Product and AI development
A Next.js application with accounts, a database and AI-assisted document generation. Structured inputs, templates and output checks support the document workflow.
Owned venture · NJ Landlord Forms
I built a document service for New Jersey landlords, connected it to payments and email, and developed a repeatable process for bringing people in through search.
App launched May 3, 2026. Search results shown for completed months. No paid advertising.
The business to build
Landlords searching for a requirement often have a task to finish: calculate a deposit, understand a notice or prepare a document. The site needed to help with that task and give people a way to buy and use the product.
Search leads to a practical guide, calculator or resource.
Structured inputs feed an AI-assisted generation process.
Payments connect customers to their products. Email follows up based on what they do next.
This is a business I built and operate, not a client engagement. My work covers the website, document workflows, database, Stripe payments, Resend email, SEO and measurement.
What I built
I designed and built the product and the systems around it, from the first search visit to document delivery and follow-up.
A Next.js application with accounts, a database and AI-assisted document generation. Structured inputs, templates and output checks support the document workflow.
Calculators for deposits, rent and property-management costs, plus a lease review tool and local registration resources. Selected calculator results feed directly into document drafts.
A repeatable content-development process built around search needs, with related guides, internal links, structured data and checks on which pages are useful enough to index.
Stripe checkout connected to purchased kits and document credits, so the product can recognize what someone owns and give them access.
Resource delivery, lead nurture and product-specific onboarding through Resend, with follow-up for unfinished documents and support for checkout recovery.
Search Console, GA4 and database records show where people arrive, which tools they use, where they sign up and whether they move into a document draft.
Search growth
Useful tools and focused content began earning traffic. Google clicks increased each month from May through August, alongside growth in organic sessions.
May to August
May to August · All search engines
286 Google clicks and 559 organic sessions. Organic search accounted for 81% of the site’s 692 sessions. This is a partial month, not a forecast.
Source: owner-supplied growth report, pulled September 22, 2026, from Google Search Console, GA4 and the production database. Google clicks and GA4 sessions are different measures. May session totals include launch testing.
A repeatable search process
I developed a process for finding the questions landlords were searching for, creating guides and tools to help, and using search performance to decide what to improve or build next.
The content connects related questions through guides, resources and useful tools. Topic selection, internal linking and performance review are part of the process, so each new page has a purpose within the site.
The work also includes reviewing existing pages and improving or limiting content that doesn’t have enough useful information.
The security deposit calculator was the site’s top page by Google clicks in the report. Other calculators and practical guides also brought people in.
Building tools gives the business another way to answer a searcher’s question and introduce the product.
| Page | Clicks | Impressions |
|---|---|---|
| Security deposit calculator | 110 | 1,558 |
| NJ Section 8 landlord guide | 93 | 3,317 |
| NJ lease agreement requirements | 69 | 4,841 |
| Wear and tear vs. damage guide | 66 | 1,767 |
Using the data to improve the product
The tools answered a question, but the journey often stopped at the result. On August 29, I connected the security deposit, rent increase and Section 8 calculators to pre-filled document drafts.
A visitor enters the details for their situation.
The result opens a document draft with relevant information already filled in.
| Measure | Aug 5–28 | Aug 29–Sep 21 |
|---|---|---|
| Site sessions | 589 | 765 |
| Unique tool users | 51 | 56 |
| Document drafts started, all entry points | 7 | 18 |
| Drafts per 100 sessions | 1.2 | 2.4 |
| Documents generated | 3 | 6 |
Four of the 18 drafts in the later period started from calculator links. Five drafts were recorded as completed; completion tracking began August 29, so there is no comparable earlier count.
These are small, observational counts, not a controlled test. Draft recovery, checkout recovery and Starter Kit document credits also launched August 29. The increase cannot be attributed to the calculator connection alone. Two earlier drafts came from one disposable email address that may have been a test. Drafts per 100 sessions is an activity ratio, not a unique-person conversion rate.
Beyond the first visit
I built the email system around the customer’s place in the journey. A new subscriber needs a useful resource. A buyer needs help using what they bought.
Visitors can request resources in guides and local registration pages, or have calculator results emailed to them. Six printable resources give people a practical reason to sign up.
Each signup records its placement, so I can see which parts of the site build the list.
A lead-nurture sequence delivers the requested resource and introduces relevant tools and guides. Links carry campaign tags so return visits can be measured.
Buyers move off the sales sequence and into onboarding for their Starter Kit or Eviction Pack. Follow-up can respond to whether they have created a document or still have unused credits.
An unfinished draft can trigger an email with a link back to the saved work. Checkout recovery links support returning to an unpaid order. Replies reach a person, and subscribers can unsubscribe directly.
The email templates and sending rules are versioned with the site. Database checks and send identifiers guard against duplicate sends, and purchase status controls which sequence a person receives.
Document generation
A document service needs controls over what it produces. I built validation and review steps into the workflow, with template-based assembly for selected notice types.
For selected notices, the system validates structured information and renders it into defined templates. That gives the application control over document structure instead of leaving the whole document to a free-form response.
Checks look for required user-supplied information and citations outside the allowed list. A failed validation prevents the document from being saved and records the failure for investigation.
Automated tests cover document-generation behavior. A separate Python checker reviews template citations and registered review fields, helping catch inconsistencies when templates change. It supports development; it is not the live validator.
A verification process uses municipal and other primary sources to support local information. Unresolved document wording is tracked for further review. Technical validation checks known rules and required information; it does not establish legal correctness on its own.
First commercial milestone
The working app launched May 3 with its database and live checkout. The first SEO-driven purchase followed on August 20. An earlier static version of the site had been deployed in February.
That is an early paid milestone. The work continues on the product, the customer journey and converting search traffic into repeatable revenue.
What this shows
This project brings product development, AI workflows and customer acquisition into the same business. If you have a useful tool in mind, I can help work out what it needs and how people will find it.
Talk about your project Explore NJ Landlord Forms