Paper quality systems put pharma licences at risk: digital SOPs, CAPA and audit trails
By Vishal Kumar, Founder and CEO · Last updated · 8 min read

In short: Pharma inspectors don't just check that you did the right thing. They check that you can prove it, with records nobody could have changed afterwards. Paper logbooks and spreadsheets make that hard. A digital quality management system (QMS) captures each deviation on a phone at the point of work, runs the corrective and preventive action (CAPA) process against approved standard operating procedures (SOPs), and keeps a tamper-evident audit trail. The US Food and Drug Administration (FDA) sets rules for electronic records in 21 CFR Part 11, and India's revised Schedule M now applies to all drug manufacturers there.
The founder told me he couldn't sleep.
He runs a pharma manufacturer. His company had lost a key regulatory approval the year before. Part of the business stopped overnight.
He wasn't worried about the science. He was worried about Monday morning, and an inspector at the gate.
What did his quality system look like? Paper logbooks. Printouts. Spreadsheets. On inspection day, some of the paper simply couldn't be found.
"Right now, they do all of this manually, and they take printouts. And on the day of the inspection, they threw the paper somewhere, and they're like, I don't know where it is."
His request was simple to say and hard to do. He wanted a system he could prove had not been tampered with in any way.
This article is what we built in response, and the rules that shaped it.
Why do paper quality systems fail inspections?
Because a regulator can't trust what it can't verify.
The FDA's guidance on data integrity says that complete, consistent and accurate data "should be attributable, legible, contemporaneously recorded, original or a true copy, and accurate", which the industry shortens to ALCOA. [1] The FDA says it wrote that guidance "in response to an increase in findings of data integrity lapses in recent inspections." [2]
Paper struggles with every letter of that:
- Attributable: a signature on a printout is easy to copy.
- Contemporaneous: people fill in logs at the end of a shift, or the end of the week.
- Original: a spreadsheet can be edited and saved with no trace of what it said before.
- Findable: a logbook in the wrong drawer may as well not exist.
When the records look shaky, inspectors assume the process is too. The FDA publishes its warning letters, and data integrity problems are a recurring theme in them. [3]
The founder put the regulator's view to me more bluntly: we don't trust your paper systems, and that Excel file looks very suspicious.
| Data integrity test | Paper | Captured at the point of work |
|---|---|---|
| Attributable | a signature on a printout is easy to copy | user and time on every entry |
| Contemporaneous | filled in at the end of a shift | captured at the moment of work |
| Original | a spreadsheet can be edited with no trace | append-only; changes stay visible |
| Findable | a logbook in the wrong drawer | searchable from any screen |
What does US FDA 21 CFR Part 11 require?
21 CFR Part 11 is the FDA rule for electronic records and electronic signatures. If you keep regulated records electronically, it applies.
For closed systems, section 11.10 lists the controls expected. Three matter most for a QMS: [4]
- Validation (11.10(a)). The system must be validated "to ensure accuracy, reliability, consistent intended performance, and the ability to discern invalid or altered records."
- Audit trails (11.10(e)). "Secure, computer-generated, time-stamped audit trails" must independently record when operators create, modify or delete records. Record changes must not obscure previously recorded information, and the trail must be kept as long as the record and be available to the FDA.
- Accountability (11.10(j)). Written policies must hold people accountable for actions taken under their electronic signatures.
In plain terms: the system must work as designed, every change must leave a footprint, and every signature must belong to someone.
What changed with India's revised Schedule M?
Schedule M sets good manufacturing practice (GMP) requirements under India's Drugs Rules, 1945.
The Government of India revised it by notification G.S.R. 922(E), dated 28 December 2023. According to the Press Information Bureau, the revised Schedule M became effective for manufacturers with turnover above ₹250 crore from 29 June 2024, and for those below ₹250 crore from 1 January 2026. [5] A crore is ten million rupees.
So from this year, every drug manufacturer in India is expected to meet it, not just the large ones.
Legal commentary on the revision describes a new pharmaceutical quality system, with corrective or preventive actions taken under it, formal change control, and detailed requirements for validating GMP-related computerised systems. [6]
India's regulator is the Central Drugs Standard Control Organisation (CDSCO). Its inspections, together with state drug controllers, decide whether a site keeps its licence.
- 29 June 2024
- revised Schedule M effective for manufacturers with turnover above ₹250 crore
- Source: PIB [5]
- 1 January 2026
- effective for manufacturers below ₹250 crore
- Source: PIB [5]
What does a digital deviation and CAPA flow look like?
Here is the flow we designed, from a breakdown on the shop floor to a closed CAPA. Each step happens on a phone, a tablet or a desk screen:
- An operator sees a fault on the line, takes a photo and reports it from their phone.
- The app shows the immediate-action checklist from the approved SOP. The operator ticks each step.
- The system opens an event and alerts engineering and safety.
- The facility manager reviews and signs off the immediate actions.
- The environment, health and safety (EHS) officer inspects, with photos as evidence.
- A deviation is opened and an impact assessment recorded.
- A root cause analysis report is completed.
- AI drafts a corrective and preventive action, using the SOP and what was fixed.
- Reviewers check the CAPA, then quality assurance (QA), then the pharma expert approves it.
- If the SOP changes, the new version is approved and the difference from the old one is recorded.
- Affected staff complete training on the new SOP in the app, and the system records the date.
- Later, the CAPA is checked: did it actually stop the problem?
That training step matters more than it looks:
"This specifically came from a compliance perspective: they have an audit trail of every change in the process, and after any CAPA, there is a training that is done, confirming that person is trained on that day so that they can do the job."
Supervisors and managers see the same events on dashboards. And there is an AI assistant inside the system that people can simply ask, "What's happening? What do I do next?"
| Step | What happens |
|---|---|
| 1. Deviation reported | photo from the operator’s phone |
| 2. Investigation | impact assessment and root cause |
| 3. Corrective action | AI drafts it from the SOP; people approve |
| 4. Preventive action | new SOP version, staff trained in the app |
| 5. Verification | did it actually stop the problem? |
What should AI do, and what must stay human?
The AI drafts. People decide.
"Ultimately, a human has to sign every single thing. But AI can take what used to be an hour of thinking about whether I should do it or not, and now it is there. You just click approve if you're happy with it."
Three rules keep that safe:
- The AI reads approved SOPs. It doesn't write them. Immediate actions come from SOPs that QA has already approved, and the screen shows which section each step came from.
- Any new check needs human approval before use. If the AI suggests a new step, it waits for sign-off.
- Every draft is visibly a draft. The person approving can edit it, and the audit trail records both versions.
The side effect surprised the client. When people know each step is tracked, they take it more seriously. As I said to an investor last week: now that things are tracked, people are more sincere about it, instead of assuming a piece of paper will do.
What makes an audit trail tamper-evident?
Part 11 asks that changes never hide what was there before. We take that literally.
- Append-only records. Nothing is overwritten. A correction is a new entry that points to the old one.
- A ledger that can't be quietly edited. Our design writes each event to a blockchain-backed log, so changing history would be detectable.
- Version history for SOPs. You can see the exact difference between the previous SOP and the new one, and who approved it.
- Evidence attached at the source. Photos and checklists are captured on the device, with time and user, at the moment of work.
A validated system, proper access control and documented procedures are still needed on top. Software alone doesn't make a site compliant.
| Entry | Who | What | Status |
|---|---|---|---|
| #1 | Operator | Fault reported, photo attached | Current |
| #2 | Facility manager | Immediate actions signed off | Current |
| #3 | AI draft | CAPA drafted from the approved SOP | Superseded by #4, still visible |
| #4 | QA | CAPA wording corrected and approved | Correction of #3 |
The tempting choice and the better choice
scan the paper logbooks into PDFs and call it digital. It's quick. But the record is still written after the fact, still hard to attribute, and still impossible to search.
capture the record at the point of work. A photo, a tick and a signature on a phone beat a perfect logbook entry written four hours later.
This isn't only for pharma. Any site that runs on SOPs and CAPA, such as food or device manufacturing, has the same problem. Our manufacturing page covers the wider picture, and our article on why compliance belongs to agents looks at the common pattern across industries.
I'll be honest about one thing. Software can't promise anyone a clean inspection. When I describe this work, I say it the way I see it:
"What's the clear answer for it? I'm not 100% sure, but it's got to be better than spreadsheets."
Your next step
If an inspector arrived tomorrow and asked for the full history of your last deviation, from first report to closed CAPA and training, how long would it take you to show them?
If the answer is longer than five minutes, start there. Get your AI roadmap.
Frequently asked questions
What is CAPA in pharma?
CAPA stands for corrective and preventive action. A corrective action fixes the cause of a problem that has happened. A preventive action stops a similar problem happening again. Each CAPA is documented, approved and later checked for effectiveness.
Does 21 CFR Part 11 apply to Indian manufacturers?
It applies to electronic records and signatures that FDA regulations require, so a manufacturer supplying the US market that keeps those records electronically needs to meet it. Get regulatory advice for your specific products and markets.
When did revised Schedule M take effect?
According to the Government of India, from 29 June 2024 for manufacturers with turnover above ₹250 crore, and from 1 January 2026 for manufacturers below ₹250 crore.
Can AI write SOPs or approve CAPAs?
In our design, no. AI drafts corrective actions using approved SOPs and the facts of the event. Qualified people review, edit and sign. New checks need approval before they are used.
Is a blockchain required for compliance?
No regulation requires a blockchain. What's required is a secure, time-stamped audit trail where changes don't obscure earlier entries. A blockchain-backed log is one way to make tampering detectable.
Sources
- US FDA, "Data Integrity and Compliance With Drug CGMP: Questions and Answers", guidance for industry, December 2018 (PDF).
- US FDA, guidance page for the above.
- US FDA, Warning Letters.
- 21 CFR 11.10, Controls for closed systems (Legal Information Institute, Cornell). Official text
- Press Information Bureau, Government of India, Ministry of Health and Family Welfare release, 24 March 2026. (An earlier release of 12 August 2025 described the conditional extension to 31 December 2025: )
- JSA Advocates and Solicitors, "Prism: Healthcare, Revised Schedule M", January 2024 (secondary source).





