They replaced a field app they disliked with software built around the way they work.

The company was using a field app its team did not like, along with manual processes for moving information from property visits into operations and client communication.

The goal was not to recreate that app feature for feature. It was to replace the fragmented process with one company-specific system built around how account managers and operations teams actually work.

Account managers work across commercial properties where Wi-Fi may not exist and cell service may be weak. A visit can include dozens of photographs. The current production record is 60 photographs on one visit, and the application is designed not to reject a valid visit simply because it is large.

The phone saves drafts and photographs locally first. The small visit record can submit independently while photographs continue uploading as the app finds signal. Uploads can resume, repeated attempts converge on the same visit, and the interface distinguishes work saved on the phone from work fully received by the server.

Issues, enhancement opportunities, notes, reasons, and photographs now enter the same structure on every property walk. That record becomes the source for the work that follows.

After submission, AI produces a structured Spanish version of the English field record and drafts action-led titles in both languages. Incomplete translation is held for repair rather than sent to a Spanish-speaking crew as if it were finished.

The property’s HubSpot record identifies the responsible production manager. Actionable findings then become bilingual Microsoft To Do tasks with photographs attached. Irrigation work can route separately. If the mapping is missing or ambiguous, the system holds or flags the work instead of guessing where it belongs.

Visits with no crew-actionable issue correctly produce no task.

An authenticated review workspace lets supervisors and account managers inspect visits, photographs, English and Spanish notes, routing status, and exceptions. The interface shows whether work was created, skipped because no crew action was required, or held because the automation needs attention.

The same stored photographs are also mirrored into a client-owned SharePoint archive, organized by property and visit date.

Botworks also built a separate PDF application around the visit record. An employee can start from a completed visit, choose the issues and opportunities worth presenting, edit the title and scope, select and reorder photographs, mark up an image with a red pen, and generate a branded PDF.

The software preserves saved drafts and generated-file history. It does not claim whether each PDF was sent or produced revenue, so the case study stops at what the system can prove.

Task routing runs every ten minutes and photo archiving runs hourly. Health checks cover the database, Microsoft identity, the task drainer, translation holds, and routing configuration. Failures are surfaced, diagnosed, repaired, and turned into tests or measurements so the same class of problem is less likely to return silently.

The company controls its data and company-specific software. That does not pretend the product can run forever without ownership. Botworks operates it while engaged and is documenting the system so future owners, including employees working with AI agents, can understand and extend it.