Medical information response automation that shows where every sentence came from
Every sentence of the reply linked to its approved source, off-label parts evidence only.
See one case, screen by screen ↓Software that takes a healthcare professional's question from the inbox to a signed reply. Medical Information Response adds seven agents that read and split the question, confirm it is unsolicited, find the evidence, compose the letter with a source on every paragraph, guard any off-label part and check every sentence — for a medical information specialist to sign.
How to choose AI solutions for life sciences →
Updated
When a pharmacist or physician asks something no single standard response covers, the medical information specialist answers in a custom letter. That means finding the right standard response, the label for the requester's country, the data on file and the published papers, pasting paragraphs into the country template, then checking every claim and number by hand and routing the letter for signature. It is hours of work and days of elapsed time, while the HCP has a patient waiting.
The risk sits in reused wording. A sentence copied from an earlier letter can carry a retired standard response, or advise an unapproved use. Off-label parts carry the most risk: the answer must state the approved use first, give the evidence including what does not support the use, stay specific to what was asked and attach references and the label — one sentence of advice can make a scientific response read as promotion. And because reviewers cannot see where a sentence came from, they re-check every one.
Estimated split of the 4.6 days from receipt to a sent custom letter by hand, and of the 1.4-day target with the solution.
Seven specialist agents read, check, find, compose, guard and verify every answer; a medical information specialist signs.
Martin Kessler, PharmD, a clinical pharmacist in nephrology at Harbor Point Regional Hospital in Baltimore, emails Northwind Bio Medical Information about a heart-failure patient on Corventa who starts haemodialysis on Monday. Here is what happens next, screen by screen, in the working solution.
The patient is 68, on Corventa 10 mg once daily, with an eGFR of 22. The inquiry reader splits his email into two parts: severe renal impairment and dialysis, which the label speaks to, and a 5 mg dose given after dialysis, which it does not. No single standard response covers both, so it routes a custom letter. The request check verifies Dr. Kessler in the HCP directory and records that the question came from him directly — unsolicited, so an off-label section may be written.
Route: “Custom letter — no standard response covers every part.”
Priya Raman watches each agent light up in turn. The evidence finder draws four sources: the Corventa Prescribing Information, standard response SRD-COR-011 v3 Use in Renal Impairment, Varga 2024, a single-dose pharmacokinetic study, and Okonkwo 2025, a case series in patients on maintenance haemodialysis. The letter writer composes in the US custom letter v7 template, one section per part of the question.
“Every paragraph carries its source · click one to read the passage.”
Section 1 is tagged PI §8.6 and §12.3 and SRD-COR-011 v3: no dose adjustment down to eGFR 30, not studied below 30 or on dialysis, and Northwind Bio has no clinical trial data in dialysis. Priya clicks paragraph 4 and the Varga 2024 results open beside it, highlighted: exposure 1.3 times higher with eGFR below 30, and less than 5% of the dose removed by a 4-hour session.
Section 2 opens with the approved dose — 10 mg once daily, or 2.5 mg with a strong CYP3A4 inhibitor — and says plainly that a 5 mg dose and timing around dialysis are not in the approved labeling. Then the evidence, its limits and the hypotension warning. The writer had carried over a sentence from a draft of letter MI-23977: “Clinicians may consider reducing the dose to 5 mg once daily, given after the dialysis session.” The guard struck it.
“Advises an unapproved dose and has no published source. An off-label section may state the evidence; it may not recommend.”
The checks are on one line: 8 of 8 sentences cited, the off-label section built from scientific evidence, 1 sentence removed by the guard, 3 references attached. The signing route shows Priya first, then Omar Haddad, MD for the off-label section, then secure email to Dr. Kessler with the case closed and archived. She signs with her password.
Meaning: “I reviewed this response and approve it as accurate, balanced and specific to the question asked.”
Because the letter contains off-label information, it waits for Omar's second signature. His screen lists what he is attesting to: specific to the question asked, the disclosure, the balance, 3 references with the label enclosed, and kept apart from promotion — sent privately to the requester only. Priya's first signature, Oct 7 at 10:24, is shown beside it.
The letter goes to Dr. Kessler by secure email with its 3 references and the Prescribing Information. The case closes and the letter, its sources, the decisions and both signatures are archived. The library keeper counts 4 dialysis dosing questions this quarter and suggests an off-label standard response to Dr. Omar Haddad.
“Sent Oct 7 10:47 by secure email with 3 references and the Prescribing Information.”
Dr. Helena Marsh, a dermatologist in Portland, asked about Veltrimab in pregnancy planning, breastfeeding and dosing every 2 weeks. Her letter has 11 sentences and 1 to decide: “Northwind Bio suggests considering stopping Veltrimab 15 weeks before a planned conception.” It came from a letter that followed SRD-VLT-014 v4, retired on Jun 12, 2026; v5 removed it because the Prescribing Information sets no period for stopping before conception. The checker offers the current wording; Priya decides.
The citation checker never accepts its own finding — a person accepts or rejects each one.
Three questions have had 4 or more custom letters in 90 days and no standard response: giving Corventa through a feeding tube (6), a washout when switching from another biologic (5) and Talmirex with untreated brain metastases (4). One click drafts a standard response from the signed letters and their sources, for its owner to approve.
Drafts only — owners approve.
Dana sees 1,184 questions answered in 30 days and 236 custom letters, 20% of questions. Every one of the 2,124 sentences checked carries a source, and all 41 questions submitted by field teams were confirmed with the requester before drafting. A flow shows today's 48 questions from how they arrived — phone, web portal, email, field medical, chat — to how they were answered and where they are now.
The same letter, seen by the four people who answer for it — what their week looked like, and what it looks like now.
Seven specialist agents read, check, find, compose, guard and verify every answer; a medical information specialist signs.
Reads each new question from the portal, email, call log, field medical or chat; checks the requester is a healthcare professional; splits the question into parts and sets product, indication, topic, country and on- or off-label for each; chooses the route.
Verifies the requester in the HCP directory, records how the question arrived and who submitted it, and when a field team submitted it on the requester's behalf, confirms with the requester that the question was their own before drafting starts.
Finds the passages that answer each part: current standard responses first, then the label for the requester's country, data on file approved for external use and peer-reviewed literature — including data that do not support an off-label use.
Writes the letter from the country template, one section per part in the order asked, linking every factual sentence to the passage it came from, with the off-label disclosure and balancing data where needed; adds references and encloses the label.
Checks every off-label section in order — disclosure, evidence, limits and safety — and removes any sentence that advises or suggests an unapproved use, has no scientific source or uses promotional or comparative wording; attaches references and the label and sends the letter for a second signature.
Re-reads every sentence against its passages: is it supported, do the numbers match, is the source current, is the off-label disclosure present, is the letter balanced, are there promotional words — and offers corrected wording from current sources.
Groups signed custom letters by product and topic; when a topic has 4 or more custom letters in 90 days and no current standard response, drafts one from the signed letters and their sources; reminds owners 30 days before a review is due.
Edits and signs. The agents propose; a named person decides.
Ask in plain words, or say what to change. Every answer points to the passage behind it.
Why did the off-label guard remove a sentence?
Part 2 of Martin Kessler's question asks about a 5 mg dose given after dialysis, which is not in the label. In an off-label section the letter may state the evidence; it may not advise. The letter writer had carried over “consider reducing the dose to 5 mg” from a draft of letter MI-23977, with no published source behind it, so the guard removed it. What stays: the single-dose study, the dialysis case series and their limits, with the Prescribing Information enclosed.
Why does Dr. Marsh's letter need a second signature?
Part 3 asks about dosing every 2 weeks, which is outside the approved dosing. Your settings send every letter with off-label content to Omar Haddad, MD for a second signature. The section opens with the disclosure, gives the extension-cohort result, then the limits and the infection rates, and the Prescribing Information is enclosed.
Flag literature older than 5 years
Done. New rule: flag any cited article published more than 5 years ago, so you can check for newer data before signing. None of the open letters cites an article older than 2024. The rule is live in Settings and recorded in the audit trail.
Draft a standard response for Corventa through a feeding tube
Six custom letters in 90 days asked this, and there is no standard response. I am drafting SRD-COR-015 v0.1 from the signed letters and their sources: the label wording, the in vitro study and its limits. It goes to Omar Haddad for review.
13 screens from the working solution, on its sample data. Pick one to see it large.
Who signs, how fast, and what the agents may cite are settings, changed by your MI manager and recorded on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Custom letters signed by | Any MI specialist | Any MI specialist · MI pharmacists only |
| Second signature on letters with off-label content | Omar Haddad, MD | Omar Haddad, MD · Dr. Maya Chen |
| Standard responses sent unchanged | On — without a second look when the question matches a current standard response exactly | On · Off |
| Questions from patients | Answered by the nurse line, never as a letter to an HCP template | On · Off |
| Custom letter clock | 5 business days | Step up or down, one business day at a time |
| Standard response letter clock | 2 business days, Prescribing Information included | Step up or down, one business day at a time |
| Field-submitted request | 1 business day for the requester to confirm the question was their own | Locked |
| Congress abstracts | Off | On, with a note that the data are not peer reviewed · Off |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal for the desk, measured in the live solution · “estimated” = our estimate · Context: FDA final guidance on communications to HCPs about scientific information on unapproved uses (January 2025) and FDA revised draft guidance on addressing misinformation and responding to unsolicited requests (July 2024). People, products, companies and hospitals named on this page are fictional — characters and sample data in the working solution.
Software that takes a healthcare professional's question from the inbox to a signed reply. Medical Information Response adds seven agents that read and split the question, confirm it is unsolicited, find the evidence, compose the letter with a source on every paragraph, guard any off-label part and check every sentence — for a medical information specialist to sign.
The inquiry reader sets on- or off-label for each part of the question, not for the question as a whole. An off-label section opens with the approved use and a plain statement that the asked use is not approved, gives only scientific evidence with its limits and safety, and attaches references and the label. The off-label guard strikes any sentence that advises, and the medical reviewer adds a second signature.
Yes. Every paragraph is tagged with its source — a standard response, a label section, a publication or data on file — and one click shows the exact passage, highlighted. The citation checker re-reads every sentence against its passage for claims, numbers, version, balance and promotional words.
Not unnoticed. Only current versions may be cited, and the citation checker flags a sentence whose wording follows a retired version, with the current wording to use instead. The specialist accepts the change or keeps the sentence with a note the medical reviewer sees.
The request check keeps the doctor's own words and records who submitted the question. Before any off-label section is written, it confirms with the doctor that the question was their own, and MI compliance reviews each confirmation. A question prompted by a sales representative never receives an off-label section.
Yes. The agents draft and never sign. A medical information specialist reviews and signs every letter, the medical reviewer adds a second signature to letters with off-label content, every checker finding is accepted or rejected by a person, and a sentence the guard removed returns only with the medical reviewer's signature.
Yes. Letter templates per country, who signs custom letters, the second signer for off-label content, the clocks for custom and standard response letters, and what the agents may cite — such as congress abstracts — are settings. Rules can also be added in plain words, for example flagging literature older than 5 years.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your SOPs, templates, standard response library and a sample of past inquiries — and it goes live once every quality gate has passed. We will walk you through it on your own inquiries first.
Every claim checked against its reference before committee.
1-min demoInvestigator-Initiated Study ReviewProposals decided in half the time with a cited review packet.
1-min demoMedical Insights ManagementEvery field insight becomes a cited theme within days, with an owner for the next step.
1-min demoWe'll run Medical Information Response on a sample of your own past inquiries and standard responses.