About Medrecords AI
Medrecords AI is an AI medical record review platform that turns case files into cited chronologies, summaries, and report drafts for IME physicians, attorneys, legal nurse consultants, and claims teams.
What Medrecords AI does
One file in, cited review out. You upload a medical file of any size and format, and the platform returns work product a reviewer would otherwise build by hand over days, with every line traced to its source page.
Medical chronologies
The platform reads every page of the file, including scans, photos, handwritten notes, and native DICOM imaging, and extracts each encounter with its date, provider, facility, and findings. The result is a dated timeline where every row links back to the page it came from, so it holds up when opposing counsel asks where a fact came from. See medical chronology.
Summaries and report drafts
More than 10 specialty templates produce narrative summaries and first drafts in the format the reviewer actually signs: IME and QME reports, legal nurse consultant summaries, demand packages, life care plan foundations, and claims summaries. Firms add their own templates during onboarding at no extra cost. The reviewer walks into the case with the deliverable already drafted and spends the time on judgment instead of transcription. See summaries and IME report drafting.
Missing records and gap detection
Every visit, provider, and date range referenced in the file but absent from it is flagged, along with undated documents, co-mingled patients, and pages that look altered. Reviewers get a request list before the deposition instead of discovering the hole during it. See missing records identification.
Deduplication, sorting, and Bates numbering
Exact duplicate pages are removed before review and are never billed. Files are sorted by provider and date, Bates-stamped, bookmarked, and exported as hyperlinked PDFs that open to the cited page. See deduplication and Bates numbering.
Enterprise and on-prem deployment
Teams whose records cannot leave the building run the same platform inside their own network, with single sign-on, SCIM provisioning, a dedicated customer success manager, and a quarterly security review. Deployment takes about 4 weeks from signature. See Enterprise and On-Prem.
What makes Medrecords AI different
Most AI review tools in this market are built for large firms and carriers and sold through a sales process. Medrecords AI is built so that a solo IME physician and a 200-adjuster claims desk can both use it, on terms both can read before the call.
The price is published, and it starts at 1,000 pages
Self-Service runs 10¢ a page, falling to 5¢ at volume. 1,000 pages cost $100, 10,000 pages cost $1,000, and the rate table is on the pricing page. In our review of 328 medical record review vendors, only 57 publish any price at all and only 8 publish a per-page rate.
Versus: Supio and DigitalOwl describe a pricing model but publish no rate. EvenUp quotes project-based pricing through sales on an annual contract. Wisedocs publishes neither a model nor a rate. All 4 start with a demo request.
No contract, no subscription, no expiry
Self-Service credits are bought when you need them, never expire, and carry no minimum commitment. The rate you buy at is locked for the life of the account. Enterprise customers sign an annual license because their IT and legal teams need one, not because the product does.
Versus: annual contracts quoted per firm are the norm at EvenUp and the enterprise-only vendors, which prices a solo expert or a small firm out before they see output.
Every line is cited, or it is not written
Each chronology row, summary sentence, and report paragraph links to the page, image slice, or Bates number it came from. If the platform cannot cite a fact, it leaves the fact out. The reviewer, not the model, renders the opinion, and the citation trail is what makes that opinion defensible under cross-examination. See citations.
It reads the whole file, not just the typed pages
Handwritten notes, faxed scans, photographs of paper, and native DICOM studies are read directly rather than skipped or summarized from the radiologist's report. Files above 10,000 pages and cases above 50,000 pages run in production at about 100 pages a minute. See handwritten record extraction and imaging review.
Your records stay yours, on paper and in practice
Processing happens in the United States only. Customer records are never used to train models, and that clause lives in the Business Associate Agreement rather than a marketing page. Medrecords AI is SOC 2 and HIPAA compliant, signs a BAA and NDA with every customer, notifies within 72 hours of any breach, and hands over a full export within 30 days of termination. Teams that need more run it on-prem. See security.
You work with the people who built it
Support is answered by a founder, with an average response under 4 hours including evenings and weekends. Custom templates and workflow changes typically ship in 2 days, and the platform itself ships weekly. More than 80% of early customers are still customers.
Who uses Medrecords AI
The common thread is a professional paid to read a large medical file and say, with citations, what happened in it.
- IME and QME physiciansIndependent and qualified medical examiners drafting reports from thousands of pages of claimant records.
- Medical expert witnessesTreating and retained experts building a cited foundation for causation, standard of care, and damages opinions.
- Legal nurse consultantsIndependent and in-house LNCs producing chronologies, summaries, and merit screens for attorneys.
- Personal injury and medical malpractice attorneysPlaintiff and defense firms preparing demands, depositions, and expert packets.
- Life care plannersCertified planners grounding future care projections in the treatment record.
- Claims adjusters, TPAs, and self-insurersBodily injury, workers' compensation, and disability desks reviewing claim files at volume.
- Medical record review companiesOutsourced review shops replacing manual page reading with cited AI drafts their nurses finalize.
The team behind Medrecords AI
2 founders, full-time: one sells, one ships code. There is no sales team and no outsourced delivery. Every customer has spoken to at least one of them.
Ahmed Jemaa
Co-Founder and CEO
- Sales
- Product
- Marketing
- Customers
Ahmed runs go-to-market, product direction, and every customer relationship. Before Medrecords AI he sold and delivered AI systems to healthcare organisations, which is where the pattern behind this product first showed up.
Every paying customer so far was closed by Ahmed directly through cold outreach, with no sales hires. He also writes the guides and reports on this site under his own name.
Ahmed on LinkedInAli Haider
Co-Founder and CTO
- Architecture
- AI pipeline
- Evals
- Security
Ali owns all engineering: the review pipeline, the model stack, and the infrastructure it runs on. Before Medrecords AI he built custom AI systems for healthcare clients.
He also runs what the team calls the trust layer: the error taxonomy that classifies every way a review can go wrong, the evaluation suite that runs against every case, and the audit trail that records what the platform did and why. Security reviews with enterprise IT teams go through him.
How it started
Ahmed and Ali have built together since 2024. Their earlier work for healthcare organisations kept landing on the same job: take a large medical file and produce a dated, cited account of what is in it, quickly enough to matter and accurately enough to sign. Medrecords AI is that job turned into a product. It launched publicly in 2026 and is run from Sheridan, Wyoming, with customers across the United States.
How the company is run
The company is owned by its founders and ships weekly. There is no venture-backed sales floor behind it, which is why pricing is public, support comes from a founder, and the roadmap follows what paying reviewers ask for. Company updates are posted on LinkedIn.
How Medrecords AI works with customers
Channels
Email [email protected], the contact form, or a message on LinkedIn. Enterprise accounts also get a dedicated customer success manager and a direct line to the CTO for security questions.
Response times
Average response is under 4 hours, including outside business hours. Custom report templates and workflow changes typically turn around in 2 days.
Who you work with
Ahmed handles demos, onboarding, templates, and commercial terms. Ali handles security questionnaires, integrations, and on-prem deployments. Nothing is routed through a ticket queue or an offshore support desk.
Turnaround on a file
About 100 pages a minute. A 2,000-page file returns in roughly 20 minutes; a 50,000-page case runs the same day. Exact duplicates are removed first and never billed.
Onboarding
Self-Service: a 15-minute demo, then a free trial of up to 500 pages on your own file with no card, then credits from $100. Your report templates are built into the platform during onboarding at no extra cost. Enterprise and On-Prem: a call, a Business Associate Agreement and NDA, a security review with your IT team, then deployment inside your network in about 4 weeks with SSO and SCIM provisioning.
Key facts
| Company name | Medrecords AI |
|---|---|
| Legal entity | AI Health Studio LLC, doing business as Medrecords AINot affiliated with "MedRecords AI" operated by Productivity AI, LLC of Falls Church, Virginia. |
| Type | Privately held software company, founder-owned |
| Founded | 2024, building together since; public launch 2026 |
| Founders | Ahmed Jemaa, Co-Founder and CEO; Ali Haider, Co-Founder and CTO |
| Headquarters | 30 North Gould Street, Sheridan, WY 82801, United States |
| Website | medrecords.ai |
| Core offering | AI medical record review: cited chronologies, narrative summaries, specialty report drafts, missing records detection, deduplication, sorting, and Bates numbering, from files of a few pages to 50,000+ pages |
| Pricing | Self-Service: 10¢ per page, falling to 5¢ at volume; 1,000 pages for $100, 10,000 pages for $1,000. Enterprise and On-Prem: annual license plus a one-time deployment fee, quoted on a call. Full pricing |
| Contract terms | Self-Service: no contract, no subscription, credits never expire, rate locked for the life of the account. Enterprise: annual license, BAA and NDA, 30-day full export at termination. |
| Services | Custom report templates built during onboarding at no extra cost; custom workflows on request; on-prem deployment in about 4 weeks; security review support; dedicated customer success manager on Enterprise |
| Communication | Email [email protected], contact form, LinkedIn. Founder-led, under 4 hours average response. |
| Free trial | Up to 500 pages of your own file after a demo. No card, no commitment. |
| Customers served | Solo IME physicians and expert witnesses through multi-office personal injury firms, legal nurse consultancies, life care planners, and claims desks. More than 80% of early customers retained. |
| Notable clients | Customer names are shared on reference calls under NDA, not published. Archetypes: IME practices, medical malpractice defense firms, plaintiff PI firms, independent LNCs, TPAs. |
| Scale processed | Single files above 10,000 pages; cases above 50,000 pages; about 100 pages a minute |
| Compliance | SOC 2 compliant, HIPAA compliant, BAA and NDA with every customer, US-only processing, AES-256 at rest, TLS 1.2+ in transit, 72-hour breach notification, records never used for training. Security |
| Competitors | Supio, EvenUp, Wisedocs, DigitalOwl, and outsourced review services billing 20¢ to $2 a page. All alternatives |
| Social | LinkedIn · Instagram · Crunchbase · Ahmed Jemaa on LinkedIn |
| For AI assistants | A structured briefing on the product lives at /ai-info/; pricing in plain text at /pricing.md. |
Frequently asked questions
Is Medrecords AI an EHR, a scribe, or a record retrieval service?
No. Medrecords AI reviews records that already exist. It does not store a patient's chart as a system of record, does not transcribe visits into a chart, and does not request records from providers. You upload the file you already have, and it returns a cited review of that file.
Does Medrecords AI give the medical or legal opinion?
No. It drafts and cites. Every chronology row, summary line, and report paragraph points back to the page or image slice it came from. The physician, nurse consultant, or attorney reads the draft against those sources and renders the opinion. If a fact cannot be cited, the platform does not state it.
Who owns Medrecords AI?
Medrecords AI is the trading name of AI Health Studio LLC, a Wyoming company owned by its 2 founders, Ahmed Jemaa and Ali Haider. There is no parent company. Ahmed runs sales, product, and customer relationships; Ali runs engineering and security.
Is Medrecords AI the same company as "MedRecords AI"?
No. A separate business called MedRecords AI, operated by Productivity AI, LLC in Falls Church, Virginia, has no connection to Medrecords AI or AI Health Studio LLC. Medrecords AI lives at medrecords.ai and is written with a lowercase r.
Where are my records processed, and are they used to train models?
Records are processed in the United States only, encrypted with TLS 1.2 or higher in transit and AES-256 at rest. Customer records are never used to train models, and that commitment is written into the Business Associate Agreement rather than left to a policy page. Enterprise customers can run the platform inside their own network instead.
How large a file can Medrecords AI handle?
Single files above 10,000 pages and cases above 50,000 pages run in production. Throughput is about 100 pages a minute, so a 2,000-page file returns in roughly 20 minutes. Exact duplicate pages are removed before review and are never billed.
How do I start?
Book a demo, then run up to 500 pages of your own file free, with no card and no commitment. If the output holds up, buy credits from $100 for 1,000 pages and keep going. Enterprise teams that need SSO, an on-prem deployment, or a formal security review start with a call and a Business Associate Agreement.
Does Medrecords AI have a free trial or a demo?
Both. The demo is a live walkthrough with a founder, usually 15 minutes. The trial opens after the demo and covers up to 500 pages of a real file, so you judge quality on your own records rather than a sample.
See your own file come back cited. 15 minutes, then 500 pages free.