# AI Mediation and Negotiation — Competitive Audit Audit date: **2026-04-23** Scope: Phase 179 Concordia task `179.1.1.1`. This audit is a dated snapshot of the AI mediation, online dispute resolution (ODR), and autonomous bargaining landscape that Phase 179 Concordia must match and surpass. It is the first artifact in the Concordia research track and feeds directly into the source matrix (`179.1.1.2`), gap analyses against Mediator.ai (`179.1.1.3`) and Pactum / Nibble (`179.1.1.4`), the bibliography (`179.1.1.5`), and the research refresh gate (`179.1.1.6`). This document is not a product endorsement, legal opinion, or competitive intelligence report. Every claim below is attributed to a URL that was either fetched or searched on 2026-04-23. Where a source could not be reached (e.g., HTTP 403 from a bot shield), the substitute source is named and the gap is called out. Where a product markets a claim the audit could not verify independently, the claim is tagged `CLAIMED` rather than restated as fact. Category legend used throughout: - `commercial` — vendor marketing or product site - `standards` — multilateral or professional-body text (UNCITRAL, NIST, EU) - `regulatory` — government or regulator text (EU AI Act, sector regulators) - `academic` — peer-reviewed paper or widely cited preprint - `benchmark` — public evaluation harness (ANAC, GeniusWeb, NegMAS) - `case-study` — documented deployment or third-party analysis - `press` — journalism or analyst coverage The full `source-matrix.md` lives alongside this file and carries per-source tagging. This audit focuses on product posture, architecture claims, evidence, weaknesses, and last-updated signals. ## Table of contents 1. [AI mediation and cooperative-negotiation products](#ai-mediation-and-cooperative-negotiation-products) 1. [Mediator.ai](#1-mediatorai) 2. [MediationAI (mediationai.app)](#2-mediationai-mediationaiapp) 3. [Bot Mediation](#3-bot-mediation) 4. [ZODR (zodr.ai)](#4-zodr-zodrai) 5. [Disputell](#5-disputell) 6. [Dyspute.ai](#6-dysputeai) 7. [TheMediator.AI](#7-themediatorai) 2. [Commercial negotiation agents](#commercial-negotiation-agents) 1. [Pactum](#8-pactum) 2. [Nibble](#9-nibble) 3. [Decentralized dispute resolution](#decentralized-dispute-resolution) 1. [Kleros](#10-kleros) 4. [ODR platform and history reference set](#odr-platform-and-history-reference-set) 1. [ODR.com (Resourceful Internet Solutions / Colin Rule)](#11-odrcom) 2. [Modria and the Tyler Technologies ODR lineage](#12-modria--tyler-odr-lineage) 3. [eBay and PayPal ODR precedent](#13-ebay--paypal-odr-precedent) 5. [Institutional ADR providers](#institutional-adr-providers) 1. [AAA mediator search and AI-Native Arbitrator](#14-aaa-mediator-search-and-ai-native-arbitrator) 2. [JAMS AI Rules](#15-jams-ai-rules) 6. [Human mediator-assist tooling (survey)](#16-human-mediator-assist-tooling-survey) 7. [Cross-product observations](#cross-product-observations) 8. [Concordia posture implications](#concordia-posture-implications) 9. [Source inventory](#source-inventory) --- ## AI mediation and cooperative-negotiation products ### 1. Mediator.ai - **Category:** Consumer / SMB cooperative-negotiation tool. Closer to a negotiation-agreement generator than a mediation-as-a-service platform. - **Core value prop (verbatim):** "Cooperative negotiation is a solvable problem." And: "Mediator.ai finds agreements that two people in conflict would both accept, often ones they hadn't thought of themselves." (https://mediator.ai/) - **Target users:** Individuals in everyday two-party disputes — roommates, cofounders splitting equity, contractor/client payment issues. Positioned at consumers and SMBs, not enterprise. - **Architecture claims:** The blog post dated 2026-04-20 describes a hybrid architecture — (a) an LLM used as a *pairwise preference ranker* to infer per-party utility functions ("would this party prefer agreement A or B?"), (b) a *genetic algorithm* over candidate agreements with crossover, mutation, and selection, and (c) *Nash bargaining* as the scoring rule (maximize the Nash product of gains over BATNA). Individual mutators are isolated Lua scripts whose selection probability adapts to performance. - **Inputs:** Private text conversations where each party describes position, financials, BATNAs, and preferences. - **Outputs:** Draft agreements with concrete terms (payment splits, equity vesting, breakup clauses with dates). - **Autonomy:** Advisory. Drafts are presented for human review and signature; no autonomous commitment path. - **Multi-party support:** Two-party only, explicit throughout. - **Privacy / confidentiality:** Parties "walk through their side with Mediator privately." Informal private-caucus pattern. No formal privacy policy, encryption posture, or data-residency claim is visible on the site. - **Human-in-the-loop:** Parties themselves are the reviewers. No professional mediator role. - **Enterprise / compliance:** None claimed. No SOC 2, GDPR, EU AI Act, or UNCITRAL references on the site. - **Pricing:** Not disclosed. - **Traction:** Illustrative vignettes only (Priya / roommate, bakery partnership). No customer logos, no funding disclosure, no published outcome data. - **Evidence quality:** The architecture description is clear but self-published. No peer-reviewed paper, no benchmark submission, no third-party validation. - **Self-disclosed open risks:** The Nash-bargaining blog post itself names six risks — unstable inferred preferences, BATNA manipulation, specification gaming, enforceability, incomplete option discovery, and leverage-seeking counterparties. These map closely to the risk vectors Concordia must cover. - **Last-updated signal:** Blog post dated 2026-04-20, three days before the audit date. - **Pull quotes:** - "Cooperative negotiation is a solvable problem." — https://mediator.ai/ - "Mediator.ai finds agreements that two people in conflict would both accept, often ones they hadn't thought of themselves." — https://mediator.ai/ - On Nash: "Identifies the agreement that maximizes the product of their gains over what they'd get by walking away." — https://mediator.ai/blog/ai-negotiation-nash-bargaining/ - **URLs accessed 2026-04-23:** https://mediator.ai/ ; https://mediator.ai/blog/ai-negotiation-nash-bargaining/ ; https://mediator.ai/examples/bakery-partnership-agreement/ (404 at fetch time) ; http://mediator.ai/examples/ (301 redirect). ### 2. MediationAI (mediationai.app) - **Category:** Consumer ODR with on-chain settlement execution. - **Core value prop (verbatim):** "AI-Powered, Fair & Fast Dispute Resolution." (https://www.mediationai.app/) - **Target users:** Individual disputants, marketplace operators wanting plug-in escrow, crypto-fluent SMBs. - **Architecture claims:** NLP plus a claimed "legal ontology" parser, a GPT-4-class legal agent, smart-contract escrow (asset-agnostic across stablecoins, NFTs, fiat), an oracle layer bridging banking rails, and ML-based fraud scoring. Strong on-chain settlement story; thin on bargaining algorithm detail. - **Inputs:** Dispute narrative, structured evidence, KYC/KYB data, asset deposits into an escrow vault. - **Outputs:** "Court-enforceable" settlement contracts, automatic fund release, on-chain transaction records. - **Autonomy:** High. AI drafts proposed resolution; smart contract executes fund release after both parties sign. Minimal human intervention. - **Multi-party support:** Two-party only (one-to-one invite flow). - **Privacy / confidentiality:** Not explicitly addressed on the homepage. - **Human-in-the-loop:** Only the parties' own signatures. No neutral mediator role. - **Enterprise / compliance:** Claims automated KYC/KYB, AML screening, real-time fraud scoring; mentions an FDIC-insured fiat on-ramp via banking partner. No SOC 2 / GDPR / EU AI Act references. - **Pricing:** "$2 per party during beta" — flat fee. - **Traction:** Beta, waitlist, "coming soon" iOS. No named customers, no outcome data. - **Evidence quality:** Marketing copy only. "Court-enforceable" is a strong claim with no case law or jurisdictional detail backing it. - **Weaknesses:** Enforceability claim unverified; blockchain escrow raises jurisdictional enforcement questions; no disclosed smart-contract audit; no confidentiality posture documented; tiny team, no funding disclosed. - **Last-updated signal:** Footer "Decentralized Technology Solutions 2025." - **Pull quotes:** - "Resolve disputes for a fraction of traditional lawyer costs—just $2 per party during beta." — https://www.mediationai.app/ - "Once funds are on-chain, they can't be clawed back through card disputes or ACH reversals." — https://www.mediationai.app/ - **URLs accessed 2026-04-23:** https://www.mediationai.app/ . ### 3. Bot Mediation - **Category:** Legal-sector ODR for plaintiff / defendant civil and employment disputes. - **Core value prop (verbatim):** "Bot Mediation™ offers plaintiffs and defendants a speedy, fair, and effective way to resolve legal disputes." (https://botmediation.com/) - **Target users:** Trial lawyers, plaintiff / defendant firms, employment and civil litigation. Southern California based (Costa Mesa, CA), founded 2023. - **Architecture claims:** "Proprietary AI technology and algorithms"; AI-driven analysis informed by "comparable case data"; AI avatars as mediator; fully synchronous secure platform. Specific model and methodology not disclosed. - **Inputs:** Case summaries and structured intake from each side. - **Outputs:** Mediator proposals, settlement recommendations, case-value insights. - **Autonomy:** Semi-autonomous. AI mediates and can refer out to a human neutral when mediation fails — escalation is an advertised feature, not a fallback bolt-on. - **Multi-party support:** Two-party plaintiff / defendant model. - **Privacy / confidentiality:** Secure platform with a published privacy policy; specifics not detailed. - **Human-in-the-loop:** Yes. Escalation path to a qualified human neutral is a marketed feature. - **Enterprise / compliance:** Terms and privacy policy published. No SOC 2 / GDPR / EU AI Act certifications disclosed. - **Pricing:** Publicly referenced at $1,500 per side, contingent on settlement, in ABA *Law Technology Today* coverage and podcast interviews. - **Traction:** Named firm testimonials include West Coast Trial Lawyers, Payne & Fears LLP, Bartko Pavia LLP, and Lerman & Pointer LLP. Featured at the 2025 ABA Techshow. A PitchBook entry exists; funding not public. - **Evidence quality:** Firm-level testimonials and ABA coverage are real, but no aggregated settlement data or academic validation. - **Weaknesses:** "Proprietary algorithms" are unspecified — no Nash, BATNA, or explicit methodology stated; "comparable case data" source and licensing not disclosed; customer base California-centric. - **Last-updated signal:** Active blog; 2025 ABA Techshow mention. - **Pull quotes:** - "Bot Mediation™ offers plaintiffs and defendants a speedy, fair, and effective way to resolve legal disputes." — https://botmediation.com/ - "We reached a settlement in about an hour, and our Bot Mediation cost just a fraction [of traditional mediation]." — homepage testimonial - **URLs accessed 2026-04-23:** https://botmediation.com/ ; ABA *Law Technology Today* coverage and podcast transcripts via search. ### 4. ZODR (zodr.ai) - **Category:** AI-augmented co-mediation workflow layered over Zoom. Tool for professional mediators, not end disputants. - **Core value prop (verbatim):** "Transforming Your Zoom Into a Professional Mediation Office." Positioning phrase in coverage: "You provide the heart; ZODR provides the memory." (https://lminetwork.com/zodr/) - **Target users:** Solo mediators, ADR neutrals, courts and tribunals, and mediation practices that run Zoom sessions. - **Architecture claims:** Customized LLM that joins Zoom silently, performs real-time transcription, produces session summaries, flags settlement avenues, and suggests impasse-breakers. Overlay on Zoom rather than a replacement platform. - **Inputs:** Live Zoom audio / video sessions, case intake, uploaded documents, calendar data. - **Outputs:** Transcripts, session summaries, settlement-option suggestions, analytics; Stripe-powered invoicing; Google Calendar scheduling. - **Autonomy:** Explicitly advisory. "The AI Co-Mediator Assistant isn't here to replace you." - **Multi-party support:** Supports multi-party Zoom sessions (standard Zoom capability); no explicit N-party bargaining logic claimed. - **Privacy / confidentiality:** "Secure ZODR dashboard," "SOC 2-aligned controls" (aligned, not certified), user authentication. Multilingual interpretation mentioned. - **Human-in-the-loop:** Core design principle — the mediator retains all decision authority. - **Enterprise / compliance:** "SOC 2-aligned" language. No certifications or regulatory attestations disclosed. No EU AI Act posture. - **Pricing:** Not disclosed; waitlist gated via LMI Network. - **Traction:** Semi-finalist, 2026 ABA Techshow Startup Alley; featured on LMI Podcast Ep. 347 (Feb 2026); real-world testing inside active mediation practices. Built by Mac Pierrelouis (attorney / mediator, LMI Network founder). - **Evidence quality:** Early-stage. Podcast / competition visibility; no published case studies or outcome metrics. - **Weaknesses:** "SOC 2-aligned" is weaker than SOC 2 certified; dependency on Zoom API policies; recording / transcription raises jurisdiction-specific consent issues that the marketing does not address. - **Last-updated signal:** LMI Network ZODR page dated February 2026. - **Fetch note:** `https://www.zodr.ai/` and `https://zodr.ai/` returned HTTP 403 to our fetcher; site is live, but bot-shielded. Quotes and facts above come from the LMI Network page and podcast transcript. - **Pull quotes:** - "ZODR comprehensively reimagines what zoom mediation should have been from the ground up if it had been designed by mediators, for mediators." — https://lminetwork.com/zodr/ - "The AI Co-Mediator Assistant isn't here to replace you. It provides 'specialized intelligence' — transcribing sessions and suggesting settlement options." — https://lminetwork.com/zodr/ - **URLs accessed 2026-04-23:** https://www.zodr.ai/ (403) ; https://zodr.ai (403) ; https://lminetwork.com/zodr/ ; https://www.lmipodcast.com/ep347-giving-zoom-mediations-a-body-introducing-zodr-ai/ . ### 5. Disputell - **Category:** Pre-mediation intake and preparation software. Not mediation itself. - **Core value prop (verbatim):** "Better Mediation Starts Before Session One." (https://www.disputell.com/) - **Target users:** Practicing mediators running their own process, and their clients. B2B to mediators. - **Architecture claims:** Separate secure links per party, guided intake forms, mediator-only structured preparation reports. No specific LLM or algorithm branding. - **Inputs:** Per-party structured narratives, priorities, desired outcomes via gated forms. - **Outputs:** Mediator-only preparation report. Participants do not see it. - **Autonomy:** Explicitly non-autonomous. "Supports preparation only and does not make decisions or recommendations." - **Multi-party support:** Dual-party isolated intake. Not evident whether more than two parties are supported. - **Privacy / confidentiality:** Party isolation via separate links; participant input is not shared with the other party; mediator report is mediator-only. This is the strongest confidentiality posture in the set — it mirrors traditional private-caucus discipline. - **Human-in-the-loop:** Mandatory. The mediator drives the process entirely. - **Enterprise / compliance:** Terms of Use and Privacy Policy referenced. No certifications. - **Pricing:** Referenced ("review pricing") but not on the homepage. - **Traction:** None disclosed. - **Evidence quality:** Narrow positioning — no recommendations, no decisions — is its own evidence of scope discipline. No outcome data. - **Weaknesses:** Small surface area; hard to differentiate from a structured form; no visible design partners or adopters. - **Last-updated signal:** Image asset timestamp `1776965435` (Unix epoch → mid-April 2026); no explicit © date. - **Pull quotes:** - "Better Mediation Starts Before Session One." — https://www.disputell.com/ - "The mediator receives structured preparation output that participants do not see… supports preparation only and does not make decisions or recommendations." — https://www.disputell.com/ - **URLs accessed 2026-04-23:** https://www.disputell.com/ . ### 6. Dyspute.ai - **Category:** Consumer / SMB asynchronous AI mediation with optional human oversight. Small-claims focused. - **Core value prop (verbatim):** "Fast, affordable dispute resolution — powered by AI, driven by YOU." (https://dyspute.ai/) - **Target users:** Consumers and SMBs; designed for "smaller, two-party disputes involving less than $25,000." Channel partners include BBBs, community mediation centers, Bar Association of San Francisco, and startup legal providers (9to5 Docs, New Era ADR). - **Architecture claims:** AI mediator named "Adri." Adri v2 launched 2026-01-13 per LawNext — a full from-scratch rebuild replacing a no-code beta. "LLMs that do not train on user data." Async-first architecture using notifications rather than live sessions. Specific models not named. - **Inputs:** Each side's narrative, facts, desired outcomes; evidence uploads for settlement negotiation. - **Outputs:** Custom demand letters, AI-generated settlement proposals, and "legally-binding settlement agreements with e-signature" on resolution. - **Autonomy:** Medium. AI-first proposal generation; human mediator can "insert themselves as needed" via a provider dashboard. - **Multi-party support:** Two-party only (explicit in LawNext coverage). - **Privacy / confidentiality:** LLMs do not train on user data; platform distinguishes shared from confidential information (caucus model). Terms and Privacy pages published. - **Human-in-the-loop:** Optional. Mediator monitors via dashboard, can intervene on stalled cases. Channel design explicitly pairs Dyspute with existing mediation providers. - **Enterprise / compliance:** No SOC 2 / GDPR / EU AI Act claims disclosed. - **Pricing:** $299 per mediation (includes basic settlement agreement and e-signature). Pilot and early-adopter discounts offered. - **Traction:** Partnerships with 9to5 Docs and New Era ADR to embed AI mediation into startup legal agreements. LawNext coverage. Provider pilots with BBB, bar associations, and community mediation centers. - **Evidence quality:** Third-party press (LawNext) adds credibility. Early, with no published outcome data. - **Weaknesses:** $25k cap and two-party limit narrow the addressable market; enforceability of AI-drafted settlement agreements varies by jurisdiction; competitive pressure from TheMediator.AI at much lower price points. - **Fetch note:** `https://dyspute.ai/` returned 403 to our fetcher. Quotes below come from LawNext and cached search extracts. - **Last-updated signal:** LawNext piece dated 2026-01-13 announcing Adri v2. - **Pull quotes:** - "Fast, affordable dispute resolution — powered by AI, driven by YOU." — https://dyspute.ai/ - "Adri listens to both sides, analyzes their positions, and generates fair settlement proposals in minutes." — https://dyspute.ai/ - "If an agreement is reached, Dyspute.ai instantly generates a legally-binding settlement agreement, ready to sign — conveniently from any device." — https://dyspute.ai/ - **URLs accessed 2026-04-23:** https://dyspute.ai/ (403) ; https://www.dyspute.ai/ (403) ; https://www.lawnext.com/2026/01/dyspute-ai-launches-adri-v2-a-24-7-asynchronous-ai-mediation-platform.html . ### 7. TheMediator.AI - **Category:** Consumer ODR mobile app for everyday personal disputes. - **Core value prop (verbatim):** "Private • Impartial • Affordable — because no argument should cost $200 per hour to resolve." (https://themediator.ai/) - **Target users:** Individuals with family, divorce, workplace, or commercial personal disputes. Explicitly not intended for complex legal or corporate matters. - **Architecture claims:** A "trained large language model" positioned as impartial evaluator. Specific model and training data not disclosed. No formal bargaining algorithm (Nash, GA) mentioned. - **Inputs:** Per-party answers to structured mediator questions via mobile app. - **Outputs:** Potential resolution recommendations; PDF export of dispute progression and outcomes for each party's records. - **Autonomy:** Advisory only. "Final decision rests with you and the other party." - **Multi-party support:** Two-party. App reaches out to the other party for response. - **Privacy / confidentiality:** Each party's conversation with the AI is private and not shared with the other (private-caucus pattern). Conversations deleted 30 days after outcome. No personal data shared unless legally required. - **Human-in-the-loop:** No professional mediator in the loop. Parties themselves are reviewers. - **Enterprise / compliance:** "Strong security safeguards" stated; no SOC 2 / GDPR / EU AI Act certification claims. - **Pricing:** $4.99 per full mediation cycle, charged to the initiator. - **Traction:** Homepage usage counters display "0" — suggesting pre-launch or very early. iOS (TestFlight) and Android (beta). - **Evidence quality:** Marketing copy. No outcome data, no third-party reviews in search. - **Weaknesses:** Usage metrics at zero; self-disclaims utility for complex legal matters such as corporate disputes; confidentiality of AI conversations depends entirely on retention policy and is not backed by formal encryption or SOC claims. - **Last-updated signal:** Footer "© 2023-2026 Underlabs Inc." Active blog posts through 2024; Google Play listing live. - **Pull quotes:** - "Private • Impartial • Affordable — because no argument should cost $200 per hour to resolve." — https://themediator.ai/ - "Not intended for complex legal matters such as corporate disputes." — https://themediator.ai/ - **URLs accessed 2026-04-23:** https://themediator.ai/ ; Google Play listing via search triangulation. --- ## Commercial negotiation agents ### 8. Pactum - **Category:** Enterprise procurement negotiation. Agentic AI for supplier negotiations. Distinct from consumer mediation, included as the clearest commercial analog to autonomous bargaining agents. - **Core value prop (verbatim):** "Procurement AI agents that assist buyers and execute supplier negotiations at scale." On the agents page: "These Agents operate 24/7/365, capturing value from mid-tier and tail spend." (https://pactum.com/ and https://pactum.com/procurement-agents) - **Target users:** Global 2000 procurement teams, CCOs, and CPOs at Fortune 500s. - **Architecture claims:** AI agents embedded in enterprise systems via standard APIs; integrate with ERP / procurement / CMS; chat-style negotiation with suppliers; policy-bounded decisioning. Specific LLMs and algorithms not disclosed publicly. Pactum's pre-LLM era used rule-based dialog; the current-gen platform is LLM-based. - **Inputs:** Requisition data, supplier master data, price lists, payment terms, rebate structures, contract clauses. Claims 2–4 week deployment. - **Outputs:** Negotiated contracts, price adjustments, payment-term extensions, rebate agreements, full audit trails of each negotiation. - **Autonomy:** Variable and configurable. "Autonomously or with buyer approval" within guardrails set by procurement policy. Highest effective autonomy in this audit. - **Multi-party support:** Bilateral per negotiation; runs many bilateral negotiations in parallel. No multi-party coalition logic. - **Privacy / confidentiality:** Trust portal at trust.pactum.com. SOC 2 Type II certified. - **Human-in-the-loop:** Procurement teams set strategy and policy; agents execute; escalation to buyer approval configurable. - **Enterprise / compliance:** SOC 2 Type II certified (confirmed on procurement-agents page). Full negotiation audit trails. GDPR / EU AI Act posture not exposed on marketing pages but presumably covered in enterprise contracts. - **Pricing:** Not disclosed (enterprise sales). - **Traction:** Named customers include Walmart, Honeywell, Bristol Myers Squibb, Veritiv, Suez, Linde, Maersk (historical), Novartis, Tetra Pak, Mediclinic, Otto, Global Industrial, Vallen. Walmart case: 3% average savings and +35 days payment terms. 60+ Global 2000 enterprises. $54M Series C in June 2025 led by Insight Partners ($100M+ total). Reported 489% YoY spend growth handled, 2.5× ARR. - **Evidence quality:** Strongest in this cohort. *Fortune*, *Sourcing Journal*, *Procurement Magazine*, named logos, disclosed outcome metrics, SOC 2 Type II, institutional venture backing. - **Weaknesses:** Not a mediation tool. Bilateral distributive negotiation, not multi-party or conflict-resolution. Category-specific (procurement spend). Specific bargaining algorithm not published. - **Last-updated signal:** © 2026 Pactum AI, Inc.; Series C news 2025; active blog. - **Pull quotes:** - "Procurement AI agents that assist buyers and execute supplier negotiations at scale." — https://pactum.com/ - "These Agents operate 24/7/365, capturing value from mid-tier and tail spend." — https://pactum.com/procurement-agents - "Pactum maintains SOC 2 Type II certification, demonstrating enterprise-grade security controls for data protection, system availability, and confidentiality." — https://pactum.com/procurement-agents - **URLs accessed 2026-04-23:** https://pactum.com/ ; https://pactum.com/procurement-agents ; triangulated via Fortune, Procurement Magazine, and Investing.com search. ### 9. Nibble - **Category:** Commerce- and procurement-side negotiation chatbot. Originally DTC e-commerce "make-an-offer" chatbot; expanded into B2B procurement. - **Core value prop (verbatim):** "The world's most experienced AI negotiation agent"; "Integrate human-centric negotiations at scale." (https://nibbletechnology.com/) - **Target users:** DTC e-commerce brands (Shopify / Magento / Adobe Commerce), B2B SaaS sellers, procurement teams using Coupa and SAP Ariba. Both buy-side and sell-side. - **Architecture claims:** "Agentic AI system" with custom LLM guardrails where "pricing functions" are kept "secure, controlled and never seen by any LLM" — i.e., the LLM handles language but a deterministic pricing engine holds the numbers. Academic-research-backed negotiation performance claimed; specific base models not named. - **Inputs:** RFQs, price lists, supplier proposals, contract term sets; on the e-commerce side, shopper offer messages. Plugins for Shopify, Magento, Adobe Commerce, Coupa, SAP Ariba, and CRMs. - **Outputs:** Completed negotiated deals (price, quantity, payment terms, rebates), buy-back agreements on e-commerce, harmonized contract terms on procurement. - **Autonomy:** High. Handles mass renegotiations autonomously via self-serve portal or API; negotiates with hundreds of suppliers or thousands of shoppers in parallel. Managed-service option available. - **Multi-party support:** Bilateral per conversation. Designed for massive parallelism ("negotiate with 100s of suppliers in minutes" / "30,000 negotiations every month"). - **Privacy / confidentiality:** "Your data is your data"; data not used to train other implementations. ISO 27001 certified. - **Human-in-the-loop:** Procurement / commerce teams set boundaries and strategy; AI executes within them. Framed as "augment, not replace." - **Enterprise / compliance:** ISO 27001 certified. Privacy policy published. No explicit SOC 2 or EU AI Act posture in fetched copy. - **Pricing:** Not disclosed on homepage; GetApp listing exists. Plugin-based deployment. - **Traction:** 350+ organizations; ~30,000 negotiations per month; 2M+ cumulative automated negotiations; deal range $5 to six figures. Plugins on Shopify App Store and Magento / Adobe Commerce marketplace. Hashtag Paid press coverage for the DTC angle. Funding not disclosed in fetched material. - **Evidence quality:** Strong self-reported usage metrics, marketplace presence, ISO 27001 certification, external press. No customer logos disclosed on the main page. - **Weaknesses:** Dual positioning (DTC vs. procurement) can blur the pitch; "academic research-backed" claim not linked to a specific publication; no SOC 2 or EU AI Act posture. - **Last-updated signal:** © 2025 Nibble Technology. - **Pull quotes:** - "Let your procurement team do more with less." — https://nibbletechnology.com/ - "Negotiation is more than numbers — it's what, when and how you say it." — https://nibbletechnology.com/ - "Your data is kept secure, fully compliant with ISO 27001." — https://nibbletechnology.com/ - **URLs accessed 2026-04-23:** https://www.nibble.website/ (timeout) ; https://nibbletechnology.com/ ; Shopify App Store and GetApp listings via search. --- ## Decentralized dispute resolution ### 10. Kleros - **Category:** Blockchain-native decentralized dispute-resolution protocol. - **Status:** Active as of 2026-04-23; still small-scale relative to mainstream ADR. - **Core value prop (verbatim):** "A decentralized arbitration service for the disputes of the new economy." (https://kleros.io/) - **Protocol / methodology:** Built on Ethereum. Disputes decided by randomly drawn jurors who stake the native **PNK** token; stake weight governs draw probability (Sybil resistance comes from capital at risk, not identity). Jurors vote privately; coherent voters — those siding with the ultimate majority, a Schelling point — are rewarded in PNK plus arbitration fees (paid in ETH / stablecoins); incoherent voters lose stake. Decisions are appealable into higher-tier **courts** in a hierarchical tree (General Court → subcourts by subject matter: Blockchain, Marketing Services, Curation, Oracle, Token Listing). Each appeal doubles the juror panel and fees. Final courts escalate to the General Court and, in extremis, to a governance fork. - **Product surface:** Kleros Court (core arbitration UI), Kleros Escrow (arbitrated two-party escrow), Curate (decentralized curated lists, used for token registries), Tokens / T2CR (token listing flows), Linguo (translation with dispute resolution), Proof of Humanity (Sybil-resistant identity registry, spun out but still entangled), and Dispute Resolver (generic arbitration frontend). Ecosystem integrations listed include Gnosis Safe, 1inch, Aragon, Ledger, Etherscan, and Polygon ID. - **Evidence of adoption (CLAIMED on kleros.io home):** ~800+ active jurors, ~150M PNK staked, 350+ ETH paid to jurors, 900+ disputes processed, 2M PNK redistributed. Protocol-lifetime figures; modest by ADR industry standards. - **Criticisms (academic and practitioner):** - Regulatory posture: Kleros is not recognized as arbitration under international agreements. Awards are not readily enforceable under the **New York Convention**; most legal scholars treat Kleros outputs as contractual private ordering, not arbitral awards. - Participant selection bias: Jurors self-select by capital; the "financial interest of jurors … may diminish the role of the rule of law." — Springer *International Cybersecurity Law Review*, 2023. - Finality and hung panels: Redistribution rules when no majority is reached are described as "unclear." - Scaling: Kleros's own "Kleros 2.0 / v2" blog (2024–2025) explicitly frames the open problem as going "from 1,000 to 1 billion cases," acknowledging current throughput is an early-adopter scale. - **Pull quotes:** - "A decentralized arbitration service for the disputes of the new economy." — https://kleros.io/ - "Kleros 2.0: Scaling from 1,000 to 1 Billion Cases." — https://blog.kleros.io/towards-kleros-v2/ - **URLs accessed 2026-04-23:** https://kleros.io/ ; https://blog.kleros.io/towards-kleros-v2/ (via search) ; https://docs.kleros.io/kleros-faq (via search). The yellow paper PDF at https://kleros.io/yellowpaper.pdf could not be parsed by the web fetcher (binary); protocol details above come from the homepage, docs, and peer-reviewed summaries. --- ## ODR platform and history reference set ### 11. ODR.com - **Category:** Commercial ODR software vendor. **Status:** Active. Operated by Resourceful Internet Solutions, Inc. (RIS), the same corporate home as Mediate.com, Arbitrate.com, Caseload Manager, and Mediate University. CEO **Colin Rule** (since ~2020). - **What the site is today:** ODR.com positions itself as "the resolutions company" selling configurable ODR software for courts, agencies, universities, ombuds offices, and businesses. Not a directory, not a marketplace. The homepage advertises case-type breadth: traffic / parking, family law (divorce, separation, custody), mediation and arbitration, ombuds / confidential intake, commercial and employment disputes. - **Claims (homepage, fetched 2026-04-23):** - "1.1B Disputes" resolved (lifetime, aggregated across platforms Rule and team have built) - "200+ Platforms" built - "25 Years" of operation - "60M+ disputes per year" at partner organizations (inherits the eBay / PayPal figure) - "92% Client Satisfaction" - Named clients: 25th District Court (Michigan), Johns Hopkins University, NY Courts, OHSU - Provenance claim: "Founded by architects who built eBay and PayPal's resolution systems" — credible given Colin Rule's documented role. - **Evidence vs. claim:** The "1.1B" and "60M+" figures reflect Rule's biographical legacy (eBay's ~60M/year figure, cumulative over decades) rather than a single ODR.com installed base. Treat the aggregate as narrative provenance, not a verifiable deployed-platform metric. - **Pull quote:** - "The resolutions company. Built by the architects of eBay and PayPal's resolution systems." — https://odr.com/ - **URLs accessed 2026-04-23:** https://odr.com/ ; https://colinrule.com/ . ### 12. Modria / Tyler ODR lineage - **Category:** Historically significant ODR platform, now embedded in a government-tech suite. **Status:** Modria as a standalone company no longer exists; the product lineage continues inside Tyler Technologies' courts and justice division, integrated with **Odyssey File & Serve**. - **Origin:** Modria was founded in 2011 by Colin Rule (former eBay / PayPal Director of ODR, 2003–2011) and Chittu Nagarajan (operator of India's largest ODR system). Modria productized the four-stage funnel Rule had built at eBay / PayPal — diagnosis → automated negotiation → mediation → arbitration — and sold it first to e-commerce firms and later to courts and government agencies. - **Acquisition:** Tyler Technologies acquired Modria in May 2017 to complement Odyssey, its dominant court case-management system. Rule joined as VP of ODR at Tyler (2017–2020). Post-acquisition, Tyler wound down Modria's e-commerce line and focused on courts and ADR organizations. - **Influence on court-based ODR:** - Utah: small-claims ODR pilot using Modria-lineage tooling. - Ohio: statewide small-claims ODR. - Michigan: primarily via Matterhorn (a separate Ann Arbor vendor), with Modria also seeding early court ODR thinking. - British Columbia Civil Resolution Tribunal (CRT): Canada's first online tribunal; both precursor BC pilots (Consumer Protection BC, Property Assessment Appeal Board) used Modria before the CRT was built. The CRT now runs custom software, but the architectural DNA persists. - **The "60M disputes / year" figure:** Canonical and repeatedly verified in academic writing: eBay / PayPal resolved ~60M consumer disputes annually by the time Rule left in 2011, more than the entire US civil court system. Attributed to Rule and Del Duca in the *Penn State Arbitration Law Review* ("eBay's De Facto Low Value High Volume Resolution Process"). Modria itself never published a comparable independent volume number — the figure is eBay's, not Modria's. - **Pull quote:** - "Tyler Technologies, a leading provider of integrated software and technology services to the public sector, has acquired Modria." — Tyler / BusinessWire, 2017-05-30 - **URLs accessed 2026-04-23:** https://www.businesswire.com/news/home/20170530005673/en/Tyler-Technologies-Acquires-Modria ; https://www.lawnext.com/2017/06/modria-innovator-online-dispute-resolution-acquired-tyler-technologies.html ; https://colinrule.com/ . ### 13. eBay / PayPal ODR precedent - **Category:** Historical reference and architectural archetype. - **Status:** Systems still operate at scale inside eBay / PayPal; the broader industry influence is the main artifact. - **Canonical history:** eBay launched dispute-resolution features in the late 1990s and partnered with **SquareTrade** (founded 1999 by Steve Abernethy and Ahmed Khaishgi) to handle buyer / seller complaints. SquareTrade handled millions of eBay cases before eBay internalized the function and acquired PayPal in 2002, building a consolidated Resolution Center across both. Colin Rule served as eBay / PayPal's first Director of ODR (2003–2011), architecting the system that, by his own published figures, resolved ~60 million disputes per year — a volume greater than the entire US civil court system. - **Architecture — the four-stage funnel** (the design pattern every subsequent ODR platform inherits): 1. **Diagnosis:** structured forms classify the dispute (item not received, item not as described, unauthorized charge, etc.) 2. **Automated negotiation:** system-mediated offer / counter-offer (full refund, partial refund, return-for-refund, replacement), often without any human intervention. 3. **Mediation:** human neutral intervenes if forms cannot close the gap. 4. **Arbitration / adjudication:** binding outcome from eBay / PayPal or a third-party arbiter. - **Key architectural lessons** (per Rule, Del Duca, UNCITRAL WG III): - Heavy reliance on structured forms, not free-text, is required for ML-free automation at internet scale. - Asynchronous, not synchronous: buyers and sellers are in different time zones. - A monetary "Money-Back Guarantee floor" rather than a numeric ceiling. - Integration with the marketplace's trust / reputation system (feedback, ratings). - Outcomes are cheap and fast, not precedent-setting. - **Regulatory influence:** eBay's experience seeded the UNCITRAL Working Group III Technical Notes on Online Dispute Resolution (2016), a non-binding framework for cross-border low-value e-commerce disputes. It also directly shaped the EU ODR Platform (operated 2016–2025, sunset July 2025) and the US National Center for State Courts' ODR guidance. - **PayPal's current dispute workflow:** buyers open a case in the Resolution Center → 20-day negotiation window → escalation to a PayPal claim → PayPal renders a binding decision applying Buyer / Seller Protection policy. - **Pull quote:** - "Over his eight years at the company, Rule developed systems that resolved approximately 60 million disputes annually, more than the U.S. civil court system." — https://colinrule.com/ - **URLs accessed 2026-04-23:** https://colinrule.com/writing/acr2008.pdf ; https://insight.dickinsonlaw.psu.edu/cgi/viewcontent.cgi?article=1060&context=arbitrationlawreview ; https://uncitral.un.org/en/texts/onlinedispute/explanatorytexts/technical_notes . --- ## Institutional ADR providers ### 14. AAA mediator search and AI-Native Arbitrator - **Category:** Major US institutional arbitration / mediation provider. - **Status:** Active. Largest US ADR institution. Actively rolling out AI. - **Mediator / arbitrator selection workflow:** Parties filing a case with AAA receive a list of proposed neutrals curated from the relevant panel (Commercial, Construction, Consumer, Employment, Labor, Healthcare, Mass Arbitration, ICDR for international). Each neutral has a disclosed CV, hourly rate, location, and conflict disclosures. Parties strike and rank from the list; AAA's case manager resolves the mutual ranking to appoint neutrals. The public-facing https://www.adr.org/Panel page returned a 404 at fetch time (2026-04-23); the party-facing search is primarily party-gated post-filing, not a public Google-style directory. - **Roster size (per AAA statements):** - ICDR (international): ~725 arbitrators / mediators across 100+ countries. - Commercial panel: "1,200+ seasoned arbitrators and mediators." - Technology panel: 223 specialists. - Totals across all panels not aggregated publicly; AAA speaks of a roster in the several thousands. - **Scope:** Commercial, Construction, Consumer, Employment, Labor, Healthcare, Energy, Mass Arbitration, International (ICDR). - **AI features:** In November 2025, AAA-ICDR announced its **AI-Native Arbitrator** in partnership with QuantumBlack (AI by McKinsey), launching documents-only construction arbitrations first. The system was trained on 1,500+ construction awards plus expert labeling. Human-in-the-loop: AI drafts / analyzes; a human arbitrator reviews, can revise, and issues the final binding award. Expansion to additional industries and higher-value claims slated for 2026. - **Rules posture:** AAA's 2024–2025 Commercial Arbitration Rules refresh included new provisions on virtual hearings, cybersecurity, and guidance on AI use by parties and counsel. Consumer Arbitration Rules remain a flashpoint in US mass-arbitration policy debates. - **Pull quotes:** - "Now is the time to embrace AI to drive positive change through speed, efficiency, and accuracy." — Bridget Mary McCormack, AAA-ICDR CEO - "By drawing on nearly a century of ADR expertise … we built a platform that delivers consistent, transparent results." — Diana Didia, AAA-ICDR CTO - **URLs accessed 2026-04-23:** https://www.adr.org/Panel (404) ; https://www.adr.org/press-releases/aaa-icdr-to-launch-ai-native-arbitrator-transforming-dispute-resolution/ ; https://www.adr.org/panel/about-our-panels/ (via search) ; https://adr.org/news-and-insights/the-aaa-s-2024-2025-arbitration-rule-changes-a-breakdown/ (via search). ### 15. JAMS AI Rules - **Category:** First dedicated ADR ruleset for AI-related disputes. - **Status:** In force. Effective **2024-06-14**. Publicly positioned as JAMS's signature differentiator vs. AAA, CPR, and ICC. - **Publication:** *JAMS Artificial Intelligence Disputes Clause and Rules*, effective 2024-06-14. JAMS is the first ADR provider to publish specialized AI arbitration rules. Principal drafters: Ryan Abbott, M.D., Esq., FCIArb (AI-law scholar at University of Surrey and UCLA) and Daniel B. Garrie, Esq. (cybersecurity / AI arbitrator). - **Scope:** Rule 1(e) defines AI broadly as "a machine-based system capable of completing tasks that would otherwise require cognition." The rules apply whenever parties opt in — either via the JAMS model AI clause in a contract or by agreement after a dispute arises. They are not self-executing; adoption is contract-driven. - **Procedural innovations:** 1. **Specialized arbitrator roster.** JAMS proposes neutrals with proven technical + legal AI backgrounds. JAMS maintains an AI neutrals list marketed separately from the general roster. 2. **Expert appointment (Rule 16.1(b)).** On joint request, the arbitrator appoints third-party AI experts from a pre-qualified JAMS list. Expert inspection occurs in "a secured environment established by the Disclosing Party," and experts "shall not transmit or remove any produced materials or information from such environment." This is a trade-secret-aware protocol for inspecting weights, training data, and source code. 3. **Expedited procedures by default.** Standard JAMS Comprehensive Rules make expedited handling opt-in; the AI Rules make it the default — 75 calendar days for percipient discovery, 105 days for expert discovery, hearing within 60 days thereafter. 4. **Confidentiality (Rule 26 + Appendix A AI Disputes Protective Order).** Automatic protective order covering confidential AI materials. - **Gaps** (per Kluwer Arbitration Blog critique, June 2024): No explicit allocation rules for training-data IP disputes, no developer-vs.-user categorization, no provisions on model-hallucination evidence or reproducibility obligations — left to arbitrator discretion. Critics argue the rules are scaffolding, not doctrine. - **Competitive positioning:** AAA's 2025 AI-Native Arbitrator is a *tool* used within existing rules. JAMS AI Rules are a *rule set* — a procedural innovation, not a technology product. CPR and ICC as of April 2026 have issued AI-use guidance but no AI-specific rule sets. - **Pull quote:** - "If jointly requested by the Parties, the Arbitrator shall designate expert(s) to inspect AI systems or related materials." — JAMS AI Rules, Rule 16.1(b) - **URLs accessed 2026-04-23:** https://www.jamsadr.com/artificial-intelligence-disputes-clause-and-rules ; https://www.jamsadr.com/news/2024/jams-announces-new-artificial-intelligence-disputes-clause-and-rules (via search) ; https://arbitrationblog.kluwerarbitration.com/2024/06/27/jams-publishes-artificial-intelligence-arbitration-rules-but-are-they-fit-for-purpose/ (via search). --- ## 16. Human mediator-assist tooling (survey) Tooling used by human mediators today, beyond the chatbot-mediator category. Each entry is short; deep dives belong in the dedicated sub-docs under `179.1.1.5`. **Caseload Manager** — Resourceful Internet Solutions / Mediate.com family. Category: cloud case management. Status: active. Claims 5,000+ professionals, 140+ mediation / human-service programs, endorsed by the National Association for Community Mediation. Used by Illinois Foreclosure Mediation statewide. Five modules: Cases, Activities, Calendar, Mail, Reports. Workflow tool, not AI. URL: https://www.caseloadmanager.com/ . **Smartsettle (iCan Systems, Canada)** — expert-system negotiation assistant. Two products: *Smartsettle ONE* (single-issue Visual Blind Bidding with nudge algorithms that reward collaborative bidding) and *Smartsettle Infinity* (multi-issue, multi-party with preference elicitation and package optimization). An "Expert Neutral Deal-closer" resolves residual gaps. Self-describes as "AI-driven" but is more accurately optimization + mechanism design, not LLM-based. Pioneered the first algorithm-resolved online court case (per ODR Africa Network). **CyberSettle** — blind bidding for insurance claims. Patented double-blind bidding: three rounds; bids within tolerance auto-settle at midpoint. Lifetime claim: ~200,000 claims, ~$1.4B total settled. Historically deployed by the NYC Comptroller's Office for claims against the City. URL: https://www.cybersettle.com/ . **Fair Outcomes, Inc.** — game-theoretic mechanism provider. Products include *Fair Buy-Sell*, *Fair Division*, *Fair Proposals* — formal mechanisms descended from Barry Nalebuff (Yale SOM) and collaborators' fair-division research (*Split the Pie*, 2022). Not AI; rigorous game theory. URL: https://www.fairoutcomes.com/ . **Picture It Settled** (Don Philbin) — litigation negotiation analytics. Uses neural networks on a corpus of 10,000+ settled cases to project opposing-side moves and settlement zones; flagship module *Settlement Prophet*. Published accuracy: within 6.6% after two rounds, 3% after three. URL: https://www.pictureitsettled.com/ . **Matterhorn by Court Innovations (Catalis)** — court ODR. Deployed in 70+ courts across 12 states; Michigan is the heaviest user with 29 District Courts. Case types: traffic, parking, civil infractions, license suspensions, small claims, family-court compliance, lesser misdemeanors, warrants, amnesty. URL: https://getmatterhorn.com/ . **Tyler ODR (ex-Modria)** — court ODR embedded in Odyssey. Tyler dominates US court case management. Post-Modria rebranding has subsumed the ODR product into Tyler's Enterprise Justice and Odyssey File & Serve suite. **Immediation (Australia)** — video-first ODR / mediation / arbitration platform. Operationally active but financially distressed: filed for bankruptcy October 2023, continues to trade as ADR Technology Pty Ltd. Used by Australian courts and major law firms. URL: https://immediation.com/ . **MODRON** — collaborative "Spaces" platform for ADR practitioners, marketed to mediators and arbitrators for case, client, document, scheduling, and settlement management. Note: the name *MODRON* is also used by an unrelated Australian AI-hardware company; the ODR MODRON is at modron.law/modron-spaces. **BC Civil Resolution Tribunal (CRT)** — public online tribunal, not a vendor, but pivotal. Canada's first online tribunal. Jurisdiction: strata property, small claims ≤ CAD $5,000, motor-vehicle injury, societies and co-ops. Precursor pilots used Modria; the production CRT runs custom software. 2024–2025 annual report is published; specific volume figures were not retrievable from the blog landing page at fetch time. **OurFamilyWizard (OFW)** — co-parenting coordination, adjacent to ODR. Launched ToneMeter AI in May 2025 — an LLM-based message rewriter that flags hostile language and proposes neutral alternatives. Beta rewrote 10,000+ messages for 2,500 co-parents; 90% satisfaction reported. Self-hosted models for privacy. URL: https://www.ourfamilywizard.com/product-features/tonemeter . **Adjacent analytics (cited by mediators, not mediator tools):** Lex Machina (LexisNexis) and Fastcase / vLex power case analytics used in pre-mediation prep but are not mediation platforms. --- ## Cross-product observations - **Autonomy spectrum:** Disputell (none) → TheMediator.AI / ZODR / Mediator.ai / Bot Mediation (advisory) → Dyspute.ai (AI-first with optional human) → MediationAI / Pactum / Nibble (high autonomy with human-set guardrails). - **Multi-party handling is a gap.** Every reviewed AI product is effectively two-party. Pactum and Nibble achieve "scale" by running many parallel bilaterals. No reviewed product claims true N-party coalition or multi-issue multi-party bargaining. Concordia's N-party coalition search and Shapley attribution in `179.4.2.7` is a clear differentiation target. - **Bargaining-theory claims are rare.** Only Mediator.ai openly cites Nash bargaining and describes a GA + LLM-ranker architecture. Everyone else markets in generic "AI mediates" terms. Concordia's explicit bargaining kernels (Nash, Kalai-Smorodinsky, NSGA-II, MCTS, CP-SAT, Bayesian, PSRO) in §179.4.2 are a durable differentiation. - **Compliance tier.** Only Pactum (SOC 2 Type II) and Nibble (ISO 27001) disclose certifications. ZODR claims "SOC 2-aligned" (weaker). Consumer ODR players disclose little. Concordia's standards baseline in §179.1.3 (UNCITRAL ODR, EU AI Act, NIST AI RMF) targets a tier not currently occupied. - **Privacy posture.** Disputell has the strongest stated private-caucus discipline among AI products. TheMediator.AI has a 30-day deletion rule. Nobody discloses physically isolated per-party prompt contexts — this is what §179.5.1.1 requires of Concordia. - **Evidence gap.** Most consumer AI products show no outcome data, no benchmark scores, no peer-reviewed backing. Pactum's customer logos and Bot Mediation's named firms are the strongest evidence in the AI cohort. Kleros, eBay / PayPal, Modria-lineage court deployments, AAA / JAMS, and the Smartsettle / CyberSettle lineage carry genuinely documented track records. - **Human-in-the-loop is often theatre.** Several AI-first products name human review but do not disclose reviewer queue SLAs, calibration, or escalation triggers. §179.5.5.1 demands these be concrete in Concordia. - **Fabrication risk on claimed metrics.** "60M disputes / year" is an eBay / PayPal figure; ODR.com inherits it as corporate provenance. Kleros lifetime stats are modest. Mediator.ai, MediationAI, and TheMediator.AI have no independent outcome numbers. Concordia's benchmark harness (§179.8.1) should not rely on any of these self-reported metrics as baselines. ## Concordia posture implications This audit reinforces several design decisions already written into Phase 179: - Keep Nash bargaining as the baseline scoring rule (`179.4.2.1`), but extend beyond Mediator.ai with multi-objective evolutionary search (`179.4.2.2`), MCTS / LATS (`179.4.2.3`), CP-SAT / MILP (`179.4.2.4`), Bayesian optimization (`179.4.2.5`), PSRO (`179.4.2.6`), and coalition stability (`179.4.2.7`). No reviewed product combines more than one of these kernels. - Treat preference inference as fundamentally uncertain. Mediator.ai's public post and LLM-bargaining critiques converge on instability, paraphrase sensitivity, order effects, and adversarial framing. §179.3.2.4 and §179.3.2.5 must be first-class, not afterthoughts. - Party isolation must be architectural, not promised. No reviewed product disclosed physically separated prompt contexts or per-party data keys. `179.5.1.1` and `179.5.1.2` are where Concordia meaningfully raises the bar. - Procurement and court-ODR pathways already have documented value (Pactum, Nibble, Modria / Tyler, Matterhorn, CRT). Concordia's §179.7.1 Maat path and §179.7.2 Themis / Aje path are well-precedented routes to measurable impact; consumer AI mediation is a narrower opportunity. - Institutional AI-rule frameworks (JAMS AI Rules, AAA AI-Native Arbitrator) are product primitives to *integrate with*, not to compete with, and `179.1.3.2` / `179.5.3` should cross-reference them. - Smart-contract settlement (§179.7.2.4) should not depend on Kleros for authoritative enforcement; treat Kleros-style adapters as optional arbitration backstops per `179.7.2.5`, with the default decision-maker remaining a qualified human reviewer or consensus mechanism. ## Source inventory URLs accessed, searched, or referenced on 2026-04-23. Categories are mapped in `docs/research/source-matrix.md` (to be produced for `179.1.1.2`). - https://mediator.ai/ — commercial - https://mediator.ai/blog/ai-negotiation-nash-bargaining/ — commercial - https://mediator.ai/examples/bakery-partnership-agreement/ — commercial (404) - https://www.mediationai.app/ — commercial - https://botmediation.com/ — commercial - https://www.americanbar.org/groups/law_practice/resources/law-technology-today/2025/ai-powered-mediation-for-efficient-legal-dispute-resolution/ — press - https://www.zodr.ai/ — commercial (403) - https://zodr.ai/ — commercial (403) - https://lminetwork.com/zodr/ — commercial - https://www.lmipodcast.com/ep347-giving-zoom-mediations-a-body-introducing-zodr-ai/ — press - https://www.disputell.com/ — commercial - https://dyspute.ai/ — commercial (403) - https://www.dyspute.ai/ — commercial (403) - https://www.lawnext.com/2026/01/dyspute-ai-launches-adri-v2-a-24-7-asynchronous-ai-mediation-platform.html — press - https://themediator.ai/ — commercial - https://play.google.com/store/apps/details?id=ca.underlabs.mediator — commercial - https://pactum.com/ — commercial - https://pactum.com/procurement-agents — commercial - https://procurementmag.com/news/pactum-secures-series-c-funding-to-drive-agentic-ai-adoption — press - https://sourcingjournal.com/topics/technology/pactum-walmart-ai-supplier-negotiation-chatbot-vendor-276400/ — press - https://nibbletechnology.com/ — commercial - https://www.nibble.website/ — commercial (timeout) - https://apps.shopify.com/nibble-chat-bot — commercial - https://hashtagpaid.com/banknotes/nibble-brings-an-ai-powered-negotiator-bot-to-ecommerce — press - https://kleros.io/ — commercial - https://blog.kleros.io/towards-kleros-v2/ — commercial - https://docs.kleros.io/kleros-faq — commercial - https://odr.com/ — commercial - https://colinrule.com/ — case-study - https://colinrule.com/writing/acr2008.pdf — academic - https://insight.dickinsonlaw.psu.edu/cgi/viewcontent.cgi?article=1060&context=arbitrationlawreview — academic - https://uncitral.un.org/en/texts/onlinedispute/explanatorytexts/technical_notes — standards - https://www.businesswire.com/news/home/20170530005673/en/Tyler-Technologies-Acquires-Modria — press - https://www.lawnext.com/2017/06/modria-innovator-online-dispute-resolution-acquired-tyler-technologies.html — press - https://www.adr.org/Panel — commercial (404) - https://www.adr.org/press-releases/aaa-icdr-to-launch-ai-native-arbitrator-transforming-dispute-resolution/ — commercial - https://www.adr.org/panel/about-our-panels/ — commercial - https://adr.org/news-and-insights/the-aaa-s-2024-2025-arbitration-rule-changes-a-breakdown/ — commercial - https://www.jamsadr.com/artificial-intelligence-disputes-clause-and-rules — commercial - https://www.jamsadr.com/news/2024/jams-announces-new-artificial-intelligence-disputes-clause-and-rules — commercial - https://arbitrationblog.kluwerarbitration.com/2024/06/27/jams-publishes-artificial-intelligence-arbitration-rules-but-are-they-fit-for-purpose/ — press - https://www.caseloadmanager.com/ — commercial - https://info.smartsettle.com/ — commercial - https://www.cybersettle.com/ — commercial - https://www.fairoutcomes.com/ — commercial - https://www.pictureitsettled.com/ — commercial - https://getmatterhorn.com/ — commercial - https://immediation.com/ — commercial - https://civilresolutionbc.ca/blog/crt-annual-report-2024-2025/ — case-study - https://www.ourfamilywizard.com/product-features/tonemeter — commercial --- *This audit is a dated snapshot. The research refresh gate in `179.1.1.6` requires this document to be re-run with a new fetch date before Phase B transitions and before every pilot; product pages, pricing, compliance posture, regulatory position, and traction numbers all move. Re-check the `(403)` and `(404)` URLs specifically, because a fresh fetch from a non-blocked origin may yield material changes.*