6,000+
Property case studies generated
Real estate marketing platform turns listing data into case studies in seconds with Claude
Built the AI capabilities behind KeyStory, turning a sold MLS listing into a written property case study, brand-matched design, and platform-ready social posts, with an in-app copilot that reads the live web.
50+
Brokerages actively generating
<40 seconds
Median time from MLS listing to publish-ready case study
Client Overview
Client: Testimonial Tree, Inc.
Industry: Real estate technology
Company size: Small
Product: Claude Sonnet 4.6
Location: Florida, United States
Testimonial Tree builds reputation and marketing software for residential real estate agents, teams, and brokerages. Its KeyStory product turns closed deals and client reviews into ready-to-publish marketing, including property case studies, social posts, and embeddable review widgets.
Challenges
A real estate agent's strongest marketing assets are the homes they have sold and the clients behind those sales. For Testimonial Tree, the challenge was that turning those assets into marketing demanded more time and design skill than most agents could spare.
Turning a sale into a story was slow, manual work
Each case study meant pulling listing facts from the MLS, sourcing local market statistics, choosing photos, and writing several hundred words of copy styled to the agent's brand. Given that effort, most agents never wrote one, and their sales stayed as raw MLS records and one-line testimonials that did little to win new clients.
Every social network demanded its own version of the post
Facebook favors a short teaser with one hero image, Instagram an energetic carousel with dense hashtags, and LinkedIn a professional, data-led tone. Agents rewrote every post for each network and also had to judge which photos would perform and where overlay text could sit without hiding key features of the home, which turned a routine post into design work.
The assistant couldn't see the web
Users needed guidance on the platform as well as help with live web content, such as summarizing a listing page or reviewing a brokerage's website. A conventional chatbot could only handle the first, which sent users to a search engine mid-task. Adding web access the usual way meant building and running crawling and scraping infrastructure alongside the product.
Solution
Simform built KeyStory’s AI capabilities on Claude Sonnet 4.6 and integrated it server-side through Anthropic’s Python SDK. One model writes the copy, reads the photos, extracts the branding, and powers the in-app assistant. A shared layer beneath all four handles the guardrails, keeps each tenant’s data separate, and tracks cost per call. Because Claude reads the listing’s actual photos alongside its facts, the writing describes the specific home that sold.
Listing data and photos in one call
When a real estate agent selects a sold listing, KeyStory pulls its details from the platform’s own database and calls the MLS API only as a fallback, fetching photos, market statistics, and community data in parallel.
Claude receives these facts, plus up to six property photos, in one call and writes each section in the agent’s first-person voice. A companion call reads the agent’s website or logo and extracts brand colors and typography, so the page matches their identity.
Exact-value rules with automatic retries
Both prompts carry the listing’s exact values, such as address, square footage, and year built, and prohibit placeholders like “N/A”, so a published page never shows a figure the listing does not support.
If a response isn’t valid JSON, the service retries once with corrective instructions. Each job runs in the background with a trace ID the app uses to check progress, and every read and write stays inside that tenant’s own section of the database.
A voice registry and vision-based photo ranking
A central registry holds each network’s voice, slide limits, and hashtag rules. Claude uses it to generate the caption, hashtags, and slide plan for Facebook, Instagram, and LinkedIn in one call, and agents can regenerate a single caption or slide on its own.
In a separate vision call at zero temperature, Claude scores each listing photo, labels what it shows, and marks where overlay text can sit, producing the same ranking on every run.
Anthropic-hosted web search and fetch
When a message contains a URL, the copilot routes it to Anthropic’s server-side web search and web fetch tools, which run inside the same API call. Search stays within the domains the user referenced, and each tool is capped at three uses per message.
If the web path fails, the copilot answers through its standard path instead. Out-of-scope questions, such as financial or legal advice, get a scripted reply that points the user to a KeyStory capability.
Business Impact
With the Claude-powered capabilities live across more than 100 tenant organizations, a closed deal now becomes a case study, platform-ready social posts, and brand-matched widgets, without the manual work that kept most agents from starting.
Case studies generated in seconds
A multi-section case study, grounded in the listing's exact figures and its actual photos, now takes a median of under 40 seconds to generate. Agents have produced more than 6,000 of them, with 99.9% completing successfully, and every price, square footage, and year built comes straight from the listing record.
A ready post for every network
Agents get a caption, hashtags, and a slide plan for Facebook, Instagram, and LinkedIn in one step, each written the way that network rewards, with the strongest photos already chosen and the space for overlay text already marked. If a caption or a single slide misses, it can be redone on its own.
Web answers inside the platform
When a question points to a web page, the copilot reads that page and answers from what it finds, so users stay inside KeyStory instead of breaking off to search. Testimonial Tree gained that capability without building or running its own crawler.
Trusted by the World's Leading Companies
“After researching and qualifying several software development service options for the development of a cross-platform application, we fortunately selected Simform as our partner. This turned out to be a very good decision overall, as their performance has been outstanding. The complete engagement from the Simform staff is remarkable. Their team diligently assesses and executes in a very competent manner.”
Sonny Cutwright, Operations Manager
Bon Appetit
“Simform rearchitected and modernized our subscription management & billing platform as part of the multi-phase development initiative. The team quickly understood our business requirements and delivered a modern, modular architecture aligned with our scaling needs. Without disrupting our operational workflows, Simform handled complex data migration with precision and supported our legacy modernization journey with seamless ERP integrations.”
Robel Yemane, Head of Engineering
Privilee
“Simform was an invaluable partner in our system modernization, moving beyond analysis to deliver true product engineering advisory. They analyzed complex legacy workflows into clear, validated, developer-ready user stories. The clarity and strategic consistency they introduced was foundational, directly influencing our cloud architecture and product design. Their work effectively accelerated our discovery phase and time to market.”
Director of Engineering
Datamark
“We partnered with Simform to centralize our preclinical research data. They understood technical data challenges and scientific workflows. They helped us build a platform for data ingestion, OCR-based extraction, standardization, analysis, and visualization. Our researchers can use natural language or a visual query builder to analyze complex datasets, while real-time charts and statistical outputs speed up discovery. We liked the ability of the team at Simform to combine strong data engineering capabilities with practical analytics experience.”
Data and Analytics Manager
Preclinical Research Analytics Platform
“The cloud migration was a great sucess. Very satisfactory, seamless and increased our productivity. The most impressive part about Simform is their dynamic and well-versed team. Anytime there was a concern, we were able to communicate and have it rectified immediately.”
President
Trusted Community Services Organization
“Simform led the discovery phase for our digital infrastructure overhaul, bringing a mix of deep technical expertise and a truly collaborative spirit. They took our complex requirements and turned them into a solid product blueprint and a clear, prioritized roadmap. Everything stayed on track thanks to their structured project management. The level of strategy and architecture they brought to the table gave us the certainty we needed to take the next steps.”
Department Manager
Eye Recommend
“Our overall experience with Simform was very positive. They worked as an end-to-end technology partner and integrated well with our engineering and operations teams. This was not just about building a mobile app; it also required careful thinking around data quality, workflows, and long-term system design. Simform brought a practical, product-minded approach and designed an Azure Data Factory pipeline to clean and organize data before it reached the app and admin dashboard.”
Marketing Manager
Patient-Facing Digital Application & Data Platform
“We’re very satisfied with Simform as our engineering partner. We wanted a platform to support the full lifecycle of our field inventory and ease operations for field reps, hospitals and analysts. They translated our complex requirements into an easy-to-use solution and handled development from UI/UX to backend and DevOps. The app works seamlessly across user groups and has a first-class, consistent and simple interface.”
Sales Manager
Field Inventory & Operations Platform
Let's talk
Hiren Dhaduk
Creating a tech product roadmap and building scalable apps for your organization.
Call Us Now
