I see a lot of company pitches in my job. I also see a lot of AI pitches.

I don’t see a lot of good ones of each.

Last week I chatted with the folks behind NetraMark Holdings (AIAI.C) and I’ve rarely moved so swiftly from “Okay, so what’s actually new here” to “Why haven’t I heard of this before” to “Are you fucking kidding me.”

The Problem Being Solved

Every investor presentation starts with a big problem, and NetraMark’s is a familiar one: most new drugs fail in clinical trials. Despite billions poured into R&D, roughly 90% of drug candidates fall short in clinical trials (asbmb.orgasbmb.org). Only about 1 in 10 drugs that enter human trials ever earn FDA approval.

Why? Often a medicine that works for some patients shows no meaningful effect in the broader trial group, or high placebo responses and side effects muddy the water. Sometimes there’s fraud involving companies brought in to run the trials. In plain English, many trials are designed too broadly and the true signal of a drug’s benefit gets lost in the noise of patient variability.

When a doctor prescribes you a drug, it’ll often come with provisos. Do you have allergies? Are you taking other drugs that might interfere with this one? Are you a redhead aged 25-39 who has been to Belize recently?

But in the trial stages, drugs are trialed against what the scientists call ‘a shit ton of people.’

NetraMark zeroes in on the issue of heterogeneity in clinical trials. The company says hidden subpopulations of patients can make or break a study’s outcome, and while the companies running the trials will weed out obvious problem candidates, it’s the unobvious problems that can kill your trial – especially at the second stage, when trials involve dozens, not thousands, of patients. In a small sample size, a two-person problem can be massive.

For example, perhaps a drug works well in younger patients but not older ones, or only in people with a certain genetic marker. If the trial isn’t designed to account for that, the overall results might look mediocre and the drug gets shelved – a costly “failure” that might have succeeded with a smarter trial design. NetraMark’s mission is to help pharma companies “know their patients” in depth (netramark.com): find those niche groups (the responders, the non-responders, the placebo-sensitive, the potentially fraudulent) before it’s too late. By identifying these subpopulations driving drug responses, placebo effects, or adverse events, the idea is to “enrich” future trials – i.e. include more of the patients who benefit and exclude the ones who don’t.

That’s not cheating – any drug hitting the market will work better on some patients than others, or have different side effects to some. By removing those who wouldn’t be getting the drug if it were approved anyway, the companies can actually get an idea of how the drug will work WITH THOSE WHO’LL LIKELY BE TAKING IT.

Is this a real problem or just academic tinkering?

The industry consensus is that it’s very real. Failed Phase II or III trials often get dissected post-hoc to see if any subset of patients did respond, but its often several indicators convinced that lessen results, not easy-to-spot single metrics.

In fact, the U.S. NIH has noted that improving the probability of success in trials is critical for both science and business (s27.q4cdn.com). Drug developers, especially in complex diseases like neuroscience and oncology, face abysmal success rates – CNS (central nervous system) drugs, for instance, succeed only half as often as the already-low industry average.

In those fields, patients are extremely diverse and trials are longer and costlier than normal, so a one-size-fits-all trial design often fails. NetraMark is tackling this by asking: What if, pre-randomization, a trial could be pre-adjusted to focus on the patients who will actually get the most benefit from the drug?

It sounds almost too good – like having an “insider info” on which patients to pick. If it works, it could rescue drugs that would otherwise flop, saving companies from expensive failures and bringing effective therapies to the right people. That’s the promise. But delivering on it is another matter, which we’ll scrutinize next.

The Business Plan

NetraMark isn’t a traditional drug developer – it’s selling a kind of AI-driven consultancy or software solution to pharma companies themselves. The business plan has two prongs: direct engagements with drug companies, and partnerships with Contract Research Organizations (CROs) that run trials.

In practice, NetraMark will take a pharma sponsor’s clinical trial data (usually from a Phase I, II, or III study) and perform an AI analysis to find those hidden patient subgroups and insights. The output might be recommendations for new inclusion/exclusion criteria – essentially, who to include or leave out next time to boost the drug’s chance of success (netramark.com). The company then charges fees for these projects. Currently, the company brings in low-to-mid six figures for one of these projects, which is cheaper than they think they can get by a large favtor, but helps sell an otherwise unknown product enough that it can begin to become known.

To scale up, NetraMark has been building a sales pipeline in the pharma and biotech industry. According to management’s March 2025 discussion, they started commercial efforts in mid-2023 and grew from 19 sales leads to over 140 leads by early 2025. That’s a huge jump, implying strong interest on paper. They also report signing five contracts with mid-sized pharma companies so far (s27.q4cdn.com), and notably inked a global partnership with Worldwide Clinical Trials, a large CRO that just sold for $2 billion, in April 2025 (prnewswire.com).

This Worldwide deal is important: it essentially gives NetraMark a channel to many biotech and pharma clients through an established player that some estimate runs more than 60 contracts a year, and if just half of those end up using AIAI’s tech, that’s a fat payout for the company. Worldwide will offer NetraMark’s AI analysis as part of its own services, initially in neuroscience and oncology trials (prnewswire.com). The CRO’s COO even stated that this could lead to “fewer required patients per trial, reduced timelines, lower costs, and ultimately increased success rates” for their clients. That’s a strong endorsement and suggests the pain point is felt industry-wide.

For the next year, management is bullish: they forecast a contract backlog of C$8–10 million (around $6–8M USD) over 12 months (30 Worldwide contracts at $300k per would run to $9 million), as both direct sales and partner-driven deals ramp up. If they hit that, it’s a big step up. However, investors should take such projections with skepticism. NetraMark’s actual recognized revenue in the first half of fiscal 2025 was only C$386k (about $290k) – albeit up ~74% year-over-year. Even if they sign $8M worth of projects, revenue is only booked as those projects reach milestones (tied to trial data readouts).

In other words, cash may come slow and “lumpy.” The company openly warns that its results will “fluctuate significantly” quarter to quarter and that any predictions are hard to make at this early stage. NetraMark had roughly C$2.1 million in cash at March 2025 after some warrant exercises, and with another few million in 35c warrants due to expire by year end, and with the share price at $1.45 at the time of writing, they could double that cash position soon.

They’ve historically operated at a loss (as you’d expect pre-marketing), so continued operations likely depend on new contracts and likely new capital raises in the future. In fact, the company acknowledges it will need to raise funds until sales can support the business. That means dilution risk for shareholders or debt if they can get it. This is typical for an early-stage tech play, but worth keeping in mind.

$2m more in warrants at 35c exercide price

So does the business model scale?

That’s an open question. The company says there’s a several week turnaround involved in getting each company’s data ready to be analysed by their system, which makes sense because data comes in all sorts of forms, but once that’s done the process is quick, repeatable, and reliable.

The CRO strategy is smart if it works, because CROs give leverage – Worldwide has 3,500 employees and works on trials in 60+ countries (prnewswire.com), and if NetraAI became a standard tool for them, NetraMark could get a slice of many trials without a huge direct sales force. The company’s new President, Josh Spiegel, came from a successful eClinical software firm that sold for $330M, and one of its board members, Dr. Angelico Carta, actually co-founded Worldwide Clinical Trials. Clearly, they have connections to pursue this channel.

The bottom line: the plan is logical – target pharma R&D’s biggest headaches and partner with the industry’s service providers – but the execution risk is high. Converting 140 “leads” into signed deals is hard in pharma, where sales cycles are long and cautious. Investors should watch how many of those leads turn into paying customers in the next year, and whether that touted ~$8M backlog materializes. If they suddenly land dozens of contracts, do they have the manpower and bandwidth (data scientists, support staff) to deliver quality results in a timely fashion? They say scalability is not their biggest issue, and that once they’ve got a few more successful projects under their belt, they expect a surge in interest – and a rise in pricing.

The Technology

Let’s cut through the buzzwords: NetraMark’s core technology is an AI engine (called NetraAI, based on their proprietary “Attractor AI” algorithm) that sifts through clinical trial data to find patterns. In plainer terms, it’s a machine learning tool that looks at all the variables you measure in a clinical trial – demographics, lab results, genetics, cognitive scores, you name it – and tries to group patients by how similarly they respond to the drug. The goal is to discover, say, this cluster of patients responded well and that cluster responded poorly, and then figure out what defines those clusters (e.g. “patients with high inflammation markers and low cognitive scores had no benefit from the drug”).

Traditional stats methods (t-tests, regressions, neural nets, etc.) struggle when data is “high-dimensional” (lots of variables, small sample size). NetraMark claims their Attractor AI is specially designed for such situations – it “pulls together” patients with similar profiles in a mathematically novel way, without overfitting to noise. Crucially, it doesn’t force every patient into an explanation, leaving some as “unexplained” outliers (which they charmingly nickname the “ugly” patients, versus the “good” and “bad” ones). By focusing on the explainable groups, they aim to deliver clear hypotheses a drug team can act on.

What does a deliverable from NetraMark actually look like? According to their materials, it includes:

• A list of key variables (biomarkers, clinical measures, etc.) that define each “persona” or subpopulation driving a response or placebo effect.
• The statistical significance (p-value) and effect size of each of those subpopulation findings.
• Specific suggestions for inclusion/exclusion criteria to use in the next trial to capture the “good” patients and avoid the “bad” ones.
• A narrative explanation via NetraGPT, their use of GPT-4-like large language models to provide a “human-readable” summary of the findings and relevant scientific literature (netramark.com).

In essence, NetraAI might tell a client something like: “Patients with biomarker X above a certain level, combined with symptom Y, were 3 times more likely to respond to the drug. If you enroll more of those and fewer patients without that profile, your next trial has a higher chance of success”.

It can also flag those who are likely placebo responders – NetraMark even integrated an academic scale called PRPS (Placebo Response Probability Scale) to help identify patients prone to placebo effects. That could help sponsors design trials to mitigate placebo noise (a big problem especially in psych and pain trials). The platform is able to unify “disparate and irregular” datasets too, meaning it can handle different data types together – for example, genetic data plus clinical scales plus imaging – which many pharma teams struggle to do internally.

Is this tech truly novel or just AI-washing?

NetraMark’s team pitches Attractor AI as a fundamentally new paradigm (the founder, Dr. Joseph Geraci, is a mathematician who’s been at it for 7+ years). To their credit, they aren’t just throwing around AI jargon without substance; they list all the usual machine learning suspects and explicitly say those methods have limitations on clinical data. The unique bit is the “attractor” concept, which sounds akin to an advanced clustering/feature selection hybrid. They have produced peer-reviewed publications applying their approach to real datasets – for instance, uncovering genetic drivers in lung cancer subtypes and addressing placebo response in schizophrenia trials (as cited on their website) (netramark.com).

This suggests there is some scientific merit, not just marketing. However, we should be wary of any claim of having no direct competitors. There are other AI companies tackling similar problems in pharma R&D. Some use different techniques (like digital twin simulations, or Bayesian approaches to adaptive trials). Even big CROs and pharma companies themselves use machine learning for patient stratification to an extent, but mostly lean on large language models (LLMs) that aren’t built for complex math.

NetraMark will have to stay ahead on accuracy and usability to avoid being leapfrogged by a well-funded rival or an in-house solution, though they have a really big moat in that, a lot of potential customers they’ve talked to have said ‘there’s no way you can extract the results you’re looking for from a second phase trial – the patient numbers and sample sizes are just too small.

In essence, the pharma companies have given up, sometimes abandoning nine-figure IP investments, because they think it can’t be done… while NetraMark is doing it.

The positive beyond that is their focus on explainability – regulators and pharma executives are notoriously skeptical of black-box AI. NetraMark emphasizing interpretable results (with an assist from NetraGPT to tie in medical literature) is savvy, because a brilliant insight is useless if the trial doctors can’t understand or trust it.

At this stage, the technology shows promise but is still unproven at scale. It’s one thing to retrospectively find patterns in a completed trial data set; it’s another to prospectively influence a trial design that later results in a successful Phase III outcome. NetraMark will need some wins – case studies where a client followed their AI guidance and then saw a drug succeed or avoid failure – to truly validate that this isn’t just cool math on paper.

Until then, the tech remains in that gray zone between “intriguing innovation” and “science project.” It certainly addresses a real scientific challenge, but the next section examines whether the industry is actually clamoring for it.

The Need

We know pharma companies hate wasting money on failed trials. A Phase III trial can cost hundreds of millions, so a tool that even slightly improves success rates could save or earn huge amounts. The clinical trials market is enormous (estimated around $21 billion in 2024 for trials operations in the U.S. alone (s27.q4cdn.com) and growing as the number of studies rises each year. Meanwhile, the field of AI in clinical trials is hot and competitive – globally it’s already a multi-billion dollar space and projected to grow at ~20%+ annually in the coming years (globenewswire.com). So on paper, there’s both a need and a market for what NetraMark offers: AI-driven insights to de-risk and speed up drug development.

But does NetraMark specifically fulfill a validated need, or are they pushing a solution in search of a problem?

The concept of patient stratification and enrichment in trials is not new. Pharma companies routinely do subgroup analyses on trial data (gender differences, genetic markers, disease subtypes, etc.). The problem is that traditional analyses often fail to find something useful because they look at the obvious potential problems, and not combinations of things because that’s REALLY hard math.

NetraMark is saying “we can find what others missed, and find it faster.”

If true, that addresses a clear pain point: many drugs might actually work if only we knew which patients to give them to. Regulatory agencies are also encouraging more precision – the FDA has pathways for “enrichment strategies” to approve drugs for narrower populations when broad ones fail. NetraMark seeking a Critical Path Innovation Meeting (CPIM) with the FDA is a smart move on this front. It shows they’re aiming to get regulators comfortable with AI-designed trial criteria. If the FDA gives positive feedback (we should hear if that meeting is granted and how it goes), it could validate that this need is recognized at high levels. In an industry as risk-averse as pharma, an FDA nod would be a green light for companies to try NetraMark’s approach.

From the customer perspective, early traction suggests genuine interest. Those five pharma contracts indicate at least some R&D teams have voted with their wallets.

The Worldwide CRO partnership is another indicator that the industry sees enough merit to integrate it into services.

Also, look at the team NetraMark assembled: senior pharma and clinical trial experts (ex-IBM business exec as CEO, former Italian FDA-equivalent chief as regulatory advisor, veteran psychiatrists and CRO founders on board. These folks wouldn’t waste their time on a fake problem – they must believe there’s a need here from their decades of experience. It’s telling that Worldwide’s COO explicitly said this AI could “increase success rates… and bring life-changing therapies to market” faster. That’s the need, in a nutshell, straight from a potential customer.

Obviously, investors should consider whether big pharma might simply develop similar capabilities in-house or whether competitors could eat NetraMark’s lunch but, to that, I point you back to the earlier point that early customers have believed their data sets to be too small to find any trends, and been surprised when the opposite is true.

What pharma company will take the time and effort to copy NetraMark when they don’t see them as competition – and wont change that view until NetraMark starts banging out wins.

Large companies like Roche or Novartis have entire data science departments working on AI for trial design. Many startups are attacking different angles (from AI-driven drug discovery to using real-world data for trial optimization). NetraMark’s niche – analyzing small, complex trial datasets for insights – is a bit esoteric, though still very important.

The question is will pharma pay an external company for this, repeatedly? NetraMark will have to demonstrate that it’s easier and better to hire them than to let an internal biostats team handle it. That likely means continuing to publish compelling results and, ideally, scoring a high-profile “save” (eg, helping a drug get approved that otherwise wouldn’t have).

In summary, the need for better trial design is very real. NetraMark’s approach aligns with what many in the industry say they want – more precision, more use of AI, fewer failures. The challenge is convincing conservative stakeholders that this relatively new company can deliver those outcomes reliably. It’s one thing to need something; it’s another to trust a small vendor to provide it. The next year will test how deep that need is, as measured by repeat business, new client signings, and any public success stories.

The Way Forward

What should an investor watch for with NetraMark in the next 12 months? In a word: execution. The company has painted an exciting picture – now they must hit some milestones to prove it’s not just hype. Here are the key things to keep an eye on:

• Contract Conversions and Revenue: Does that projected C$8–10M backlog materialize, and do at least a few million of it convert to actual recognized revenue? Remember, management expects “strong momentum” across both direct and partner channels. Investors should look for evidence of that momentum in quarterly updates – e.g., number of active projects, size of backlog, etc., not just optimistic language.
• Delivery on the Worldwide Trials Partnership: This deal could be transformative if it yields substantial business. Watch for any indication of how it’s going – are Worldwide’s clients actually opting to use NetraAI in their trials? Ideally, NetraMark might announce the first study or customer that came through Worldwide. The CRO route is a force-multiplier if executed. Also notable: a co-founder of Worldwide sits on NetraMark’s team, which could both spur the integration and potentially make NetraMark an attractive acquisition target for a CRO if the tech proves out.
• Regulatory Engagement (FDA CPIM): By mid-2025, NetraMark expected feedback from the FDA on their Critical Path meeting request. If the FDA meeting happens and NetraMark gets positive signals (or any public mention from FDA about AI in trial design including them), that would be a credibility boost. Conversely, if nothing is said about it again, one wonders if it quietly didn’t go anywhere. Investors should watch for any updates on regulatory front – even EMA or other agencies, since one of NetraMark’s officers used to be at the EMA. Regulatory acceptance could be a game-changer, turning the technology from “nice-to-have” to almost a recommended practice in trial planning.
• Case Studies or Clinical Outcomes: This is perhaps the most important long-term indicator. In the short 12-month horizon, we might or might not see a clear “trial saved by NetraMark” story, since trials take time. But even an anecdotal success (e.g., “Company X used NetraAI insights to redesign their Phase II trial and saw improved results”) would provide powerful validation.
• Competitive and Market Developments: The AI-for-pharma space is evolving rapidly. Investors should watch how NetraMark’s offering stacks up against peers. Are larger analytics companies encroaching on their turf? NetraMark’s value prop needs to remain differentiated – perhaps through superior explainability or niche focus on tough disease areas. The company’s initial focus on CNS and oncology trials made sense due to high need there.
• Financial Health: NetraMark is still burning cash, so keep an eye on its runway.

The Blue Sky

• New verticals: The company is exploring options in using their tech not just to look for faults in clnical trials, but also to track location fraud, a big problem in trial make up, and something they’re already seeing results in.
• Licensing opportunities: Could the company shift quickly to licensing their tech to bigger players and riding out the future with a fat monthly royalty cheque?
• Equity opportunities: Often a drug company that fails in clinical trials goes under financially. Could NetraMark save a trial for a company that’s otherwise out of money, by working for equity in the finished IP?

There are definitely risks, but I see them as thus:

1. That the tech doesn’t work, something they appear to be demonstrating isn’t the case
2. That competitors try to overtake them, unlikely when it’d be cheaper to just buy the company
3. That they run out of money, something I think is unlikely considering the low cost nature of the projects, the low staffing needs to run them, and the fact that money is beginning to roll in. If your worst case scenario is doing a raise right as you’re kicking into business development, that’s not terrible.

In conclusion, NetraMark Holdings offers an ambitious solution to a very real problem in drug development. They have a plausible plan and an experienced team, which already sets them apart from the many AI startups that are, frankly, all talk. However, as investors we must be skeptical of rosy projections. The technology, while intriguing, still needs to prove that it can consistently deliver value. The business model – combining direct pharma sales with CRO partnerships – makes sense, but it hinges on actually closing deals and generating repeat business in an industry that moves cautiously.

Over the next year, the “straight talk” is this: watch for concrete signs of market traction and technical validation. Is NetraMark signing more clients and generating revenue at the pace they promise? Are their partners and customers vocal about the benefits they’re seeing?

The pieces are in place: big problem, clever tech, savvy team. Now it’s about execution in the real world of pharma R&D. Investors should remain cautiously optimistic but verify every claim against actual results. In the coming 12 months, success for NetraMark will be best measured not by press releases or projected backlogs, but by evidence that its AI insights are becoming an integral part of how trials are designed – and that the company can turn that into a growing, sustainable business. Keep your eye on the data (and the cash), and you’ll cut through the fluff to see if this story is truly starting to pan out.

— Chris Parry

FULL DICLOSURE: No commercial connection.