Founded by a former Xometry VP of AI and a former Bristol Myers Squibb Lead AI/Tech Counsel, the U.S.-based company replaces black-box legal AI with an end-to-end negotiation engine whose every risk decision can be traced, audited and run inside a company’s own cloud.

CHICAGO, IL, October 27th, 2026, ZEX PR WIRE — RiskVantage AI, a U.S.-based legal AI company building an end-to-end transaction engine for enterprise contracts and M&A, today introduced RiskVantage Negotiate, its explainable AI platform for enterprise contract negotiation. The company will showcase the platform internationally at the Dubai AI Festival, October 26 and 27.

Enterprises spend an estimated $75 billion a year on contract review, negotiation and M&A legal work, and most of it is still done by hand. A contract arrives on the other party’s paper, waits in a legal queue, and is read clause by clause, with results that depend on who reviews it. World Commerce & Contracting research puts the average value lost to poor contract management at about 9% of revenue. Applied to the Fortune 500, that is an estimated $1.7 trillion a year.

RiskVantage built its platform to change that, with an architecture designed for the legal teams and regulated industries that cannot accept unexplained or unverifiable AI output.

Beyond the wrapper

Most legal AI tools for contract review wrap a general-purpose large language model. RiskVantage takes a different approach in three ways.

Transparent, deterministic risk decisions. RiskVantage uses a neuro-symbolic architecture: risk decisions are made in a knowledge graph, and language models only read contracts and draft language. No risk decision is made by a language model. Teams state their preferred position, acceptable fallbacks and walk-away point once, and every incoming clause is graded against that ladder. Each finding names the rule that was fired, so legal teams can review and audit every decision.

Sovereign, cost-efficient deployment. The platform runs on small language models adapted for legal tasks, on commodity hardware. In Khazraee’s words, a domain-adapted 9-billion-parameter model is “not smart enough” to wander outside its sandbox. In an internal benchmark on legal concept detection and extraction, RiskVantage’s domain-adapted small language model achieved about 97% accuracy, performing on par with frontier models. It is also inexpensive to run. Most legal AI tools rely on frontier models whose inference costs grow with every page reviewed, an economic model that is hard to sustain at enterprise scale. Using small language models lets organizations deploy them inside their own private cloud or on premises, which matters for pharma, healthcare, financial services, defense and other regulated sectors with strict data requirements.

End-to-end negotiation, not a standalone assistant. RiskVantage covers the full negotiation, from intake and review to redlines, counterparty exchange and signature, and connects to the systems enterprises already use. RiskVantage has filed a provisional patent (April 2025) covering its neuro-symbolic risk classification, ontology-driven reasoning and automated redlining.

Platform features

RiskVantage Negotiate, the company’s core product, is live. In internal testing, it reduced first-draft issues by 60% compared with manual review. It includes:

• Intake builder: captures deal context, such as deal value or whether sensitive data is involved, which sets how strictly each clause is reviewed.
• Risk configurator: uses AI to build a company’s risk profile automatically from its existing contracts, policies and SOPs, and keeps each position marked unreviewed until a human confirms it.
• Deviation reports and redlines: compares each clause with the playbook, explains every deviation, and produces a tracked-changes redline ready to send.
• Full Microsoft Word compatibility in the browser: legal teams and counterparties work in the format they already use.
• Counterparty portal: a shared space to exchange drafts and negotiate with the other side.
• Internal sharing with subject-matter experts: routes specific clauses to colleagues in areas such as privacy, security or finance.
• Audit trail: a record of every decision across the negotiation.
• Enterprise integrations: connects to ERP, contract lifecycle management (CLM), e-signature and communications tools such as Slack.

The platform currently covers different contract types including MSAs, SaaS agreements, SOWs, NDAs, IT services, software licenses, proof of concept and IT orders. Two M&A modules are in development: RiskVantage Diligence, which analyzes data rooms against an acquirer’s baseline, and RiskVantage Integrate, which builds a plan to align acquired contracts with the parent company.

Founder quotes

“Every enterprise contract goes through the same broken process: a queue, a clause-by-clause read, and an outcome that depends on who happened to review it,” said Mark Afshar, CEO and co-founder of RiskVantage AI and former Lead AI/Tech Counsel at Bristol Myers Squibb, where he closed more than 2,000 technology transactions. “We built RiskVantage from inside the deal room, so legal and procurement teams can see exactly why a clause was flagged and act on it with confidence.”

“If a rule can govern it, you should never use an LLM,” said Emad Khazraee, CTO and co-founder of RiskVantage AI and former VP of AI at Xometry. “Legal risk is exactly that kind of problem. We put the reasoning in a knowledge graph built from each company’s own playbook and use small language models only to read and write. You don’t use a missile to hit a fly.”

Visit RiskVantage at Dubai AI Festival

Attendees can see live demonstrations of RiskVantage Negotiate at booth S95, Dubai World Trade Center, October 26 and 27. To book a meeting with the founders during the festival, contact contact@riskvantage.ai.

About RiskVantage AI

RiskVantage AI is a Chicago-based company building an enterprise transaction engine that turns legal risk into strategic advantage. Its neuro-symbolic platform combines a knowledge graph of each company’s playbook with domain-adapted small language models to deliver explainable, auditable contract negotiation that can run in private cloud or on-premises environments. RiskVantage was co-founded by Mark Afshar (CEO), former Lead AI/Tech Counsel at Bristol Myers Squibb and a Penn Law alumnus, and Emad Khazraee (CTO), former VP of AI at Xometry and a former data science and AI leader at Turing, a unicorn startup. Khazraee was also an information science professor and a Harvard Law Fellow, and holds a PhD in Information Sciences. Learn more at riskvantage.ai.

TwinMeta has introduced the CPTI (Creative Power Type Indicator), a self-report assessment designed to help creators understand their individual creative working styles in the AI era. Available at genesistwinker.xyz, CPTI explores four dimensions: ideation, expression, creative goals, and work process. Rather than ranking creativity through scores or productivity, the assessment focuses on how individuals generate ideas, express themselves, and approach creative tasks.TwinMeta has also published a related research paper on SSRN, outlining CPTI’s theoretical foundation and development process. The study reviewed 39 academic sources and developed 20 preliminary assessment items. Currently in its early research stage, CPTI will require further empirical studies to evaluate its reliability and validity. The initiative aims to encourage greater self-awareness and a broader understanding of human creativity in an increasingly AI-driven environment.

Lewes, DE, United States, 27th Oct 2026 – TwinMeta has introduced the CPTI (Creative Power Type Indicator), a self-report assessment designed to help creators better understand their individual working styles, and has published a related research paper on the academic research-sharing platform SSRN.

CPTI examines how individuals approach creativity across four dimensions: ideation, expression, creative goals, and work process. The assessment is available at genesistwinker.xyz.

CPTI was developed around a central question: how should human creativity be understood in the age of AI?

As AI increasingly generates text, images, and video, the human role in deciding what to create and what meaning to give it is becoming even more important. However, judging creativity only by the quality of the final output or the speed of production may not fully capture the different processes and choices involved in how individuals create.

Rather than assigning a single score to determine whether someone is more or less creative, CPTI explores the approaches individuals tend to prefer when generating ideas, expressing them, and developing their work.

The assessment is intended to help people understand differences in creative style without ranking one approach above another. For example, someone who explores a wide range of possibilities may work differently from someone who prefers to develop a single idea in greater depth.

CPTI is structured around four areas: Ideation Tendency, which examines how individuals explore ideas; Expression Tendency, which looks at how ideas are translated into concrete forms; Creative Goal Tendency, which considers the purpose behind creative work; and Work Process Tendency, which explores how individuals approach experimentation and completion.

Through these dimensions, CPTI is designed to encourage creators to reflect on their own choices and creative processes.

The research paper published on SSRN outlines the theoretical foundation and development process behind CPTI. The study reviewed 39 academic sources and used them to develop 20 preliminary assessment items.

The project is currently at an early, literature-based research stage. The reliability and validity of the assessment are expected to be evaluated through future empirical studies.

Rather than simply asking creators to produce more, CPTI focuses on giving individuals an opportunity to understand how they create.

It represents an effort to explore creativity in the AI era not only through productivity, but also through the different perspectives, preferences, and working styles that shape each person’s creative process.

More information about the CPTI creative tendency assessment is available at genesistwinker.xyz.

Media Contact

Organization: TWINMETA INC

Contact Person: Yongil Choi

Website: http://twinmeta.xyz

Email: Send Email

Contact Number: +821033588983

Address: Coastal Highway Lewes DE

Address 2: 16192

City: Lewes

State: DE

Country: United States

Release id: 50203

View source version on King Newswire:
TwinMeta Introduces CPTI Creative Tendency Assessment, Publishes Related Research on SSRN