Company reportedLive since Oct 2024Added Oct 2026

AppFolio

AppFolio's Realm-X assistant saves property managers over 10 hours a week, the company reports

AppFolio built Realm-X, an assistant inside its property management software that generates reports and drafts messages to residents and vendors. After a June beta it went to all customers in October 2024; AppFolio says users save over 10 hours a week and 26 seconds per message.

Key figures

Copied exactly from the sources. Copy any figure with its citation.

over 10 hours per week

Average weekly time saved by Realm-X users on tasks on their to-do lists (AppFolio's October 2024 press release)1

26

seconds

Average time saved per Realm-X-generated message compared with a message composed without its help, over the last 30 days before the release1

11.9

hours a week

Estimated weekly time savings by property managers on communication-related tasks (AWS case study)2

~40%

to ~80%

Performance of Realm-X's text-to-data feature (turning a plain-language question into a data query) after adding dynamic few-shot prompting (LangChain customer story)3

What it took

CostNot disclosed
TeamNot disclosed
TimelineNot disclosed

Cost, team size and timeline were not disclosed in the sources.

The problem

Property managers spend much of their day on routine work: pulling reports, answering residents and writing to vendors. An AWS case study about AppFolio says communication can consume up to half of a property manager's working day, and that answering a question like the occupancy or vacancy rate used to mean asking in-house tech support for a custom report.

What they built

AppFolio added Realm-X to its property management software. Realm-X Assistant is a conversational helper that generates reports, manages vendor interactions and drafts personalized emails and text messages. Realm-X Messages is a redesigned inbox that sorts resident messages and suggests clear, personalized replies. A third part, Flows, automates workflows for some plans. This case covers the Assistant and Messages.

How they did it

AppFolio began a beta in June and, by its October 2024 conference, made the Assistant and Messages available to all customers. It used LangChain tools to build and monitor the assistant, including LangSmith to trace traffic and collect feedback when users submit an action Realm-X drafted. An AWS case study, which carries no publication date, says AppFolio tested several models on Amazon Bedrock and picked Amazon Nova Pro for a balance of speed, cost and quality. It does not say when that choice was made. Few-shot prompting, which shows the model relevant worked examples for each request, lifted its text-to-data feature.

Results

AppFolio reports an average of over 10 hours per week saved by Realm-X users on items from their to-do lists. In the 30 days before the release, customers sent almost half a million messages drafted by Realm-X, each saving 26 seconds on average. AWS gives an estimated 11.9 hours a week saved on communication-related tasks. LangChain reports text-to-data performance rising from ~40% to ~80%.

What to copy

Run a beta, publish the saving per task as well as per week, and keep your model choice flexible so you can retest as new models appear.

Lessons for your project

  1. Use a beta to measure time saved and adoption before a full release. AppFolio ran a beta from June and says it released widely once adoption was strong.

  2. Choose a measure users can feel, such as time saved per message or per week, and say how it is stated: an average, an estimate or a user report.

  3. Plan to swap models. AppFolio built so model providers could be changed without code changes and evaluated models on speed, cost and quality before settling.

  4. Feed good examples into prompts. AppFolio credits dynamic few-shot prompting, which pulls relevant examples per query, for the text-to-data gain.

Caveats

  • The time-saved figures are AppFolio's own and no method is given. 'Over 10 hours per week' is an average of Realm-X users; the 11.9 hours is an AWS 'estimated' figure from an undated vendor case study. The two are not the same measurement and are not combined.
  • The AppFolio release does not name a model. The AWS case study says Amazon Nova Pro powers Realm-X Messages and Realm-X Assistant. The AWS page shows no publication date and links a '2025' benchmark report, so it appears to postdate the October 2024 release. It is therefore unclear which model produced the October 2024 figures.
  • The release gives general availability of Realm-X Assistant and Messages on 30 October 2024 after a beta launch 'in June'. The deployed date uses the October announcement; the release does not print the beta year beside June.
  • The ~40% to ~80% text-to-data figure is from a LangChain customer story, a vendor that sells the tools it describes, and gives no test set or method.
  • The press release was read from an archived copy (web.archive.org) because the live page timed out.
  • A quote from a customer in the release (a property operations specialist at Fairlawn) says the tool replaced over 18 hours of weekly work. It is one customer's statement and is not used as a metric.

Sources

  1. AppFolio is Powering the Future of Real Estate Using Generative AI and Amazon Nova

    Amazon Web Services · Undated · Vendor case study · aws.amazon.com

Summarized in our own words from the sources above. Spotted an error? See corrections.