a clearing rule, whether a fund should use a dealer at all. It does not replace the fund's lawyers, its playbook or its approvers; it gives them the reading, the precedent and the draft. A fund that negotiates a handful of ISDAs a year with no playbook may get more from writing the playbook first.

ISDA Negotiation’s audit trail: who conceded what, why, and who checked it

Audit trail screen: append-only, who did what, when, why and on which evidence, people and agents alike, each entry with a fingerprint chained to the one before. 22 of 22 entries for October 16 to 19, 41,880 since go-live, retained 10 years. Suggested questions: who conceded beyond the playbook this year; show every decision on NEG-2026-0147; did anyone approve their own work. Entries include Victoria Hale approving set-off exception EX-2026-0412, Mei Lin Tan approving collateral terms, Tomás Herrera approving threshold counters, Claire Whitcombe deciding the remaining terms with 6 counters accepted, 13 kept and set-off conceded as an exception, and Intake agent entries logging new dealer drafts, one below the 90% auto-match threshold sent to the analyst.
Fictional sample in ISDA Negotiation. Internal audit can ask the trail in plain words, verify the chain and export it, read-only.

Questions people ask

Is ISDA negotiation software the same as contract lifecycle management?

No. Contract lifecycle management covers many kinds of contract across their life; ISDA negotiation software is specialised in negotiating derivatives documentation term by term, against a playbook and precedent, with credit and collateral approvals.

Can ISDA negotiation software handle the uncleared margin documents?

It should, because initial margin terms interact with the Schedule and the variation margin CSA. ISDA Negotiation reads the 2018 IM CSA and the custodian's account control agreement with the rest of the package and flags differences, such as a notification time of 10:00 a.m. in one CSA and 11:00 a.m. in the other.

Who should own ISDA negotiation software inside a fund?

Legal, with the deputy general counsel or head of derivatives documentation as the business owner, because the playbook, exceptions and releases sit there. Credit risk and treasury own their routing rules, and an administrator who cannot approve terms manages settings.

Does ISDA negotiation software work with outside counsel?

It can. In ISDA Negotiation outside counsel see only the negotiations shared with them, and their work goes into the same trail as the fund's own lawyers'.

What does ISDA negotiation software cost?

Vendors don't publish prices, and the right comparison is the fund's own cost of a negotiation: lawyer and approver hours per first turn, turns per agreement, and the value of concessions made under deadline. Measure those first, so any price can be judged against them.

ISDA Negotiation.
See it on your own files.

An agentic solution from Botminds. Compares a dealer's ISDA Schedule and CSA to your playbook and drafts the redline for legal Book a demo, or explore how it works.