Product/UX Design

From long forms to AI-assisted drafting: redefining legal document creation

Our legal document feature felt just like everyone else’s — long forms, confusing legal language, and not much flexibility.

I told my team that we could do better. I led a brainstorming workshop with key stakeholders, and together we mapped out all the ways we could improve the experience.

That’s how we landed on a new direction: a smarter, AI-assisted document builder. Now, users can edit clauses directly, ask an AI assistant to tweak language for them, or skip the manual customization entirely by just telling the AI what they need.
Person asking an AI chatbot to generate a Force Majeure clause and the AI responds with the clause
My role
Lead UI/UX Designer, Product Designer, UX Researcher
Timeline
3 weeks
Disclaimer
Some project details were omitted to protect client confidentiality.

Why redesign our legal document template generator?

Most business formation portals offer access to legal document templates like NDAs, Operating Agreements, and Employment Contracts. Ours was no different.

I wanted to challenge industry norms and ask if this was really the only way to customize document templates. What if we could do better and find a user-friendly way to do it?

I found that:

  • 28% of customer support tickets were related to confusion or frustration with the document creation flow.
  • 17 min is the average completion time for legal document forms.
  • 41% drop-off rate before document generation was evident.
  • We have a missed opportunity to differentiate ourselves from competitors.

We also wanted to improve these metrics by creating an intuitive flow that ensures users complete the document generation.

Results

  • Positive user feedback on usability tests, validating the design decisions to keep iterating on the feature.
  • The company adopted a new product philosophy focused on discovering effective and user-friendly solutions, moving away from traditional methods that may not serve users' needs.
  • We began asking: Why are we doing it this way? Could we do it better?
  • Our documents experience became a standout differentiator in sales demos.

Hypothesis

What if legal documents could adapt to users — not the other way around?

We believed a smarter, more flexible interface could reduce user drop-off rate, increase trust, and create competitive differentiation.

Rethinking the status quo

I prepared a competitive analysis and demoed four template creation tools: Legal Zoom, Rocket Lawyer, LegalTemplates.net, and LawDepot. I walked stakeholders through common pitfalls and user frustrations, emphasizing that our product faced the same problems.

Then, I pitched a new approach: a smarter document experience powered by guided UX and AI.

My pitch focused on “What if we did it better?”

Competitive analysis

Upon analyzing Legal Zoom, Rocket Lawyer, LegalTemplates.net, and LawDepot's document creation tool, I found these pitfalls:

  • The customization forms vary in length. The longer forms are time-consuming and have excessive steps, which can lead to user frustration.
  • The short forms promote quicker document generation, but it's difficult to customize the document as a result. Less inputs means less opportunities to tailor the document to your needs.
  • There is guidance provided when it comes to legal jargon that the average user may not understand. However, Legal Zoom sometimes fails to explain certain legal terms.
Legal Zoom document template customization page
LegalTemplates.net document template customization page
RocketLawyer document template customization page
LawDepot document template customization page

Brainstorming workshop

To spark innovation, I led a cross-functional brainstorming session followed by an affinity mapping exercise to group together ideas. We then turned these ideas into potential solutions.

I also used How Might We questions to study the problem:

  • How might we ensure generated clauses and documents are compliant with local laws?
  • How might we allow users to customize clauses easier?
  • How might we move beyond the generic form-fill approach?
  • How might we use modern tools like AI to adapt to users better?
  • How might we integrate legal experts into the process so that users can have an attorney review the generated document?

Here are key ideas from this workshop:

  • Create a document notarization feature.
  • Educate users during the entire document generation process for better outcomes.
  • Custom clause creation and modification with AI assistance.
  • Have the tool adapt to state laws and stay up to date with new laws.
  • Clause and legal jargon explanation using simple words.
  • Risk analysis to let users know if the document is missing anything.
  • Compliance review of generated documents using Al and human intervention.
  • Collaboration tools that allow attorneys and document recipients to leave comments on document clauses, streamlining negotiation processes.

Problem statement

Our legal document flow followed the industry standard of long, rigid forms. Users found it confusing, often dropped off, or reached out to support.

Hypothesis

What if legal document generators could adapt to users — not the other way around?

Storyboard

I conducted a storyboard workshop to develop a shared understanding of the experience, business goals, and user pain points.

This was one of the ways I was able to pitch my ideas, gather feedback, and explain concepts to team members with little to no design knowledge.

Outcomes:

  • We re-framed the problem. It was no longer about making forms faster. It was now about making legal documents less intimidating.
  • We identified emotions like “I don’t know what this means”, “I’m scared to get this wrong”, and “Do I need a lawyer for this?”
  • We found out that most user confusion happens before they visit our website, prompting me to design the feature in a way that it looks inviting and easy to digest rather than create an overwhelming UI.

Panel 1: Property manager Savannah tells a tenant their application is approved and she'll send the lease.

Panel 2: Savannah researches lease agreements online but finds it confusing.

Panel 3: Worried about the deadline, she looks for templates.

Panel 4: She finds a lease agreement template website.

Panel 5: She customizes the template, but decides to pay for an attorney review.

Panel 6: An attorney reviews the document and suggests edits via chat.

Panel 7: Savannah downloads and sends the finalized lease agreement.

Panel 8: The tenant signs the lease and moves in.

A comic strip with eight panels shows a property manager, Savannah, creating a lease agreement for a new tenant.

Prototypes

After I performed several iterations, this is the design I created that meets both user and business needs. I used input from my team, subject matter experts, and brainstorming workshop participants.

Template page

Lease agreement template description page

Customization page

The clause customization is located in one tab, whereas the clause description is in another. I included common clauses for each use case. For example, if the user creates a lease agreement, they will see common clauses found in most of these documents.

This is to educate them and guide them.

For the clause explanation tab, users are able to see a simplified version of each clause. They can either click on the clause located on the document preview, or select it from the clause list. Once they perform any of these actions, the accordion expands to reveal the explanation.

It's part of our efforts to educate users on legal terms and clauses, guiding them through the process to make the document generation possible.

Template customization page: clause explanation tab selected, displaying the list of clauses  within accordions. One is expanded revealing the clause explanation
Template customization page: clause customization tab selected, displaying a list of commonly used clauses, an add clause button, and a run risk assessment button

Add and customize clause

The user can add a clause manually, or with AI assistance. I design it this way to account for users who want control, and users who want more assistance.

Manually customize clauses

Some users don't know the exact name of a clause due to little to no legal knowledge. I made it easy for them to search for clauses by just having them type a description of what they're looking for.

For example, a landlord can type "a clause that details the consequences of breaking the lease early", and it will show a list of clauses that match the description.

The customization is a fill in the blanks format, which is easier and faster than a long form. They can also further customize the clause.

Legal terms are highlighted so that the user can view the definition by hovering over it. Or, they can use the toggle to view a simplified version of the clause.

Add Clause dialog displaying a list of commonly used clauses
Add Clause dialog with search results for "lease duration" loading
Add Clause dialog with search results for "lease duration"
Add Clause dialog with clause preview, further customizations, and a preview of the clause
Add Clause dialog with clause preview, further customizations, and a preview of the clause using simplified language. The "show simplified version" toggle is on

Customize clause using AI

The user has the option to skip the manual editing and use AI chat instead. They can enter prompts to generate clauses and customize anything they need by telling the chatbot.

Add clause dialog with Generate Clauses with AI tab selected, showing an AI chatbot generating a Force Majeure clause
Add clause dialog with Generate Clauses with AI tab selected, showing a chatbot that generated a Force Majeure clause with "Add clause to document" and "Restore original version" buttons below it

Testing and validation

To validate my design decisions and ensure we're on the right path, I conducted moderated usability tests with 6 participants that use our product. We used a mid-fidelity prototype of the mockups I included in this case study.

Tasks included creating a lease agreement using both methods, editing a clause, and viewing the simplified version of clauses.

I focused on qualitative feedback because I'm looking for data regarding their feelings, emotions, and biases to keep iterating before launching the feature. I focused on observing:

  • Ease of use and comprehension
  • Reactions to AI-generated language
  • Moments of hesitation or confusion
  • Emotional response (confidence, frustration, delight)

Key findings:

  • All users felt like the process was quicker than the long form they have used for our document generation flow.
  • Most users preferred to add clauses manually rather than using the AI chat. They explained how they're skeptical of AI in general. They also stated how they trust an attorney rather than a chatbot.
  • Users who used the chatbot described it as advanced, adaptable, and quick. They felt like it eliminated unneccesary steps.
  • One concern raised was trust: users wanted to ensure AI-generated clauses were still legally valid. This insight prompted me to incorporate legal disclaimers. I told users we are adding an attorney review for an additional cost and they seemed interested in it.

What we learned

The tests confirmed that users were open to a new approach. The expiterience made them feel more in control and more confident about the documents they were generating.

It also validated our belief that a smarter experience doesn’t have to sacrifice accuracy or reliability. We will continue iterating as we gather more data and conduct further tests.