Skip to content
NexusBook a demo

Customer story · Real estate

An industrial real-estate operator

Contract intelligence for an industrial real-estate operator

One assistant across tenant and supplier contracts. It answers only from the documents, always cites them and surfaces contradictions rather than smoothing them over.

Executive summary

The situation

Tenant obligations and supplier services live in separate libraries, but questions often span both. Useful answers must preserve evidence and reveal contradictions rather than resolve them by guesswork.

What was built

An assistant searches both libraries, correlates obligations with services and returns cited answers. Conflicts are surfaced, and consequential findings are routed for human verification.

Why it matters

Gives people a way to examine promises alongside contracted services. It supports contract review without replacing legal or commercial judgement with an assistant’s interpretation.

Primary users
People reviewing contractual obligations
Core systems/data
Tenant and supplier contracts
Human control point
Human verification of consequential findings
Nexus role
Retrieval, correlation and verification routing

The challenge

Two contract worlds that must agree.

Tenant agreements describe obligations on one side; supplier agreements describe services on the other. Questions about whether they align can require reading across both libraries.

The risk is not only failing to find a clause. A smooth answer can also conceal disagreement between documents. The assistant needed to keep sources and contradictions visible while leaving consequential interpretation to people.

What we built

An assistant across both libraries.

Nexus provides an assistant that searches tenant and supplier contracts together, connecting obligations with the services supporting them. Answers come only from the document base and cite their source contracts. The evidence stays available for inspection, rather than disappearing into a generated summary.

Where the documents disagree, the assistant surfaces the contradiction instead of constructing a convenient compromise. Consequential findings go to human verification before action. It is a review aid that locates and connects evidence, not an authority that settles contractual conflicts on its own.

How the workflow runs

Step by step.

  1. 01

    Ask

    A user asks a question about obligations or services. Tenant and supplier contracts are searched together.

  2. 02

    Correlate

    Tenant obligations are matched to supplier services. The answer cites the contracts it relies on.

  3. 03

    Flag and verify

    Contradictions are surfaced and consequential findings go to human verification.

What the business team sees

Cited answers and flagged conflicts.

Users see cited answers across both libraries, including the passages supporting a correlation. Contradictions become findings to examine, not details quietly absorbed into a smooth answer. The interface is a starting point for review with the actual contracts in view.

Illustrative business interface
Contract assistant
  • Questions across both libraries
  • Cited contract passages
  • Tenant / supplier correlation
  • Contradictions surfaced
  • Verification queue

How Nexus works underneath

Correlate the evidence. Keep the judgement with people.

Grounded retrieval ties answers to the corpus, while cross-library correlation links obligations to services. Citations and contradiction handling preserve differences in the source material. Nexus routes consequential findings to a person so a generated answer does not become an unreviewed decision.

Architecture

How the layers fit together.

Business team

Contract assistant

Questions · citations · contradictions · verification

Nexus orchestration

  1. 01Grounded retrieval
  2. 02Cross-library correlation
  3. 03Contradiction detection
  4. 04Citation
  5. 05Verification routing

Connected systems and data

  • Tenant contracts
  • Supplier contracts
Answers only from the contracts. Conflicts are shown. People verify what matters.

Human control and audit

Findings are verified by people.

The assistant does not resolve contradictions on its own. It shows them, with citations, and sends consequential findings for human verification.

What this changes

  • Questions can span both libraries. One query can examine tenant obligations alongside supplier services.

  • Disagreement remains visible. Conflicting provisions are surfaced with citations.

  • Findings have a review boundary. Consequential conclusions reach human verification rather than autonomous action.

Your process

Discuss a similar workflow

Tell us about the documents, systems and decisions involved. We will discuss where Nexus orchestration fits and where your team stays in control.