Shifting bottlenecks: Key takeaways from ACRO's regulatory summit for 2027
“
The technology is ready. Regulators are increasingly ready. The thing slowing clinical research down is us.
Jump to section
- Why aren't regulators the bottleneck anymore?
- Where should AI go first?
- Why is trust a workflow problem?
- Why does interoperability decide the race?
- The four signals at a glance
- The 2027 clinical operations plan
- The mission hasn't changed
- Frequently asked questions
2026 State of Clinical Trial Technology Report
What 400+ site, sponsor and CRO leaders said about AI and disconnected systems.
Get the reportIn September 2026, ACRO convened its Global Clinical Trials Regulatory Summit in Washington, DC. Regulators from the US, UK, EU, Japan, India and Australia shared a room with CRO, sponsor and technology leaders for candid conversations.
My biggest takeaway? The technology is ready. Regulators are increasingly ready. The thing slowing clinical research down is us.
That is a much less comfortable conclusion than "the regulations need to change." It is also a much more useful one. As clinical operations teams start 2027 planning, four signals from the summit are worth paying attention to.
Short answer
Regulators already allow sponsors to do less unnecessary work than most of them do. For 2027, the biggest gains in clinical operations will come from retiring legacy processes and pointing AI at operational data that can move between sites, sponsors and CROs.
Why aren't regulators the bottleneck in clinical trials anymore?
Because regulators in several regions already allow less work than most sponsors actually do.
One of the more memorable moments of the summit came when a regulator described an effort they launched to reduce burden in clinical trials. The unexpected obstacle to their success? Sponsors.
Across several regions, regulators already allow approaches that can materially reduce unnecessary work, including risk-proportionate monitoring and more selective safety data collection, but sponsor uptake has remained slow.
ICH E6(R3) only makes the gap harder to ignore. Quality by design and proportionality are now the expectation, and they've outgrown the conference panel. And yet, plenty of organizations are still layering new technology on top of old processes. Upload the PDF to the eISF. Print it. Sign it in wet ink. Scan it. Upload it again. Congratulations: we digitized the filing cabinet.
56%
of clinical trial documentation still originates on paper.
Source: Florence's 2026 State of Clinical Trial Technology Report
For 2027, "What will regulators allow?" matters less than "Which of our own processes are we still defending because 'that's how we've always done it'?"
Where should AI go first in clinical trial operations?
Into the operational work that fills a coordinator's and PI's week.
There was broad agreement that the highest-value near-term use of AI in clinical research is operational work: startup packets, document QC, regulatory binder upkeep, reconciliation and the other work that quietly eats a coordinator's and PI's week.
This work also generates an enormous amount of data. Every signature, audit trail entry, document version and completed task leaves behind an operational signal. Historically, we've captured most of it to prove the work happened. Very little of it has helped us run the trial better. This has to change.
In 2027, the organizations that get this right will use those signals to identify startup delays earlier, see when sites are overloaded, anticipate missing work and route tasks before a problem becomes a deviation.
The caveat is that AI on top of fragmented data simply produces confident errors faster. The prerequisite for success here is connected workflows where documents and data can move between sites, sponsors and CROs without someone becoming the human API.
We have spent years describing fragmentation as "too many logins." The bigger problem is that the data cannot travel.
Why is sponsor and site trust a workflow problem?
Because every certified-copy request and duplicate entry exists only because the sponsor doesn't trust the site's record.
Trust was the summit's through-line, whether in data, in regulators, across nations or among stakeholders. Most of the room treated that as a policy problem, but I'd argue it is a workflow problem, and it shows up first at the site.
Every request for a certified copy, every duplicate entry into a sponsor system, every reconciliation of the same record across three platforms is a small vote of no confidence. The sponsor does not fully trust the site's record, so it builds its own. Multiply that across a portfolio and you get the coordinator burden everyone agrees we need to cut.
AI raises the stakes. When an algorithm flags a missing document or drafts a reconciliation, the auditor's first question will be "who checked this, and where is that recorded?" Teams that can answer from the audit trail will get to scale AI. Teams answering from a spreadsheet of good intentions will not.
For 2027, I'd track how often your people re-verify work. Every redundant check is trust you have not yet built into the workflow.
Key takeaway
When AI flags a missing document or drafts a reconciliation, auditors will ask who checked it and where that review is recorded. Teams that can answer from the audit trail are the ones that will get to scale AI.
Why does interoperability decide the global race for trials?
Because harmonized policy can't speed up trials if the systems underneath still can't exchange data.
Another thing that came through clearly in Washington is that countries are not waiting around. We heard national strategies from India, Australia, the UK, the EU and Japan, all focused in different ways on speeding trial startup, modernizing infrastructure and making their research environments more attractive.
The US is having its own version of that conversation right now with Operation TrialBlazer (announced by HHS in June 2026) and other efforts focused on Phase I competitiveness, regulatory modernization and broader trial networks.
Everyone wants more trials. Everyone wants faster startup. Everyone wants technology to help. But harmonized policy only gets us so far if the systems underneath it still cannot talk to one another.
A site should not have to manually shuttle the same documents, credentials, training records and study information from one partner's platform to another. That is not interoperability. That is people doing interoperability. The markets and research networks that gain an advantage will be the ones where trial information can move with far less friction.
Interoperability is the quiet infrastructure underneath all the policy talk.
The four signals at a glance
| Signal | What the summit showed | What to do in 2027 |
|---|---|---|
| Regulation | Regulators already allow risk-proportionate monitoring and selective safety data collection | Retire steps the regulation doesn't require |
| AI | The highest-value near-term use is operational work | Point AI at connected operational data first |
| Trust | Duplicate checks exist because sponsors don't trust site records | Track how often your teams re-verify work |
| Interoperability | Countries are competing on faster trial startup | Ask vendors whether data can leave their system cleanly |
What should clinical operations leaders put on their 2027 plan?
If you lead clinical operations, here is my short list:
- 1.Audit your own friction first.Pick three processes where the regulation allows less work than you actually do, and retire the extra steps.
- 2.Treat operational data as a product.Decide which workflow signals you want to see before the study starts, not after it slips.
- 3.Build AI governance before AI.Map the chain from data to decision, including exactly where a human signs off.
- 4.Ask vendors about portability before features.If data cannot leave a system cleanly, what you've bought is a silo with a nice interface.
- 5.Bring sites into the design.The people closest to the work know which steps are compliance and which are folklore.
The mission hasn't changed
My colleague Andrea Bastek, VP of Market Strategy, summed up the summit better than I can:
"AI and technology should take the manual work off our plates so we can focus on the mission that brought us here, excellence in clinical trials."
Andrea Bastek, VP of Market Strategy, Florence Healthcare
The regulators have opened the door. 2027 is the year we find out whether the industry walks through it.
Keep the conversation going at Research Revolution 2026
October 25 to 27 in Atlanta.
Frequently asked questions
What happened at ACRO's Global Clinical Trials Regulatory Summit?+
ACRO brought regulators from the US, UK, EU, Japan, India and Australia together with CRO, sponsor and technology leaders in Washington, DC on September 22 to 23, 2026. Florence's Catherine Gregor came away convinced that the technology and regulators are ready, and that industry processes are now the main thing slowing clinical research down.
Why are regulators no longer the bottleneck in clinical trials?+
Regulators in several regions already allow approaches that reduce unnecessary work, such as risk-proportionate monitoring and more selective safety data collection. Sponsor uptake has been slow. Many organizations still layer new technology on top of old paper processes instead of retiring the extra steps.
Where should clinical operations teams use AI first?+
In operations. Summit participants broadly agreed that the highest-value near-term use of AI is work like startup packets, document QC, regulatory binder upkeep and reconciliation. AI needs connected workflows to work, because AI running on fragmented data produces confident errors faster.
How does ICH E6(R3) affect clinical operations planning for 2027?+
ICH E6(R3) makes quality by design and proportionality the baseline expectation for clinical trials. For 2027, that means identifying processes where the regulation allows less work than your team actually does and removing the extra steps.
What should be on a clinical operations plan for 2027?+
Catherine Gregor recommends auditing your own process friction first, treating operational data as a product, building AI governance before deploying AI, asking vendors about data portability before features, and bringing sites into process design.
Sources
- ACRO 2026 Global Clinical Trials Regulatory Summit
- Operation TrialBlazer, U.S. Department of Health and Human Services
- ICH E6(R3) Guideline for Good Clinical Practice
- 2026 State of Clinical Trial Technology Report, Florence Healthcare
You May Also Like
