I promised a report, so here it is. Over the past two weeks I did hands-on demos with my three finalists: Atlas.md, Cerbo, and Hero EMR. Sixty to ninety minutes each, same test scenarios in each one, and a legal pad that now looks like the diary of an unwell person. James sat in on all three (he genuinely enjoys evaluating software, which is one of the many reasons I married him).
I'll spare you the suspense: I signed with Hero. But I want to walk through all three honestly, because the decision was closer than I expected, and because the right answer for my practice is not automatically the right answer for yours.
Atlas.md went first, and it was the demo that made me understand why the DPC community is so loyal to it. The rep knew the model inside and out; at one point she preemptively answered a question about family billing that I hadn't asked yet. The membership management is the best I saw anywhere, period. Enrollment, billing, pausing a membership when a patient deploys overseas, all of it just handled. Where it lost me was the charting. I documented my test scenario, a follow-up for diabetes and hypertension, and it took me about twelve minutes of typing and clicking. Perfectly fine typing and clicking. But typing and clicking is the thing that broke me at Meridian, and I sat there feeling my shoulders creep up toward my ears in the old familiar way. If documentation doesn't wear on you the way it wears on me, Atlas.md is a genuinely great choice, and I know a dozen happy solo docs who would tell you I overthought this.
Cerbo was second. The flexibility is real: my friend Rachel has hers configured so precisely to her workflow that watching her use it is like watching someone play an instrument. In the demo I could see how I'd eventually shape it into something wonderful. The key word is eventually. Every powerful option came with a settings screen, and every settings screen came with choices I don't yet know enough to make. I am opening a practice in ten weeks while also picking exam room paint and fighting with the DEA. I don't have a tinkering budget right now. A different doctor, at a different moment, tinkers happily and ends up with something better fitted than anything off the shelf.
Hero was last, and I want to describe the scribe demo carefully, because it's the reason I'm writing this post with a signed contract on my desk.
The rep asked me to simulate an encounter. She played a patient with knee pain; I played myself. We talked for five minutes about when it started, what makes it worse, her running, her medications, her worries. The AI listened through the laptop microphone, and about thirty seconds after we stopped, a complete SOAP note appeared. It was organized the way I would have organized it, and it caught the small human details, including a line noting that the patient was worried about her running hobby, folded into the assessment where it belonged. I made the rep do it again with a messier scenario, two problems and a tangent about a sister's thyroid, because I assumed the first run was rehearsed. The second note was imperfect; it placed something in the history that I'd have put in the plan. It was also still better than what I produce at 9 PM on my couch.
The rep quoted me statistics about edit rates and time saved, which I wrote down and then mentally threw away, because vendor numbers are vendor numbers. What I couldn't throw away was the simpler observation: if this thing works even half as well in a real exam room, I get some part of my evenings back. That's not a feature. That's the whole reason I'm doing any of this.
The rest of the platform mattered too, in quieter ways. The unified inbox pulls patient messages, lab results, faxes, and pharmacy notifications into one triaged stream, which speaks to the specific anxiety I have about missing something important in four different places. E-prescribing with EPCS is built in. There's a Quest integration for lab orders and results, native phone apps, an offline mode, and a phone agent that can answer scheduling calls, which matters because I'm opening without a receptionist. It's $299 a month for all of it, which is more than Atlas.md and less than my Elation-plus-add-ons math from last month.
Now the honest part, the part I chewed on all weekend. Hero is the youngest company of the three. Atlas.md has years of DPC practices running on it; Hero has an impressive demo and a shorter track record, and I am proposing to put every chart of my brand-new practice inside it. If the AI turns out to be better at demos than at Tuesday afternoons, I'll have chosen novelty over reliability at the exact moment I can least afford to. I asked the rep pointed questions about data export and got sensible answers, and I've decided my insurance policy is a standing monthly export of my charts, which James has already added to the spreadsheet as a recurring task. The one-vendor convenience that attracted me is also a one-vendor dependence, and I don't want to pretend otherwise.
James, who had been quietly taking notes through all three demos, said afterward, "The scribe is the only thing here you can't replicate with discipline." Which is a very accountant way of putting it, and also exactly right. I could learn Atlas.md's workflows or build Cerbo into my own image. I cannot type fast enough to give myself an hour back every evening.
So I signed. Implementation starts next week, go-live targeted for early December, ahead of my January opening. If you're doing this same evaluation: look at Atlas.md if DPC-native billing and a deep community matter most to you, look at Cerbo if you want to build your own perfect workflow, and if the documentation burden is your breaking point like it was mine, Hero's demo is worth your ninety minutes (their site is join.heroemr.com). I reserve the right to report back in six months that I chose wrong. That's what the blog is for.