I run the billing operations for independent medical practices, and I build the software that runs them. Those two things are usually done by different people at different companies, which is why most practices wait weeks for a report that should take a minute.
I founded Bonfire Revenue in November 2016 and still run it. We handle revenue cycle for independent multi-specialty practices across the country — behavioral health, internal medicine, dental, nurse-practitioner-owned clinics, skilled nursing — on contingency, billed as a percentage of what we actually collect. If the claims don't get paid, we don't get paid. That arrangement is the reason everything below exists.
Early on I did the work by hand like everyone else: eligibility checks in one portal, claim scrubbing in another, patient statements in a third, and a spreadsheet to tell me whether any of it had worked. Hiring more people to do that faster was the obvious move and the wrong one. So I learned to build instead. Across four platforms there are now more than 200 production automations behind the firm — 59 in Make and Boost.space, roughly 70 in Pabbly Connect, roughly 80 in TaskMagic — plus several hundred more built inside the project systems we run day to day. They cover patient statements and payment collection, client onboarding, staff provisioning, denial routing, content and campaign delivery, and the KPI reporting every client sees. I architected all of it and I maintain all of it.
The credentials are real but they're not the interesting part. I'm a Certified Professional Biller and a Certified Billing and Coding Specialist, which means I can read a denial and tell you whether it's a coding problem, a credentialing problem, or an eligibility problem. The useful part is that I can then go build the thing that stops it from happening again — and I know which of those three teams to route it to while I do.
What I'm looking for now is a role where both halves count: revenue operations, business systems, or solutions work at a company that would rather have one person who understands the domain and can ship than two who each only do one.
Preston BoutsisUnedited screen recordings from real build and training sessions — my own voice, my own hands, no slides. Every frame was scanned for client and patient data before it went up; segments that showed any were cut.
The Article Gen Reference sheet is the spine of the publishing system — category, title, keywords, meta description, rendered HTML, article URL, image media link, prompt ID, author, shortlink, webhook status, and the email subject and body each piece gets repurposed into. This walks through copying published subject and body cells into the dynamic content that feeds the email edition.
A TidyCal booking fires the trigger, and a Pabbly router splits it down two conditional routes depending on what was booked — each one setting its own filters and action steps before it reaches Gmail and the CRM. This is the pattern behind every intake flow I build: one trigger, explicit branches, no silent fallthrough.
Prompting Gemini Pro to generate a Google Apps Script, then doing the part the model can't — configuring it as a manually triggered function that posts a payload to a Boost.space custom webhook, and registering the hook on the receiving end so the scenario actually catches it.
Bonfire Dash is the white-labeled portal our clients log into — dashboard, staff, calendar, tasks, projects, CRM, content, forms, files, automations, onboarding, LMS, training videos. Here I'm generating Instacharts URLs for a client's KPIs and embedding them into a portal page through iFrame blocks.
The same article-generation automations in Taskmagic, taught in Spanish to the team members who run them. Conversational Spanish is on my résumé; this is what that actually buys an employer — I can hand a system to the people who will operate it without routing it through a translator.
Working through SEO metadata on the firm's site in the Wix editor — titles, descriptions, and structure on a page that exists to convert a practice owner into an onboarding call.
Updating images and raw HTML inside a live article in Grigora, including the featured-image alt text. The articles arrive through the automated pipeline; this is the human pass that happens after the approval gate.
Short destination URLs with aliases, UTM parameters, custom social preview cards, deep links, scheduled publication and expiry, and password protection. Every outbound campaign link is built here, which is how the attribution survives the trip.
Bulk-updating staff profile photos across the organization's Zoho Mail user directory — the unglamorous administration that makes a distributed team look like one company in a client's inbox.
Six of the two hundred, picked because they show range rather than repetition. Each diagram is the real shape of the workflow — module counts, branch widths, and approval gates as they actually run in production.
Replaced manual patient billing across the entire client book. One trigger generates the statement, sends it by email and SMS, attaches a hosted payment link, and writes the result back to that client's KPI sheet. Nobody on my team touches a patient statement anymore.
Takes a topic off a queue, generates an SEO-structured article, renders styled HTML, and publishes to the CMS through its API — behind a human approval gate, because an unreviewed article about medical billing is a liability. Each published piece is then repurposed into an email edition and syndicated to LinkedIn and Facebook. Runs unattended, twice daily.
A Chrome and Edge extension that validates patient contact data for medical practices and flags unverifiable records before they enter the pipeline. Clean records become qualified leads in the CRM with automatic campaign enrollment; the rest are held and flagged rather than quietly dialed, which is what keeps the outbound side TCPA-defensible.
A single intake form builds a client's entire nested document tree — 16 folders, correctly permissioned — shares it with the right people, spawns a KPI sheet from template, and creates the company and contact records downstream. The staff version routes new hires to the right team's workspace off a six-way branch. Onboarding used to be a checklist somebody forgot.
Moved the firm's full automation estate between platforms by exporting every scenario as JSON and transforming it programmatically — pruning platform-specific modules at any nesting depth, nulling stale connection and hook references, and stripping regenerable metadata to cut payload size 60%. Hand-rebuilding 59 scenarios would have taken weeks and introduced drift. This was reviewable, repeatable, and diffable.
CRM architecture and integration, automated quote and proposal generation, and multi-channel campaigns across email, SMS, and physical direct mail. Every outbound path sits behind a compliance gate for TCPA and A2P 10DLC — in healthcare the consequence of a sloppy campaign isn't a low open rate, it's a complaint. This engine is what drove the 220% and 310% client growth years.
Live sessions in 2023 for billers, office managers and physicians, each built around the denials that specialty was actually getting. Open Q&A throughout. Recordings available on request — attendee questions are in them, so they are not posted publicly.
Where to set charges on bracing and durable medical equipment, and how to defend the number. Walked through referencing the Medicare fee schedule, FairHealth estimates for a given zip code, and contracted payer schedules, then landed on charging 120–130% of Medicare as the working norm.
The modality codes chiropractic clinics bill every day and the ones they get wrong. Covered unit rules on timed codes, the bundling denials that follow manual therapy, and when a modifier actually fixes the claim versus when it just delays it.
What to do when a payer says the panel is closed. The out-of-network economics that make waiting survivable, the transparency obligations that come with charging OON rates, and using patient nominations to force a panel open from the demand side.
Built around the denial codes pediatric practices actually receive. Spent most of the session on modifier-inconsistency rejections — verifying payer telehealth modifier requirements, knowing when 59 or XE is required on a bundle, and checking bilateral standards before resubmitting.
The front-end work that decides whether a claim gets paid: eligibility verification, coverage checks, and prior authorizations, with the plan-specific traps that catch primary care — Medicare Advantage rules, Medicaid managed-care plans, and payer stipulations tied to a diabetes diagnosis.
A second session on the same audience, reframed around undercharging rather than denials. Supplies billed at cost or not at all, injections and medications submitted without NDCs, supervisory protocol not documented, annual wellness visits bundled into the wrong encounter.
Not a list of logos. Everything here is something I've built in, administered, or migrated between in production, under a contingency contract where a broken workflow costs the firm money the same week.
Claims Adjuster, UIG — 2016–2017. Auto insurance. I read claims against policy language and decided what was payable. Different industry, same posture as a denial review, and it is the only year of my career spent on the paying side of a claim rather than the submitting side. That is worth more to me now than it sounds.
Sales Team Lead, Ingram Medical — 2014–2016. Inside sales on diabetic supplies against a daily quota of 22. Ranked consistently in the top five for conversion across a sales organization of roughly 150 staff, and promoted to Sales Team Lead for my final six months, running a team of 17 reps.
preston@bonfirerev.com
Salt Lake City, UT — open to remote