Spellbook AI for Lawyers: What to Check in a Review

Spellbook is an AI drafting assistant designed for legal agreements, built to work within document workflows attorneys use. Before a law firm relies on it for client work, a structured review tells you how it performs on the agreements your practice handles, where it needs guidance to match the firm's standards, and which review steps are still required. Tools like fenbs keep the approved drafting instructions and evaluation decisions outside any single tool, so the standards from the evaluation carry into every session that follows.

This page is about how to evaluate Spellbook, not a definitive review of its features. For current information on what Spellbook does and how it is licensed, visit Spellbook's own website directly. For broader AI drafting tools, see AI agents for lawyers and AI for lawyers drafting.

What Spellbook does and where it fits

Spellbook is designed to assist lawyers with contract drafting and review. Its integration into document workflows is its main selling point: the AI assistance is available where the drafting happens. Check Spellbook's website for current details on which platforms it supports and which practice areas its suggestions cover.

The relevant question for a firm considering Spellbook is not what it claims to do, but how it performs on the specific agreements your practice handles most frequently. A tool that performs well on technology licensing agreements may perform differently on construction contracts or employment agreements.

Spellbook is not a legal research tool and does not verify citations. It is a drafting assistant. The attorney's review of suggested language against the applicable law, the client's requirements, and the firm's standards remains necessary after every session.

How to run a proper evaluation

A useful evaluation uses real documents from your practice, not example contracts from a tutorial. Choose three to five agreements that represent the firm's typical work: different agreement types, different complexity levels, and different governing law states.

For each test document, use Spellbook to review or generate language as you normally would. Note what it suggests, compare it to your firm's preferred language, and record where the suggestions needed revision and why. Do the same for a document where you draft from scratch against a client brief.

Run the evaluation before making a purchasing decision. Free trial periods are designed for this purpose. A structured test on real work is more informative than a demo on example documents.

Review criteria for the evaluation

Apply consistent criteria across each test document so the evaluation is comparable. The following criteria cover the areas where AI drafting tools most commonly require revision in legal practice.

• Governing law accuracy: does the suggested language reflect the applicable jurisdiction's requirements?
• Scope and coverage: does the suggested language address the specific scenario, or is it generic?
• Firm style: does the tone, structure, and level of formality match what the firm uses?
• Non-standard terms: does the tool handle instructions about unusual or client-specific provisions?
• Consistency: does it apply the same approach across similar clause types in the same document?

Record the result for each criterion on each test document. The pattern across multiple documents tells you more than any single result.

Data handling and confidentiality

Legal AI tools that integrate into document workflows have access to the content of those documents. Before using Spellbook with client agreements, check Spellbook's current data handling terms and privacy policy. Understand what data is processed, where it is stored, and whether it is used for model training.

Your state bar may have issued guidance on using AI tools with client documents. Check the current guidance from your bar rather than relying on general summaries. Confidentiality obligations under Model Rule 1.6 apply to the choice of tools as well as to the content of communications.

Keeping evaluation decisions and approved instructions

The value of a structured evaluation comes from applying what you learned. If the evaluation showed that Spellbook needs specific instructions to produce jurisdiction-specific language, those instructions should load at the start of every subsequent session. If certain clause types need additional attorney scrutiny, that should be a standing review rule.

These evaluation outcomes are practice decisions, not personal notes. The junior associate using Spellbook six months after the evaluation should operate under the same standards the evaluation established. That requires recording the decisions where they are accessible, not leaving them in the evaluating attorney's memory.

How to do this with fenbs

1. Create a free account and set up a project for the evaluation or for the practice area.
2. Add the evaluation results as tasks: the document types tested, the criteria applied, and the outcomes.
3. Add the drafting instructions that improve Spellbook's output as Skills and prompts. These are the instructions to load at the start of each drafting session.
4. Add the review standards as Decisions and rules: which clause types need additional scrutiny, which document types require partner review, what the data handling decision was.
5. Connect a different AI assistant from Integrations to read the evaluation decisions and approved instructions.

Fenbs does not connect to Spellbook directly. It holds the instructions and decisions that guide how any AI drafting tool is used. For the broader context of keeping AI work consistent across tools, see what an AI brain is.