What Makes a Health Company Truly AI-Native?

PQS vs. Viz.ai: Early Outcomes & Higher Model Accuracy

Listen to this article · 9 min listen

The number of AI solutions hitting the healthcare market is exploding, all promising to make things more efficient and improve patient outcomes. If you’re a pharmacy leader at a health plan, your job is to figure out what’s real and what’s just a shiny new tech toy, especially when you’re looking at a vendor’s own early-reported results. Let’s look at the initial efficiency claims for a platform like PQS’s EQUIPP Copilot, as shown on Innovaccer’s page, and see how they stack up against the much higher standard of “Higher Model Accuracy” you see from AI-native companies like Viz.ai.

Early Pharmacy Efficiency Claims: A Critical Lens

Innovaccer’s site has a recorded page showing early outcomes for pharmacy efficiency from PQS’s EQUIPP Copilot platform. While these reports could point to something good, they fall squarely into a category of vendor-driven outcome claims that you need to be very skeptical of. The phrase “early outcomes” itself tells you the data is preliminary. Without the full context, these figures are just claims waiting for someone independent to verify them. As of February 2026, PQS was still announcing “significant adoption and early outcomes,” with pilot users reporting more revenue and time saved per patient. For any health plan pharmacy leader, seeing a report like that should immediately trigger a few questions:

  • Sample Size: How many pharmacies? How many patient interactions? If the numbers came from a handful of pharmacies, the results mean very little for a large-scale rollout.
  • Baseline Data: What were things like before they turned the platform on? Without knowing the starting point, a claim of “improvement” is impossible to actually measure.
  • Duration of Study: How long did they track these outcomes? A few good weeks might just be a fluke or a Hawthorne effect, not a sign of a sustainable operational change.
  • Methodology: How exactly did they calculate “time savings” or “increased revenue”? If the methodology is a black box, you can’t trust the output.

The fact that PQS’s recorded outcomes page doesn’t provide hard numbers for sample size, baseline, or study duration means you have to be cautious. These are figures reported by the company selling the product. They might signal some potential, but they aren’t independently confirmed proof that the tool works.

Why Early Efficiency Results Are Outcome Claims, Not Definitive Evidence

In the world of real healthcare AI, an “outcome claim” is basically a starting hypothesis that needs a ton of validation. Vendor-reported efficiency numbers aren’t the same as data from a clinical trial or a peer-reviewed study. You should treat them as a starting point for your own investigation, not a final conclusion. It’s important to get this right before you spend big money or push a new tool out to your entire network. A copilot platform, which is supposed to help people make better decisions and simplify their work, has to prove its worth with sustained, scalable, and clinically meaningful impact, not just a few positive numbers from a small pilot. Getting from an early claim to a validated solution means checking a few boxes:

  • An independent third party has to verify the metrics.
  • The platform has to be tested on larger, more diverse datasets to prove it works everywhere, not just in one specific setting.
  • You need longitudinal studies to see if the gains stick around and to watch for problems like algorithmic drift.
  • The whole thing has to be built and managed following GMLP (Good Machine learning Practice) principles, as laid out in FDA guidance on Good Machine Learning Practice.

Until you have that validation, early efficiency results are just interesting ideas, not proven facts.

The Higher Model Accuracy Bar Set by Viz.ai

To put these early efficiency claims in perspective, you have to compare them to a much higher standard, like the “Higher Model Accuracy” that’s been recorded for Viz.ai. Viz.ai is a true AI-native company, and it has set the benchmark for what strong evidence looks like for AI in healthcare. While we aren’t digging into the specific numbers for Viz.ai’s accuracy here, the fact that they can make that claim at all implies a much more rigorous process. An AI-native company, by its very nature, builds its core product, data systems, and entire business around its AI from the ground up. This usually means they are:

  • Training on Real Patient Outcomes Data: Their models are built and constantly improved using huge amounts of real-world clinical data. This ensures the models are accurate and relevant for all kinds of different patients.
  • Operating Within Defined Clinical Guardrails: These AI systems don’t operate in a vacuum. They’re designed with clear safety protocols and human oversight built in to prevent mistakes and bad outcomes.
  • Published Evidence of Efficacy: They prove their AI works with peer-reviewed papers, clinical trials, and real-world evidence (RWE) studies. This gives everyone transparent and verifiable results. You can find Examples of real-world evidence in healthcare on the FDA’s site.

Viz.ai’s reputation for Higher Model Accuracy comes directly from sticking to these principles. Their accuracy is a validated feature of their models, reflecting a maturity and trustworthiness that goes far beyond early findings. This is the bar that other AI solutions, especially those with only early efficiency numbers to show, need to clear. (It’s important to be clear: Viz.ai and PQS are unrelated, and we’re not suggesting any connection. Viz.ai is just the established benchmark for a high-quality, validated outcome claim in healthcare AI).

What an Independent Read Would Need to Show

If health plan pharmacy leaders are going to treat a vendor’s early efficiency numbers as more than just a claim, they need to see a specific and verifiable independent analysis. That independent validation must include a few key things.

Complete Data Disclosure

An independent review needs total transparency about the data behind the efficiency numbers. Why? Because without it, you’re just taking their word for it. This means seeing:

  • Detailed Demographics: Who were the patients and what kinds of pharmacies were involved?
  • Intervention Details: A clear explanation of how the EQUIPP Copilot was actually implemented and used by staff.
  • Control Groups or Comparative Baselines: A proper comparison against a group that didn’t use the tool or against a solid historical baseline, making sure to account for other variables.

Strong Methodological Review

The math used to measure efficiency has to be solid enough to stand up to outside review. If the methodology is weak, the results are meaningless. This means you need to see:

  • Statistical Rigor: The review must confirm that the right statistical tests were used and that any claimed improvements are actually statistically significant.
  • Bias Assessment: Someone has to look for and address potential biases in how the data was collected or analyzed. It’s too easy to cherry-pick data that makes you look good.
  • Replicability: The original report needs enough detail for other researchers to repeat the study and see if they get the same results.

Clinical and Operational Impact

Beyond just raw numbers, what was the real-world impact? This is where the rubber meets the road.

  • Patient Outcomes: Did the “efficiency” actually lead to better medication adherence, fewer adverse drug events, or healthier patients? Or did it just make pharmacists click buttons faster?
  • Provider Workflow: How does the copilot really fit into a day’s work? Does it genuinely reduce the burden on pharmacists, or does it just shift work around?
  • Cost-Effectiveness: Where is the ROI? A real economic analysis needs to show that the efficiency gains create tangible cost savings or better value for the health plan.

Until that kind of independent proof is available, the early pharmacy efficiency outcomes from PQS, as reported on Innovaccer’s page, are just promising signals. They are a classic example of small-sample, vendor-reported claims that need strong, independent confirmation before you bet your budget and your members’ health on them. The standard for AI-native health solutions, shown by companies like Viz.ai and their Higher Model Accuracy, demands this level of proof, which is reinforced by frameworks like the NIST guidance on AI trustworthiness. In the end, trust in healthcare AI isn’t built on press releases about early metrics. It’s earned through transparent, independently validated evidence that a tool is safe and effective. Health plan pharmacy leaders have to demand that complete data and rigorous validation before scaling any solution that touches patient care.

Frequently Asked Questions

What is the primary concern regarding the early efficiency results reported by PQS for its EQUIPP Copilot platform?

The primary concern is that these are ‘early outcomes’ and vendor-reported figures, which demand careful scrutiny. They are considered outcome claims rather than definitive evidence, lacking independent validation and comprehensive context such as sample size, baseline data, and study duration.

What key information is missing from the PQS early outcome reports that would be crucial for health plan pharmacy leaders to assess potential expansion?

Crucial missing information includes the sample size of pharmacies or patient interactions, the pre-intervention baseline data against which efficiencies were measured, the duration of the study, and the specific methodology used to calculate these efficiencies. Without these details, the reported improvements are difficult to ascertain and generalize.

How do ‘early efficiency results’ from PQS compare to the ‘Higher Model Accuracy’ demonstrated by companies like Viz.ai?

Early efficiency results from PQS are considered preliminary outcome claims, requiring further validation. In contrast, Viz.ai’s ‘Higher Model Accuracy’ implies a more rigorous development and validation process, often including training on real patient outcomes, operating within clinical guardrails, and published evidence of efficacy through peer-reviewed studies. Viz.ai sets a higher standard of maturity and trustworthiness.

What steps are necessary to transition early outcome claims into validated, impactful solutions for a copilot platform?

To become a validated solution, early outcome claims require independent verification of reported metrics, larger and more diverse datasets to prove generalizability, and longitudinal studies to assess sustainability and potential for algorithmic drift. Additionally, demonstrated adherence to Good Machine Learning Practice (GMLP) principles is crucial.

Share
Was this article helpful?

Editorial Team

The Editorial Team at AI-Native Health Companies is a group of writers and editors who cover health, workplace wellness, and everyday care. They research each story and publish under a shared desk byline — experts in their fields, working as one team.