TL;DR: Refill workflow automation for Elation Health practices works by having an AI agent read each incoming refill request, check the chart against your protocols (last visit, labs, medication type, controlled-substance flags), and queue the clean ones for provider sign-off inside Elation. Providers still approve every refill. Controlled substances, overdue visits and anything ambiguous go to a person.
What does refill workflow automation for Elation Health actually do?
Think of it as a pre-screening layer that sits in front of your providers' sign-off. An AI refill agent reads each incoming request, pulls the relevant chart data, compares it with the rules your practice has written down, and sorts the request into one of three buckets: ready for sign-off, needs a human look, or not eligible.
It doesn't replace Elation, and it doesn't replace the provider. The refill still gets approved, modified or denied in Elation, and the action lands in the patient's chart like any other.
What changes is who does the chart-checking. Today that's usually a medical assistant or nurse who opens the chart, scans for the last visit date, looks for a recent lab, confirms the drug isn't on a restricted list, and then routes the request to a provider. The agent does that same pass in seconds and hands the provider a request that's already been screened.
Here's the scope worth holding in your head:
- Automated: reading the request, matching it to the patient and medication, checking protocol rules, drafting the recommended response.
- Still human: the clinical decision to approve, every controlled-substance request, and any exception your protocol doesn't clearly cover.
If someone pitches you "fully autonomous refills," ask who signs. In a well-run practice, a licensed provider does.
How do refill requests reach your practice?
Refill requests don't arrive through one clean pipe. Most practices see four.
Pharmacy requests come electronically. In Elation, pharmacy refill requests are filed to the patient's chart, assigned to a provider, and shown in the Rx Requests inbox on Practice Home. Requests older than 30 days are archived automatically, which matters if your queue ever backs up.
Patient portal messages are free text. "Can I get my lisinopril refilled?" has to be matched to a medication, a dose and a pharmacy.
Phone calls turn into notes or tasks, usually typed by whoever answered.
Faxes still show up from some pharmacies, and they're the slowest to process because someone has to read and key them in.
An agent earns its keep on the messy channels. The pharmacy request is already structured. The portal message, the voicemail and the fax are where staff lose time. A good implementation normalizes all four into the same record: patient, medication, pharmacy, request date and channel. Then every request goes through the same checks, regardless of how it came in.
One failure mode to plan for: patient matching. If a fax lists a nickname or a transposed birthdate, the agent should flag a low-confidence match and stop, not guess.
What does the agent check before a refill reaches a provider?
This is where your protocols become rules. Most practices already have a refill policy, often a laminated sheet or a page in the staff handbook. Automation forces you to make it explicit.
A typical check list looks like this:
- Last visit date. Many practices require a visit within the past 6 or 12 months for chronic medications.
- Required monitoring. If a medication needs a lab (a thyroid medication needing a TSH, for example), is there a recent result?
- Medication type. Is it a maintenance drug on your approved list, or something that always needs a provider's eyes?
- Controlled-substance flag. Anything scheduled gets routed out of the automated path.
- Quantity and refill count. Is the request consistent with the original prescription and the pharmacy's records?
- Duplicates. Has this same request already been answered this week?
Here's an illustrative example, not a benchmark. Say a six-provider primary care group gets 40 refill requests a day, and each takes about three minutes of staff time to chart-check and route. That's two hours of screening daily, before any provider sees a single request. If the agent clears even half of the volume as ready-for-sign-off, you've bought back an hour of staff time every day. Swap in your own numbers; the arithmetic is the point.
Elation's own design shows what "ready" means. Its Bulk Refill feature preselects only refill-ready medications: non-controlled substances with a name, sig and quantity already specified. Data quality drives automation, too. Medications that were only documented or imported may not be complete enough to act on.
Where does the provider stay in the loop?
Everywhere that matters. The agent prepares; the provider decides.
In Elation, the response options for non-controlled refills are approve, approve with modifications, change the number of refills, deny, deny with a reason, or reassign to another provider. An automation layer doesn't need new buttons. It needs to put a recommended action in front of the provider so that approving takes seconds, not minutes.
In practice, the provider's view should include:
- The request and its source
- What the agent checked and what it found (last visit, lab dates, flags)
- The recommended response, with the reason
- Anything unusual, called out at the top
Providers who see a request that's already summarized tend to trust it faster. Providers who see a black-box "approved" label tend to rebuild the whole check themselves, which erases the savings.
There's also a policy decision hiding here: which refills, if any, can a staff delegate or a protocol approve without a provider reviewing each one? Some practices have standing-order protocols for low-risk maintenance drugs. Others require a provider to touch every refill. Refill workflow automation for Elation Health works under either rule. It just has to follow the one you've chosen, and your compliance lead should sign off on that choice before go-live.
If you want a practical test, ask a vendor to show you the audit trail for a single refill: who or what checked what, when, and who made the final call. If they can't show one, keep looking.
What still needs a human?
More than vendors usually admit. Plan for these four.
Controlled substances. Federal rules are strict. Under the DEA's regulations, no Schedule II prescription may be refilled, and Schedule III and IV prescriptions can't be refilled more than five times or more than six months after issuance. Elation treats these separately too: its documentation notes that Rx delegates aren't allowed to ePrescribe controlled substances, and the provider has to complete the response with their EPCS token. An agent can sort and tee up the request. It can't sign it.
Overdue visits. If the patient hasn't been seen in 14 months and your policy says 12, the agent shouldn't quietly approve a refill. It should route a task to scheduling: book the visit, and (if your provider allows) send a short bridge supply request for the provider to consider.
Ambiguous requests. A patient who asks for "my blood pressure pill" and has three on the list needs a human to clarify.
Clinical red flags. A new symptom mentioned in a portal message, a dose change request, an interaction concern. The agent's job is to notice and escalate, not to interpret.
The honest version of the pitch: automation takes the routine middle of the distribution and leaves the hard edges to people. That's usually the majority of volume by count, and a smaller share by clinical risk. Ask any vendor, Honey Health included, for the exception rate on a live deployment, not a demo.
How do exceptions escalate, and what changes for your staff?
Exceptions need a destination, not just a flag. The common design is a small set of named queues:
- Provider review: clinical questions, dose changes, anything with a red flag
- EPCS queue: controlled substances for the provider to handle with their token
- Scheduling task: overdue visits and missing labs
- Front-desk follow-up: patient didn't specify the drug, pharmacy mismatch, missing information
Each exception carries its reason, so the person who picks it up doesn't start from scratch.
For your front desk, the change is mostly less retyping. Fewer faxes to key in, fewer "which pharmacy?" callbacks, and a clear queue for the ones that really need a human. Phone staff stop acting as refill routers and go back to answering patients.
For clinical staff, the screening pass shrinks. Medical assistants and nurses spend less time opening charts to check dates and more time on the exceptions that need judgment. Expect a transition period: people who've owned the refill queue for years may distrust a new layer. Run it in shadow mode first, where the agent recommends and staff compare it with what they would have done, for a couple of weeks.
For operators, set up three numbers before launch so you can tell whether it's working: average time from request to response, share of requests cleared without staff chart review, and the number of corrections providers make to the agent's recommendations. If corrections stay high, your protocol rules need tuning, not your providers.
On the broader economics, the 2024 CAQH Index estimates the industry could save about $20 billion a year by moving manual and partially electronic administrative transactions to fully electronic ones. That report covers medical and dental transactions such as claims and eligibility, not refills, so treat it as context for the size of the administrative problem, not a refill benchmark.
How does Honey Health fit alongside Elation?
Honey Health is one example of this pattern: a Refill Management agent that works alongside the EHR rather than replacing it. The agent reads incoming requests from the channels you use, checks them against your protocols, and prepares the work for provider sign-off while the record of what happened stays in Elation.
When you evaluate any option, including ours, the questions that separate a real implementation from a demo are practical ones:
- Does it write back to the chart so there's one source of truth, or does it keep a parallel record?
- Can you edit the protocol rules yourself, or is every change a support ticket?
- What does the exception queue look like, and how are controlled substances handled?
- Can you see an audit trail for any single refill?
- How long does it take to run in shadow mode before it touches live requests?
A useful rule of thumb from the American Medical Association's practice redesign team is that reducing refill touches upstream helps too. The AMA's Marie Brown has said that writing 90-day prescriptions with four refills would cut refill work in half, saving the physician about an hour and the team another hour or more. That's her estimate, not a measured study, but it points at something real: automation and smarter prescribing habits stack. If your providers can write longer supplies for stable patients, your agent has less to process.
Frequently Asked Questions
Does refill automation replace the provider's approval?
No. The agent screens and recommends; a licensed provider approves, modifies or denies in Elation. Some practices allow protocol-based approvals for low-risk maintenance drugs under a standing order, but that's your policy decision, not something the software should decide for you. Controlled substances always stay with the provider.
Can an AI agent handle controlled-substance refills?
It can sort and prepare them, but it can't sign them. Schedule II prescriptions can't be refilled at all under federal rules, and Elation requires the provider's EPCS token to complete controlled-substance responses. Treat any promise to "automate" controlled refills as a red flag and ask exactly what that means.
What data does the agent need from Elation?
At minimum: the patient's medication list, prescription history, last visit date, relevant lab results, allergies and the pharmacy on file. The cleaner your Elation medication list is, the better refill workflow automation for Elation Health performs. Imported or undocumented medications that lack a name, sig or quantity are hard to act on, for people and software alike.
How long does setup take?
It depends on how clear your refill protocols are. Writing down your rules (visit intervals, monitoring requirements, approved drug lists) is often the longest step. For refill workflow automation for Elation Health, plan on a shadow-mode period of a few weeks, where the agent recommends and staff compare, before anything goes live.
How do we measure whether it's working?
Track time from request to response, the share of requests cleared without staff chart review, and the rate at which providers correct the agent's recommendations. Add patient callbacks about refill status. If response time drops and corrections stay low, the rules are right.
