AI sales reps for regulated medicine
Reach every HCP with a rep that can't go off-script.
Every word a zenrep sends on web and Telegram is checked against your MLR-approved corpus and paired with fair balance before it reaches the HCP — and every conversation, on every channel, is logged for your compliance team. It extends how far your approved message travels, without adding another territory, vehicle, or headcount.
HCP · inbound10:42:03
“What's the once-daily dose of Veltarin, and does food matter?”
Compliance gate · pre-send4 checks
Verdict — approved for delivery
Zenrep · delivered10:42:05
“Veltarin is 40 mg once daily, with or without food. The most common adverse reactions are headache and mild nausea, and it's contraindicated in known hypersensitivity — see the full prescribing information. Want me to send the approved labeling to your inbox?”
Veltarin is fictional, shown for illustration only.
The territory math changed. The discipline didn't.
As of 2016, only 44% of US physicians were rep-accessible, down from roughly 80% in 2008.1 The rooms a field force can reach keep narrowing, and every sentence spoken inside them is a regulated statement. Most teams respond by narrowing further: fewer touches, tighter scripts, quieter territories. A zenrep works the territory the other way. Every HCP, including the rural and low-volume clinicians a field force never gets to, around the clock — with the discipline built into how it speaks rather than reviewed after the fact.
1. ZS AccessMonitor 2016, via PR Newswire — prnewswire.com/news-releases/as-doctors-keep-closing-doors-o…
Reach more, risk the same.
VP / CCO / Brand. Cover the whole territory without growing the org chart.
Book a demoThe risk is designed out, not reviewed after the fact.
CMO / Reg Affairs / Reviewers. Never off-label. Fair balance and ISI every time. Refuses rather than guesses.
How the gate decidesDefensible by design.
CISO / Legal / Procurement. Residency matched to jurisdiction, per-tenant isolation, an append-only audit trail.
Read the trust postureGate verification metrics publish here as they accumulate.
Every answer arrives with a receipt.
On web and Telegram, live today, every reply passes the compliance gate before a single word reaches the clinician. Below: one question, one answer, and the record the gate writes while it works. The video avatar is a preview today — audited after the fact, with the fully gated version on the near-term path.
HCP asks
“What's the once-daily dose of Veltarin, and does food matter?”
Zenrep answers · after the gate
“Veltarin is 40 mg once daily, with or without food. The most common adverse reactions are headache and mild nausea, and it's contraindicated in patients with known hypersensitivity to the active ingredient — see the full prescribing information for complete warnings. Want me to send the approved labeling to your inbox?”
Fair balance paired automatically; every passage traces to the approved corpus. Veltarin is fictional, shown for illustration only.
Live gated web chat against the fictional Veltarin corpus — every reply runs the same pre-send gate and writes the same audit record.
1 / 4
It can only speak approved material.
There's no general medical model improvising underneath. A zenrep cannot assert anything that isn't in your medically and legally approved source material — that corpus is the ceiling on what it can say.
2 / 4
Every claim is checked before it's delivered.
Each turn passes the pipeline: in scope, fair-balanced, grounded in an approved passage, and citable. Then, and only then, is it spoken.
3 / 4
It refuses rather than guesses.
When a claim can't be grounded in approved material, the gate is fail-closed: the zenrep declines, redirects, or routes the question to a human. Refusing correctly is treated as a success, not a failure.
4 / 4
Every turn is audit-traceable.
Each exchange records what was asked, what was answered, what the gate decided, and the exact approved passage it was anchored to — reconstructable for your compliance team any time, and exportable as CSV.
Don't take our word for it. Text a rep.
Sarah is a zenrep detailing VELTARIN — a fictional product built to show the gate. Ask a real dosing or safety question and watch the reply get assembled from the approved corpus, checked claim by claim, and written to the audit record — turn by turn. Nothing scripted.
Sarah · zenrep
Compliance-gated
✓Verified against the approved corpus · logged
The work a zenrep takes off your team.
The whole territory.
Every HCP on the map, including the clinicians no route plan reaches. Coverage stops being a function of headcount.
Around the clock.
HCPs ask questions at 9 p.m., between patients, on weekends. A zenrep answers the moment they ask — on-label, or with a clean refusal routed to your team.
A continuous record.
Per-HCP continuity: a zenrep persists each clinician's thread so the record stays continuous and your team can pick it back up.
Compounding per-HCP memory that carries prior context forward into new conversations is on the roadmap — not present today.
The same answer, every time.
A human can drift from approved messaging under pressure. A zenrep's reply is assembled from your approved corpus, and on web and Telegram it is verified claim by claim before it sends. Consistency is a property of the system, not of training.
Also in the platform: two-layer adverse-event detection with pharmacovigilance routing, escalation paging to your team, and jurisdiction and product-class rulesets (PAAB / Health Canada and FDA OPDP, drug and device) enforced on every turn. Sample-request intake is on the roadmap. The full platform
The whole job.
A field rep's job runs longer than a call list. Below is the whole of it, and where Zenreps stands on each line. The aim, stated plainly: the reach of many reps with the touch of one — on the clock evenings and weekends, keeping a verbatim record, carrying no commission incentive. The status marks are literal. We would rather you read "roadmap" here than discover it in a contract.
Detail the product, in writing.
Gated text detailing on web chat and Telegram. Never off-script: every claim comes from your approved material and is checked before it is sent. Every turn is written to an append-only audit record as it happens.
A human rep holds one conversation at a time. A zenrep is in every conversation it is asked into, simultaneously.
Detail the product, face to face.
Video detailing by avatar — demoable today, with every word audited after the fact. The fully gated version, checked before it speaks, is the near-term path.
A visit with no drive between offices. The calendar, not the highway, sets the limit.
Take the phone call.
Voice over phone. A gate-enforced voice pipeline is in development; it is not generally available.
Phones do not close at five. Neither will this.
Work the inbox.
Email and SMS as detailing channels, behind the same gate as everything above.
An 11 p.m. question can get an 11 p.m. answer. Evenings and weekends are working hours by construction.
Open the conversation.
Proactive outreach: identifying the clinicians who should hear about the product, and reaching out first.
A territory is usually a limit of hours and highway. Here, it would be as long as the list.
Book the appointment.
Scheduling with office staff, reminders, and the inevitable reschedules.
Bookable whenever the office has the calendar open — before clinic and after close — once this ships.
Order lunch, and host it.
Lunch-and-learns: catering ordered, meeting hosted, product detailed inside the gate — for several offices in the same noon hour.
A human rep gets one noon a day. Nothing in this design is limited to one office per noon.
Order the samples.
Sample requests placed in the conversation where they are asked for.
Designed to file at the moment of the ask, not from the car afterward.
Write it up.
Structured post-call notes and CRM write-back. The raw material is live today: every conversation is already recorded verbatim, turn by turn.
The note derives from a verbatim record, not from memory in a parking lot.
Remember.
Conversation persistence is live: no prior exchange is forgotten. Compounding per-clinician memory — each conversation informed by every one before it — is roadmap.
Every contraindication, study endpoint, and technical specification lives in the approved corpus it answers from — and on the gated text channels, every reply is verified against that material before send.
One rule travels with every line above. Each new channel ships behind the same compliance gate that governs the first one. And when hospitality ships, it ships logged as a reportable transfer of value by design — the machine that orders the lunch will be the machine that files the disclosure.
That is the whole job. We publish where we stand on every line of it, because a pharma buyer should be able to audit the ambition.
The arithmetic of a field force.
The left column is published research on what one US field rep costs and can do, shown with each figure's year and source. The right column describes what a zenrep is structurally — properties of software, not measured performance. A zenrep is a coworker, not a substitute: this table is about the work software should carry so your reps can carry the relationships.
| Dimension | One human field rep | One zenrep |
|---|---|---|
| Fully loaded annual cost | Industry research puts the average yearly cost per rep at ~$260,000, up to $450,000 depending on geography, therapeutic area, and physician type.1 | Software. No salary, bonus, benefits, fleet, or T&E line items. structural |
| Hours of availability | Bounded by office hours and windshield time in one territory; a fleet car alone runs about $14,000 a year.7 | 24/7, including the after-clinic hours when HCPs actually have time. structural |
| Concurrent conversations | One at a time. The benchmark inside rep makes 46 dials for 5.8 quality conversations a day.3 | Not bounded by headcount — capacity scales with demand. structural |
| Getting through the door | As of 2016, 44% of US physicians were rep-accessible, down from roughly 80% in 2008; 18% were severely restricted.2 | Pull-based: the HCP opens a link on their own schedule. No front desk to get past. structural |
| Cost per HCP interaction | Rep pay alone works out to $20–57 per call — before the vehicle, samples, and management overhead.4 | The marginal cost of one more conversation is compute — a fraction of a field call, with no vehicle, samples, or management overhead. structural |
| Ramp time | 3.3 months average for inside reps3; up to three months for pharma hires to get up to speed1; 5.42 months just to fill a vacant field role.5 | Zero ramp once the approved corpus is loaded; the "new rep" is a copy of the trained one. structural |
| Turnover | Roughly 26% annual sales-force turnover; $97,690 average cost to replace.5 Gallup puts replacement at 0.5–2x annual salary.6 | Doesn't resign, doesn't take the territory to a competitor, no backfill cost. structural |
| Record-keeping | Call notes are self-reported after the fact; completeness depends on rep discipline at the end of a driving day. structural observation, unsourced | Every conversation logged verbatim, automatically, the moment it happens — into an append-only audit trail. structural |
| Message consistency | A human can drift from approved messaging under pressure; consistency depends on training and monitoring. structural observation, unsourced | Cannot draw on a general model: replies are assembled from your approved corpus and, on web and Telegram, verified claim by claim before send. structural |
| Scaling the team | Months per hire (3.69–5.42 months to fill5), plus recruiting, onboarding, and fleet provisioning; downsizing means severance and lost relationships. | Scales up or down instantly. No hiring cycle, no fleet, no severance. structural |
1. Arx Research — Compensation for Pharma Sales Representatives, District and Regional Sales Managers (published 2017; 67 respondents at 29 US life-science companies): arxresearch.com/compensation-for-pharma-sales-reps-dms-rms
2. ZS AccessMonitor 2016, via PR Newswire — As Doctors Keep Closing Doors on Pharma Reps: prnewswire.com/news-releases/as-doctors-keep-closing-doors-o…
3. The Bridge Group, 2016 Sales Development Metrics and Compensation Benchmark Report, via For Entrepreneurs (David Skok) — sales-industry SDR benchmark, not pharma-specific: forentrepreneurs.com/bridge-group-sales-development
4. MedCepts — Value of Pharmaceutical Rep Per Physician Visit; rep compensation only, not fully loaded: medcepts.com/pages/value-of-independent-pharmaceutical-reps
5. DePaul University Center for Sales Leadership — 2015–2016 Sales Effectiveness & Sales Acceleration Survey; general B2B sales, not pharma-specific: cdn2.hubspot.net/hubfs/3319111/ConnectLeader_June2017/PDFs/C…
6. Gallup — This Fixable Problem Costs U.S. Businesses $1 Trillion (2019): gallup.com/workplace/247391/fixable-problem-costs-businesses…
7. Cardata — Fleet Vehicles: What Do They Really Cost? (vendor worked example, Jeep Cherokee fleet program): cardata.co/blog/fleet-vehicles-real-cost
Source notes
Figures are shown with their survey years and are not inflation-adjusted; several predate 2020. DePaul, Gallup, and Bridge Group figures are general sales-industry benchmarks, not pharma-specific. The ZS access data shown is from 2016, the most recent year we verified against a primary public source; access is widely reported to have tightened since. The MedCepts per-call figure is rep pay only. Rows marked structural are design properties of software, not measured performance.
Gate verification metrics publish here as they accumulate.
Two regulated verticals, one discipline.
Drug promotion, PAAB- and OPDP-bound.
Reach every prescriber on-label, at launch scale, without growing the field force. Off-label refusal, fair balance and ISI, adverse-event capture — enforced on every turn on the gated text channels (web and Telegram).
Device promotion, FDA- and Health Canada-bound.
Consistent, approved product answers across the install base: IFU-bounded, no off-label use claims, malfunction and adverse-event routing, a full audit trail.
What a zenrep will not do.
Hard limits, engineered in — behavior the gate is built to refuse, not policies it tries to follow.
- Make a claim that isn't in your approved corpus
- Send efficacy without the required safety information
- Guess when a question goes off-corpus — it refuses
- Speak off-label to sound helpful
- Talk to patients — HCP-facing only, by design
HCP · inbound
“Can I use Veltarin for an off-label indication?”
Zenrep · declined & routed
“I can only speak to Veltarin's approved indication and labeling, so I can't advise on off-label use. I can route this to a member of the medical team — would that help?”
Fail-closed
Veltarin is fictional, shown for illustration only.
Isolated by tenant.
Row-level security, per-tenant secrets, per-tenant corpus composition. No customer's corpus or audit log crosses into another's.
Residency that fits the law.
Canadian tenants run in ca-central-1 (Montreal) today; a US region stands up in-region when a US tenant onboards. Data residency by layer is detailed on the trust page.
Append-only application record.
Append-only audit trail with database-enforced edit and deletion restrictions. Reconstruct any conversation; export per-turn audit evidence as CSV.
Posture, on the record.
Every reply is grounded against your MLR-approved corpus and written to an append-only, per-turn audit record with database-enforced edit and deletion restrictions. Data residency by layer, subprocessors, and certification status are detailed on the trust page — rounded down.
The full security posture, including subprocessors
Procurement questions, answered straight.
Where does our data live?
Canadian tenants run in ca-central-1 (Montreal) today; a US region stands up in-region when a US tenant onboards. Row-level security isolates every tenant.
Do you support the EU and UK?
Not live today. The regulatory research is complete and provisional rulesets are drafted; counsel review and productization remain before any EU or UK tenant goes live. Canada (PAAB) and the US (OPDP) are live, and EU is a near-term priority.
Does it integrate with our CRM?
On the roadmap. Today every interaction lands in your audit trail and admin console, exportable as CSV.
Can it talk to patients?
No. HCP-facing only, by design. Patient-facing support is on the roadmap.
Are you PAAB pre-cleared?
PAAB pre-clears materials, not live conversations. Your corpus is the pre-cleared material; we describe our posture as PAAB-aligned.
SOC 2?
On our roadmap: controls mapped today, formal Type I scoping begins at first deployment. No audit is engaged today; status is published on the trust page as it progresses.
Inspect the mechanism.
Zenreps doesn't ask you to trust a logo wall. What you can inspect on this page is the mechanism itself: watch the gate write the receipt above, then request a design-partner demo and read a real one.
How it's built
- B.01Fail-closed in production: a reply with one ungroundable claim is refused whole, not trimmed.
- B.02Structural refusals (pediatric-on-adult, cross-tenant probes) fire before any model runs; off-label claims are refused at the pre-send verifier on the gated text channels.
- B.03Append-only audit log with database-level tamper protection; CSV export with injection neutralization.
- B.04Jurisdiction and product-class rulesets enforced per turn; a product the system cannot classify is refused rather than mis-classed.
- B.05Onboarding is "here is our MLR-approved corpus" — load it. Zero code changes.
Turn the reach loose, not the risk.
A demo runs against a corpus that looks like yours. Bring your MLR team and your hardest off-label probe.
Not ready to book?
Deployments are scoped one at a time. If you're EU-based, say so — EU is a near-term priority. We'll only email you about getting started.