Skip to content

How a New York Home Care Agency Cut Admin Work With Custom Software and AI

Work was spread across HHAeXchange, phones, fax, text and an old nursing system, with staff re-typing between them. What we built for one New York home care agency since early 2024, and where another agency should start.

Since 2024as the agency's engineering team, still ongoing
20+projects shipped
4,000+patient records migrated
iOS and Androidcaregiver app live on both stores
Key takeaways
  • One New York home care agency, more than 20 projects since early 2024, and the work is still going.
  • HHAeXchange is connected to everything else: the task system, phones and text, so referrals, timesheets and calls become assigned tasks without re-typing.
  • A nursing assessment platform replaced a legacy system: 4,000+ patient records moved over, in production for about a year.
  • Physician orders close the loop themselves: sent by fax with QR codes, matched back to the right patient when they return signed.
  • AI on the phones and in the office: voice agents, text campaigns, private assistants grounded in the agency's own policies, and a co-pilot for case managers.
  • Voice agents that can run without per-minute AI fees: the models can run on the agency's own hardware, bought once.

A New York home care agency had its work spread across HHAeXchange, a task manager, phones, text, fax and an old nursing system, with staff re-typing the same details between them. Since early 2024, over more than 20 projects, we connected those systems and built what was missing: a nursing assessment platform, AI phone agents and text campaigns, private AI assistants for staff, and a caregiver hiring app.

Home care runs on paperwork, phones and a scheduling system that does not talk to anything else. This agency had all three, plus a care-coordination arm working inside hospital systems. The client is not named here; everything below is drawn from our own project records. If you run an agency and want the short version, see what we build for home care agencies.

The agency, and where the hours were going

The client is a New York State licensed home care services agency (LHCSA), with nurses, home health aides and personal care aides, serving Medicaid and managed long-term care (MLTC) patients, together with a care-coordination business whose case managers document members in Epic and Foothold. Between them they employ office coordinators, nurses, case managers and a large caregiver workforce, and work with outside partner companies that refer patients and aides.

When we started, the work was spread across HHAeXchange, a task manager, phones, SMS, email and fax, an older nursing system and a lot of manual re-typing between them. The owner wanted two things: fewer hours lost to admin, and AI that actually helped staff not another tool to log into.

Nursing assessments and physician orders, without paper or re-typing

The largest single project is a nursing assessment platform that replaced the agency's legacy system. Nurses get their assigned assessments on a phone or tablet; managers review and approve them or send them back with comments.

  • A full patient record with more than 20 clinical sections: diagnoses with ICD-10 search, medications, allergies, falls, hospitalisations, equipment, mental health, pain, plan of care, and certification periods with expiry tracking.
  • Digital forms that match the paper ones, printed as PDFs: clinical assessment, medical and verbal orders, aide plan of care, emergency kardex, medication profile, supervisory visits, the HRA forms and discharge. Fields prefill from the patient record and sync across forms, and every change keeps its history.
  • Signatures on screen for nurse, patient, witness and representative, and an assessment cannot be closed until every task is complete and signed.
  • A physician-order loop that closes itself. Orders are generated with QR codes and sent by fax or email, one at a time or in bulk. When the signed order comes back, the system reads the QR code and attaches it to the right patient and assessment.
  • AI in the clinical flow: AI-drafted progress notes, and a medication-label scanner that reads a photo of the bottle.
  • HHAeXchange data such as contracts and visit schedules flows straight into the patient profile.

More than 4,000 patient records were migrated from the old system, and the platform has been in daily production use for about a year.

AI phone agents and text campaigns for patients and caregivers

A home care office spends much of its day on the phone. We put AI on the first touch and made sure every conversation ends up somewhere a person will see it.

  • An AI voice agent that works through call lists of patients, aides and partners, books appointments against staff availability with reminders, offers "press 1 for a callback", and transfers calls when it should. Every call becomes a task with an AI summary and the transcript.
  • An employment-verification line. Outside organisations call to verify a caregiver's employment. The AI answers, checks the caregiver in HHAeXchange, and texts back a PDF verification letter, around the clock.
  • AI text campaigns that send the first message and follow-ups, handle opt-in, opt-out and "not interested" automatically, answer questions from the agency's knowledge base, and hand the conversation to a person when needed.
  • A member-outreach platform for a Medicaid plan partner: multilingual AI calls and texts within allowed calling windows, attempt limits, opt-outs, callbacks and a warm hand-off to a community health worker, with campaign scheduling and usage billing.
  • An option with no per-minute AI fees. The phone agents can run with the language model, speech recognition and speech synthesis on the agency's own hardware. That means paying for the hardware once instead of paying a voice provider for every minute of every call. Phone numbers, and premium voices where used, are still billed.
  • The phone system wired in: call recordings and transcripts, and texts to dedicated lines, create inquiry tasks automatically.

If you are weighing up an AI phone agent for your own office, our guide to setting up and testing an AI receptionist covers what to check first.

Private AI assistants that know the agency's own policies

The agency did not want staff pasting patient details into a public chatbot. So every assistant we built answers from the agency's own material and respects who is allowed to see what.

  • AI support and training centres for both businesses. Staff ask questions in plain English and get answers from the agency's policies, workflows and training videos, with links to the source. Permissions by team and role are enforced inside the AI, not just in the menu.
  • Training by conversation. Admins upload a policy or a training video; the AI builds the module and staff train by chatting with it, with progress logged.
  • A knowledge base that updates itself from the task system, spreadsheets, website pages and media on a schedule, so the assistants and voice agents never answer from stale files.
  • An assistant that runs on the agency's own hardware. A local AI model on a machine in the office, connected through tool servers to HHAeXchange, the task system, file storage, SMS and voice. Staff can ask what a caregiver earned last month or tell it to call a patient in Spanish.
  • A co-pilot for case managers in Epic and Foothold. It works alongside the case manager in the browser, pulls the member's details from the task system by Medicaid ID, drafts encounter notes using task-specific instructions, and completes audit spreadsheets against the agency's own audit guide.

When the AI vendor retired the interface these assistants were built on, we moved every assistant across three portals to the new one with its files, knowledge and tools intact.

A caregiver hiring app and self-service

  • A caregiver hiring app on iOS and Android: sign up, a seven-step application that can be saved and finished later, credential and physical-exam uploads, direct deposit, and an onboarding tracker. It is published on the App Store and Google Play.
  • Self-service for caregivers: timesheets, pay stubs and W-2 forms, which caregivers can also ask the chatbot about.
  • Compliance in one view: TB, flu, annual physical, evaluations and supervisory visits synced from HHAeXchange.
  • A recruiter text inbox with new, updated and pending statuses, so no applicant's message is missed.

HHAeXchange integration: referrals, timesheets and calls in one place

  • A referral app and portal for partner companies. Partners refer patients and aides from their phone, with voice notes and signed compliance forms, and each referral becomes a tracked task for the care team. The portal also handles partner billing.
  • A portal for managed long-term care plans: tickets, referrals, questions and invoices in one place instead of email threads.
  • HHAeXchange integrations across the board: coordinator sync, patient and caregiver search, visits, timesheets and Medicaid ID lookups.
  • Automatic tasks: timesheets, referrals, job applications, calls and texts become de-duplicated tasks with retries, so nothing falls through when an integration hiccups. An AI agent writes a daily report of high-risk tasks.
  • A company-wide move from one task manager to another in about seven weeks, including all the automation hidden inside the agency's websites and forms.

How we work with a home care agency

This is not one big project with a handover at the end. It is a standing engineering team: a senior engineer leads, work ships in small releases that are tested on production, and the next priority is agreed as the last one lands. Every hour is logged against a task in our own time tracker, so there is a record of the work behind each month. You can see how that works on how we run the work.

The systems involved: HHAeXchange, Epic, Foothold, a task manager, file storage, SMS and voice providers, cloud AI models and local ones, and native iOS and Android apps.

If you run a home care agency: where to start

Most agencies do not need all of this, and nobody should start with all of it. The useful first step is usually one workflow that eats the most staff hours: physician orders, the phones, caregiver onboarding or re-typing between HHAeXchange and everything else. Book a free 30-minute review and we will tell you which one we would do first, and what it would take. There is more on what we build for home care agencies.

Frequently asked questions

What software does a home care agency need beyond HHAeXchange?

HHAeXchange handles scheduling, visits and billing. This agency also needed nursing assessments and physician orders, phones and text messages that create tasks, caregiver hiring and onboarding, and partner referrals. We built those as custom tools and connected each one to HHAeXchange so nothing is typed twice.

Can you integrate with HHAeXchange?

Yes. For this agency we pull patients, caregivers, visit schedules, contracts, pay rates and compliance items from HHAeXchange into custom tools, AI assistants and phone agents, and keep fields such as the coordinator in sync with the task system.

Do you work with Medicare-certified home health agencies?

Our deepest experience is with a licensed home care services agency (LHCSA) serving Medicaid and MLTC patients. Nursing assessments, physician orders, HHAeXchange integrations and AI phone work carry over directly to a certified home health agency, but we have not built Medicare-specific tools such as OASIS or Medicare billing, so we would scope those with you first.

What does an AI phone agent cost a home care agency to run?

It depends on how it is hosted. Most services bill for every minute of every call. The agency's phone agents can also run with the language model, speech recognition and speech synthesis on hardware it owns, so it would pay for the hardware once instead of per-minute AI fees. Phone numbers are still billed, and so are premium voices if it uses them.

Can an AI phone agent work for a home care agency?

Yes, for the right calls. Here it handles outbound call lists, appointment booking with reminders, callbacks and employment-verification calls, and every call becomes a task with a summary and transcript so a person can follow up.

Is it safe to use AI with patient and caregiver data?

It depends on how it is built. This agency's assistants answer only from the agency's own documents and systems and respect role and group permissions. One assistant runs on a local model on hardware the agency owns, so what staff ask it stays on site. Staff are not pasting patient details into a public chatbot.

Is the software HIPAA compliant?

HIPAA compliance is not a label on software; it depends on how a system is built, hosted and run, and on the agreements between the agency and each vendor. We design for it with role-based access, change history, and options such as local AI models, and we go through the specifics for your agency on a call rather than making a blanket claim here.

Does this replace EVV or our scheduling system?

No. Visits continue to be scheduled and recorded in HHAeXchange. Our tools read visit, timesheet and compliance data from it and use it elsewhere, for example in the patient record and the caregiver app.

Can you work alongside Epic and Foothold?

Yes. The care-coordination team documents in Epic and Foothold, and we built a co-pilot that helps case managers draft encounter notes and complete audit spreadsheets while they work in those systems.

Do we have to replace the software we already use?

No. Most of this work connects what the agency already had, such as HHAeXchange, its task manager, phones and SMS, and adds custom tools only where nothing off the shelf fitted, such as the nursing assessment platform.

How much does custom home care software cost?

We work on a monthly plan, not a fixed project price: $1,200 a month for 40 hours, $2,100 for 80, or $3,600 for 160, and you can change plan or cancel at any time. Most agencies start with one workflow, so the first plan depends on how much you want moving at once.

How do you work with a home care agency long term?

As an ongoing engineering team, not a one-off project: a senior engineer leads, work is planned in small releases, and every hour is logged against a task in our time tracker.