The AI-Enabled Research Toolkit: Google Alerts, OpenAI and Similari

In a sense, R&D professionals are spoilt for choice when it comes to automated search tools. There is now a wide (and growing) range of solutions that can streamline the process of monitoring technical data, and extracting insights from it. 

Here, we’re unpacking three solutions that augment human search capabilities: Google Alerts, ChatGPT3 and, of course, Similari, to find out where their strengths and weaknesses lie. 

Google Alerts: your trusty research sidekick for the last 20 years 

Google Alerts can be a powerful tool for gathering information about new and emerging trends, technologies, and ideas that impact innovation strategy. 

By simply signing in and setting some parameters, users can track industry keywords, flag key trends and stay up to date with the news in their field. They can even keep tabs on competitors, provided the information makes it into the news cycle (more on this later).

And it’s all served to them, daily, weekly, or however they prefer, via email. All in all, it’s a useful tool that can help innovators to at least keep up with the curve, even if staying ahead of it remains just out of reach, for reasons we’re about to explore. 

Everyone has a blindspot (even Google)

For all its (completely free) benefits, there are some limitations to keep in mind when it comes to using Google Alerts for innovation intelligence.

Available sources: vast, but limited in important ways

Sources are limited to pages indexed by Google, and within that, mostly news sources (recall what we said earlier about competitors). That may be enough for certain use cases, but it leaves a lot of relevant information out of the frame, especially when it comes to scientific and technical literature. 

Not everything worth knowing makes it into the news – and if it isn’t there, it won’t make it into a daily Google Alert.

The brute facts aren’t enough on their own

Users have noted long standing challenges like false positives (unrelated content that matches the keywords) or false negatives (relevant content that does not contain the exact keywords). 

Additionally, Google Alerts does not offer much customization, so users can’t combine multiple searches into a single alert. And because it’s email-only, there’s no easy way to combine all emails and news stories into a single source. Reporting on the facts, and analyzing their significance – that’s left up to researchers to do the old-fashioned way.

From data to insight: the missing step

Perhaps most importantly, Google Alerts can’t automate that crucial step from data to insight – and insights are what fuels innovation. The work of figuring out the story the data is telling remains with the user. In other words, it’s a valuable tool for exploration, but not exploitation.

ChatGPT3 and the future of AI-enabled research

Meanwhile, at this very moment, the internet’s favorite chatbot is having human-like conversations with millions of users. What are they talking about? In short, everything, including a wide range of business use cases: automating marketing & sales, debugging code, and most importantly for our purposes – R&D. 

But it’s not just the impressive NLP that has the whole world abuzz. It’s also the extraordinary ease of use that ChatGPT3 provides. The user simply gives an input, asks a question, and sits back while the generative AI works its magic. Unlike email-based alerts, this experience is conversational. The user can ask follow-up questions or demand justification. 

Scientists and researchers can curate their own corpus of technical information and feed it to the system, and leverage its computing power through that simple interface. 

ChatGPT’s limits are harder to find, but they exist

Since it appeared in 2022, ChatGPT has been used by millions of people, and it’s received intense scrutiny in the process. This has provided useful input for the platform itself, while pointing up its limitations. 

The problem of data (again): historical and reactive 

As we saw with Google Alerts, historical data can only take you so far. ChatGPT has limited visibility past 2021. And while that could change in the future, its text-based output is best suited to providing a historical account of past data.

Innovators seek uncommon knowledge, not the common ground

ChatGPT is adept at drawing on billions of data points to come to a single answer. It can tell users, usually with great accuracy, where the common ground lies in a specific scientific dispute. It can even point you to its sources (if you ask nicely). This is all extremely impressive. But it’s not what researchers need to help them innovate, because innovation is about challenging the status quo, and finding new ways to interpret the data. 

And even with a private corpus of texts to work with, ChatGPT’s output is always text-based. That’s ideal for marketing, or generating source code, but not for scientific research, where data visualization plays a vital role in decision making. 

Similari: insights at your fingertips

We would be remiss to not tackle the question of where Similari fits in this picture. In short, Similari allows researchers to go deep where other tools favor breadth. In a matter of minutes, researchers can configure Similari to start watching, learning and generating deep insights about their specific field of inquiry. 

Picking up where search analytics leaves off: an insight mechanism that learns over time

Traditional search analytics platforms excel at finding data and presenting it in a predefined way. But they leave the heavy lifting to the researcher or analyst who must spend time curating that information, deciding what is and isn’t relevant, and juggling multiple disparate data sources. Similari replaces that workflow with a unified, live feed of up-to-date insights.

And over time, Similari learns from the human user, absorbing their preferences and imitating their behaviors – all without explicit instruction, thanks to its sophisticated underlying ML framework. That’s how Similari is able to slash manual data monitoring time by 80% or more, while actually enhancing research outcomes and fuelling innovation. 

Harnessing the power of AI to know everything, all the time

Augmenting human capabilities with AI is now a priority shared by businesses in almost every industry. And as the market matures, it’s providing more and more targeted solutions for highly specified business needs. Similari is one of them. And, for innovation professionals at least, it’s one of the most effective. 


To learn more about how Similari enhances and streamlines research for R&D and innovation teams in the life sciences and beyond, schedule a demo with our team. We’re ready to show you everything you could be missing with traditional search analytics – and much more besides.

What’s Next for the US Cannabis Market? Key Trends, Challenges, and Opportunities

The legal cannabis industry has emerged from its recent slump and appears ready for massive growth in the near future. It’s been a rough ride getting here, and companies who have weathered the storm now have the opportunity to turn things around.

But what do piecemeal legalization and broad investor pullback mean for businesses? And how should these trends impact the decisions being made right now?

In this article, we’re exploring the case for optimism, pointing out some of the pitfalls, and recommending the best way forward. 

The case for cautious optimism: signs of recovery and opportunities for the future

The US legal cannabis industry took a beating in 2022, with the largest publicly-traded companies reporting collective losses of $550 million. But analysts are now expecting a rebound, with sales projected to increase to $50.7 billion by 2028.

This growth is expected to come largely from the piecemeal, state-by-state legalization that is now taking hold. New states that are now beginning to allow sales: Connecticut and Missouri for adult-use, and Mississippi for medical use. It’s likely that as these regulatory changes open new markets, the industry will eventually right-size itself relative to the level of demand. That could allow cannabis-lifestyle and therapeutic businesses to expand their operations into newly legal markets.

Opportunities for biomedical industries

It’s not just recreational or lifestyle use that benefits from a more relaxed regulatory environment. It also creates more opportunity for biomedical and pharmaceutical companies to investigate cannabis for therapeutic applications. Pfizer’s acquisition of Arena Pharmaceuticals was just one major signal of this increasing interest in cannabinoids. 

Cannabinoids are already used in a wide variety of therapeutic applications spanning pain management, immune diseases and improving patient outcomes in chemo-therapy. Legalization is making it easier for researchers to access and work with cannabis, even though at the federal level, significant obstacles remain. 

But overall, the industry looks ripe for product innovation and accelerated growth. 

Why the cannabis market isn’t fully out of the woods just yet

The legal cannabis industry has to reckon with the same broad investor pullback affecting almost every other industry. Investors are – understandably – demanding more proof of future profitability, which may be especially difficult for operators in mature markets where prices have leveled out. In addition, inflation and rising costs will continue to impact the bottom line. 

With funding and M&A on a downward trajectory that shows no signs of reversing anytime soon, the industry’s best chance lies in strategic partnerships. 

Better together: why partnerships are so crucial for cannabis companies

Partnerships have the potential to act as a force multiplier to differentiate companies, expand their reach, and navigate complex regulatory challenges. 2023 has already been hailed the “Year of Partnerships” by industry leaders due to the recognition that the industry will need to pull together to overcome challenges and make the most of the potential the market is showing. 

Partnering for compliant distribution: MSOs and co-manufacturers 

As we mentioned earlier, legalization is rolling out state by state, but federal prohibitions still make interstate commerce illegal. This is where Multi-State Operators (MSOs) play a pivotal role, enabling companies to expand their presence to other states, without running afoul of federal rules. 

 Co-manufacturing partnerships offer another way for companies to expand without overextending budgets. These deals enable companies to pool resources to enter new territories or start new product lines. This can also help to avoid the costs of setting up new manufacturing operations, leveraging partners’ facilities instead.

Accelerating biotech research through partnerships with authorized manufacturers

Partnerships are also enabling progress in biotech research and the creation of cannabis-based medicines. The recent partnership between biotech company Bioharvest Sciences and Royal Emerald Pharmaceuticals is one example of this type of partnership. Through this alliance, Bioharvest Sciences can more easily sell to research institutions federally, leveraging Royal Emerald Pharmaceutical’s ability to legally produce marijuana in the United States.  

Cut through the noise with AI-enabled insights

To succeed in an increasingly saturated market, cannabis companies will need to take a hard look at operational efficiency, and cut costs aggressively. But even more importantly, they will need to find ways to know what is happening in the market, moment to moment, in order to move in on opportunities as they arise.

Similari empowers companies with the always-on market intelligence they’ll need to make agile decisions in a constantly evolving commercial landscape. Through its next-generation insights mechanism, Similari gives researchers and innovators real-time updates on competitors, technological advances and critical changes to the regulatory environment, adapting to users’ needs and preferences over time. 

Schedule your very own demo with our team to see how Similari brings order out of chaos, with timely, reliable insights. 

Breaking the Bank to Save Lives: the Burgeoning Costs of Drug Development

It’s no secret that drug development costs are ballooning, while the number of products that actually make it market is decreasing. This phenomenon has become a kind of truism in the life sciences – so much so that it has spawned its own “law”. The so-called “Eroom’s Law” states that, adjusted for inflation, the cost of drug production doubles every 9 years.

That may or may not be technically true. But what’s undeniable is that managing the costs of drug development is now a critical priority for the pharmaceutical industry. Advanced AI holds the promise of shortening the timeframe and reducing the overall costs of drug production through in silico modeling. But a long process of research and ideation has to take place before an asset reaches that stage. 

And traditionally, that process has been resource-intensive, both in terms of cost and personnel. In this article, we’re taking a look at how this stage can be optimized and accelerated with AI-enabled search capacities that augment human research teams. By saving time and cost here, companies can move from white space to asset more quickly and more confidently than ever before. 

But what’s really new? Drug development has never been cheap

The high costs of drug development are nothing new. Neither are the factors that go into it: extensive research and development, clinical trials, regulatory approval, and marketing. Drug development has also always required investment in infrastructure and personnel. Back in 2021, the average cost of developing a single asset was estimated at just over $2 billion

But while the fundamental inputs that go into that cost haven’t changed, their relationship to market conditions have. 

Cost up, ROI down: the conundrum for Pharma

According to a Deloitte study, 2023 is putting drug developers between the rock of higher development costs and the hard place of lower sales forecasts. In 2022, costs rose to pre-pandemic levels, with sales dropping almost as low as 2019. Looking beyond 2023, pharmaceutical companies in the United States will also need to be agile enough to adjust their clinical and commercial strategies as the Inflation Reduction Act changes the commercial landscape for prescription drugs.

The perfect storm of scarce talent and tight budgets

Another factor complicating the question of cost is the prevailing state of the market. The economic outlook for 2023 is bleak, and companies across the board are being forced to trim the fat, and find ways to do much more, sometimes with much less. At the same time, employees still hold the upper hand in a historically tight labor market. Skill shortages are being felt everywhere from high-tech to marketing. 

For business leaders, the problem is made even more insoluble by global employee mobility and remote work: highly skilled professionals are able to be selective about who they work for, and they’re confident enough to strike out on their own in search of better conditions. 

Managing costs with artificial intelligence 

These challenges underscore the importance of leveraging AI solutions to manage costs, optimize processes, and make up for personnel shortages. Here’s how AI-enabled research impacts every stage in the development process.

Hit identification: automatically sorting the hits from the misses

The average cumulative cost for this initial stage is around $102 million. AI can help to control this cost by analyzing large datasets of biological and chemical data before it reaches human researchers. They’re then able to focus their efforts on only the most promising targets. This can radically enhance high-throughput screening, by accurately picking the best compounds for analysis.

Progressing hits to leads and optimizing them

Tools like Similari make it possible for researchers to instantly survey the landscape and technical literature to accurately assess the viability of leads. By automating away the bulk of their otherwise manual search time, these tools accelerate the hit-to-lead and lead-optimization phases. And by leveraging proactive analytics, researchers can respond to emergent threats or new findings, to adjust their strategy accordingly. 

Streamlining preclinical development & clinical trials

Always-on market surveillance allows researchers to “fail early” (and cheaply) by knowing the results of prior research. With the volume of clinical data growing exponentially year on year, AI offers a way for human teams to cut through the noise and identify salient insights that can make or break their innovation initiatives. 

Streamlining approvals: the last hurdle 

By the time an asset reaches the approval stage, the cumulative cost will have reached around $3 billion. AI has a crucial role to play in managing these ballooning costs, before, and during the approvals phase.

Before an asset gets here, AI like Similari can help reduce the costs of regulatory approval by predicting the likelihood of approval and identifying potential safety issues early on. This helps companies make better decisions about which drugs to invest in and reduce the risk of costly regulatory failures. 

Companies can also leverage market intelligence to identify potential partners with the requisite expertise and experience to achieve compliance. By entering licensing or partnership agreements with these (usually larger) entities, startups in particular stand to gain a great deal when it comes to expedited approval and go-to-market. 

Breaking free of Eroom’s Law starts with data

The way out of the bind is to improve the quality and the quantity of data available to decision makers. Similari puts thousands of clinical trials, along with millions of patents, articles and press releases at their fingertips. R&D professionals and innovation leaders are leveraging Similari’s next-generation AI capabilities to accelerate drug development, and improve outcomes with accurate and up-to-date insights, moment to moment. 

Using NLP to Address Data Overload in the Life Sciences Industry

Globally, we’re producing more data than ever before. But the events that catalyzed and accelerated that historic increase caught us by surprise, so the methods that we use to handle it are still catching up. None of this is news to anybody, but it’s less obvious why this is a problem.

After all, data is the lifeblood of innovation, and the better informed we are, the higher the quality of our decision making. But if you don’t have the right tools to organize that data and extract insights from it, it’s very much a case of “water, water, everywhere, but not a drop to drink.”

Data overload: too much of a good thing?

What we’ve described is the unfortunate situation that many life sciences businesses find themselves in, and now it has a name: data overload. Companies that rely on old, manual methods, may find themselves overburdened by the volume and velocity of data. So how should companies be leveraging big data, without triggering data overload? 

The answer lies in AI technologies like Natural Language Processing (NLP), that enable teams to digest huge amounts of information, quickly, and distill actionable insights from the data stream. 

In this article, we’re going to explore the key problems that data overload causes, and how solutions like Similari leverage AI and NLP to address these challenges. 

Data overload woes for R&D and innovation teams

Data overload happens when there is too much data to effectively process, analyze, and make decisions from. In the age of “Big Data”, companies urgently need a way to sift this information before they can use it. Ultimately, it’s not about how much data you have, but how you use it to achieve your goals.

Back in 2001, Doug Laney defined Big Data using three essential characteristics:

Volume: the amount of data that needs to be processed and stored

Velocity: the rate of real-time data creation

Variety: this data can be structured, semi-structured or unstructured

And with Big Data now bigger, faster and more varied than ever before, businesses need to guard against risk factors like these:

Inefficiency

More clinical trials, publications and drug patents are good news in a sense: they increase the amount of information available. But the sheer volume of available data makes it difficult for businesses to process it, let alone extract insights by selecting the most relevant and actionable information.

This inefficiency arises because traditional search methods are manual and reactive (or historical). They involve sifting through the results of events that have already taken place. But in a dynamic environment (remember: big data is generated in real time), this isn’t enough. What do you do when one set of scientific results supersedes a previous one? AI resolves this problem neatly – we’ll get to that a little later. 

Missed opportunities

In the face of increased volume and velocity of data, companies that rely on manual or database search methods inevitably have to choose between limiting the scope or the depth of their research. When your focus is too narrow, or too shallow, you risk missing out on key market trends and opportunities. But the opposite problem is equally dangerous: choice paralysis is a very real problem, as articulated in the so-called Hick’s Law: an excess of options can actually slow down decision-making. This is crippling for humans, but not for AI (more on this later). 

Poor decisions

Perhaps worse than missed opportunities and decision paralysis, data overload can lead companies to take decisive steps – but in the wrong direction. Fixating on a single data point, or misunderstanding facts in their particular context, can lead to inaccurate conclusions.

Addressing data overload with Natural Language Processing

NLP is an interdisciplinary field that brings together artificial intelligence, linguistics and computer science. It focuses on the interaction between computers and human language. NLP techniques are used to analyze, understand, and generate human language in a way that AI can understand and process. If it’s sophisticated enough, NLP can read and comprehend written texts, extracting the most salient points and distilling insights for the human decision-maker using the system.

NLP alleviates the burden of data overload by automating away mundane, time consuming workflows, allowing data intelligence professionals to focus on making sound decisions. Crucially, it also replaces reactive search with proactive insight: when facts change, an AI-powered system can register the change almost instantaneously. As we discussed earlier, that overcomes one of the most profound weaknesses of traditional methods. 

Key use cases for NLP in the life sciences

Literature mining: R&D teams can use NLP techniques to extract information from large volumes of scientific literature, such as research papers and patents. This can be used to identify new drug targets, understand disease mechanisms, and track the progress of scientific research.

Clinical trial management: NLP can be used to extract information from clinical trial protocols and reports, such as inclusion and exclusion criteria, adverse event reports, and treatment efficacy. This information can be used to select appropriate sites and improve the design of clinical trials. It can also help companies avoid costly and redundant dead ends. 

Drug discovery: NLP can be used to extract information about chemical compounds, proteins and genes to identify new drug targets, and better understand the mechanism of action of existing drugs. And it can get this all done fast – and more accurately – than any human could, by automating key components of the process:

  • Surveying biomedical literature for specific genes related to therapeutic outcomes
  • Identifying white spaces for specific disease targets
  • Searching patent data concerning specific technologies

Further applications for NLP in life sciences research and innovation 

Executive buy-in: R&D and innovation can cut through the noise of data overload and find the precise insights that bolster the business case for each new initiative. Using this information, they can motivate resource allocation to the C Suite based on sound, up-to-date insights. 

Build partnerships: Using NLP techniques, business development and partnership teams can parse scientific literature and extract key information from these documents, such as the names of researchers, institutions, and technologies. All of this data can be used to identify potential partnerships and opportunities for collaboration. 

From data overload to data motherlode: how Similari harmonizes data to supercharge innovation

The proliferation of data isn’t going to slow down – in fact, we can expect to see over 180 zetabytes in 2025, as the curve gets steeper. Next generation AI technologies make it possible to keep up, by translating huge swathes of data into graphs and presentations that make sense for the humans who use them.

Similari empowers scientists, business leaders and data professionals to harmonize data, reveal insights and make the sound, innovative decisions that will shape the future of life sciences.  

Get in touch with our team to learn more, or schedule a demo to see exactly what Similiari can do.