BlackBoiler’s Veris Merges Deterministic and Generative AI for Legal Contract Review

The new Microsoft Word add-in could redefine how lawyers and AI agents collaborate on redlining—without leaving their workflow.

If you’ve ever reviewed a contract, you know the drill: endless back-and-forth, tracking changes, and wrestling with conflicting edits. BlackBoiler’s new Veris tool, launched today, aims to streamline this by combining two AI approaches—deterministic redlining and generative suggestions—directly inside Microsoft Word.

Here’s why it matters for AI agents: Veris doesn’t just automate edits; it explains them. That transparency could make it a bridge between human legal teams and AI-assisted workflows, especially for enterprises where traceability is non-negotiable.

How Veris Works—and Why It’s Different

BlackBoiler’s legacy tool used deterministic AI—rule-based systems that follow predefined logic—to redline contracts.

Veris keeps that precision but adds a generative AI layer that suggests alternative language, explains edits, and even flags potential negotiation points according to LawSites.

Key features:

• Deterministic redlining: Edits are made based on clear rules (e.g., striking unfavorable clauses).
• Generative suggestions: AI proposes alternative wording, with explanations in comments.
• Negotiation flags: Highlights terms likely to trigger pushback from the other party.

This hybrid approach could reduce the “black box” problem in pure generative AI, where edits arrive without rationale. For AI agents handling contract workflows, that interpretability is critical.

The Bigger Picture for AI-Assisted Legal Work

Veris isn’t just a productivity tool—it’s a test case for how deterministic and generative AI can coexist in high-stakes environments. Legal teams (and the AI agents they deploy) need both speed and accountability. By embedding this in Word, BlackBoiler sidesteps the friction of switching platforms, making adoption smoother.

One open question: How will firms balance AI suggestions with human judgment? Veris doesn’t auto-apply generative edits; it leaves the final call to the user. That may reassure cautious adopters but could limit time savings.

For developers building agent-infrastructure for legal tech, Veris’ API (if offered) could be a template for blending structured and unstructured AI outputs. Watch this space.

Need to compare tools? Explore other AI agents tackling document-heavy workflows.