Venrock has released the results of its 2026 survey of over 200 (so, 201) healthcare-related experts. These are folks from across the spectrum including startups and investors to HCPs and academics and life sciences.

When your own skin is in the game (literally) healthcare experts say they trust AI with 39% wanting a physician in the conversation.

Ambient documentation for HCPs has been seen as a big opportunity for AI in healthcare. It is still high but now experts are looking at revenue management (prior authorizations for both providers and insurance).

This question was "Where is AI actually going to move the needle in 2026? [Select up to two]."

A lot of experts expect providers and insurers to "arm up" with powerful AI tools for prior authorization management. They see this turning into a bot-vs-bot war that might not result in savings.

This question was "How will AI reshape the reimbursement landscape over the next 18 months? [Select all that apply]."

This report from Venrock includes all of the raw data at the end plus links to previous years which is the gold standard. 

Klick's Guardrail is a tool for help in creating compliant, documented submissions to MLR teams. We analyzed 40,144 MLR review submissions across 6,907 projects, 110 client accounts, and 25 material types, using our intranet platform (Genome) submission and timesheet records from August 2020 through January 2026 to see how it performs.

The TL;DR version is that it has an estimated rejection rate of around 12% compared to the industry average of 26%.

When looking at industry MLR rejection rates for individual submissions from Veeva and Guardrail we get:

To any pharma marketer these rejection numbers will seem low, but consider that most projects actually have multiple submissions involved. For a project package of 10 submissions Klick estimates that the rejection rates are amplified (1-[individual acceptance rate]¹⁰ = rate for all 10 items to be accepted).

And this improvement in approvals results in Klick's teams saving 16 days on full MLR cycle times.

Jay Goldman, EVP, Product Portfolio, gave us permission to highlight a feature that blew me away (no personal opinions -Ed.).

Guardrail now has the ability to create AI-generated video storyboards directly from video files with no text transcript. This moves the toolset from a purely text-based workflow to incorporate video natively.

The output of the video-to-storyboard conversion is nearly submission-ready.

Each frame in the submission has detailed relevant information:

Heidi is an AI ambient scribe vendor so they wanted to know how HCPs feel about the technology in 2026. They surveyed 1,823 clinicians across 25 countries in May of this year. Note that these respondents came from Heidi's internal mailing list so there will likely be bias in favor of AI in the data.

Most of the HCPs surveyed are confident but a sizeable minority have concerns that outweigh the positives.

There is no indication in the report that this data is only the HCPs that disagreed above. Even HCPs who are all in on AI will be concerned about the technology in some ways, they just feel that the benefits make it worthwhile.

The top concerns expressed by HCPs were:

Some other stats from the report:

The folks over at OpenEvidence seem to have reviewed the NOHARM study we featured in the Wire last week. In that study there was data about the AI tools doctors choose.

Anthropic makes some of its usage data available online so we looked at "Healthcare Practitioners and Technical" to see the ranking. Cardiologists were out in the lead. The numbers are small because they represent portion of all Claude usage.

A study from earlier this year will resonate with anyone struggling to keep up with the pace of change in workplace AI. Researchers surveyed 2,500 knowledge workers from the UK, US, Canada, Germany, and Spain in March 2026 to gauge their opinions about AI. The researchers found that "73% of workers acknowledge AI’s efficiency."

What emerges from this research is a very complex and very human story. There are also challenges in workplace adoption of AI.

However, these same employees want their companies to invest in more AI.

What we're seeing is the turmoil in a workforce adapting to the new tools faster than has ever been asked before. Organizations that can help employees through this transition stand to earn the most benefits from the new technologies.

BrightEdge keeps a tally on search engines and AI engines for referrals. The company shows search as a point-in-time bar chart but gives a trend for AI. We grabbed the data and charted out the search engine data.

Of that 1.5% at the bottom of that chart you then get the referrals from the dedicated AI engines.

GEO is important, of course, and Google's AI Overviews are the dark horse in this race, but it's interesting to not see much change in the search engine's dominance ... so far.

Some research from a company called geoSurge looks at how AI models choose brands to feature in results. In their testing 69% of results didn't include brands and 31% did.

Of those results with brands the ones in memory showed more often

The researchers separated results into nine categories. Unfortunately "Fitness & Wellness" is the closest to health.

The researchers do acknowledge the limitations in their methodology, and anyone wanting to see the full description should review the full article. 

This is more evidence that brand matters in the world of AI search and referrals.

This story has little direct insight, and the defendants "Hims & Hers" almost certainly make unsubscribing difficult using dark patterns, but it is a good reminder that the FTC still considers website pixels protected health data.

This may be an outcome of the FTC's Healthcare Task Force we reported on in March. Similar to then, this is a situation that pharma marketers need to monitor, and ensure that any marketing pixels being used are compliant.

These two charts show a story of correlation and usage expansion. Normally we would keep the Emarketer headlines and other material but for scaling purposes we removed them this time (see story for full image). This data is for US users.

On the left we see correlation: high internet users tend to use the internet more. This results in AI users also using search more.

More interesting is that the total number of minutes for both AI and Google is going up. This shows that AI isn't taking usage away from Google (although it could also mean that AI Overviews on Google are generating more minutes...).

Gallup surveyed 22,573 American workers in May to get a view into how AI is being used in businesses. For the first time usage has passed 50%.

For B2B marketers it's important to see that search and research is almost 50% among AI users. If you're looking to sell GEO in, this is the chart.

In a newer report on AI usage, Gallup also shows that employees whose managers support AI have higher efficiency and productivity.

It is taking a long time to dethrone Meta. 

Social Media Today published its ranking of the various social platforms. Some of the numbers have to be estimated because the companies don't release monthly active user (MAU) totals.

Doing social in Pharma? Start with Facebook, then Instagram, TikTok, Reddit, and LinkedIn depending on your audiences.