This is on the homepage using 146 words.

“Bims: Biomed News” may appear like just another current awareness system. We think of it as an expertise sharing system. That’s because we request our users to make weekly selections and the selections are publicly available. This is unprecedented in the biomedical domain.

The team is Thomas Krichel and Gavin McStay. Thomas’ claim to fame is being the founder of the RePEc digital library. He is a former university faculty in the USA and the UK. He now works as a digital librarian in Novosibirsk, Russia. Gavin is the project director. He holds a PhD in Biochemistry from the University of Bristol. He held academic positions in the USA and is an associate professor at Staffordshire University.

In 1998 Thomas created “NEP: New Economics Papers” to improve current awareness about economics working papers in RePEc. Bims is an implementation of the same idea in the biomedical area using PubMed. It uses much of the same software. Thomas’ idea was to reach beyond the economics domain. Thomas picked the biomedical domain because it is large and fast moving. Thomas looked for four years for a partner before stumbling across Gavin at a networking event in New York City.

Thomas created the domain. Gavin has subject expertise.

At this time, we aim to serve academic users. We are not just solving a single problem. We are addressing two user needs at the same time. (1) They need to know about the latest papers in their area. (2) They need to be recognized as subject experts.

In the longer run, we will also serve users that do not have the second need. In fact, if they are patients, they may want to stay anonymous.

We intend to use the funds for renting a server and to pay for Thomas’ labour for two years. Thomas will not undertake any other paid work for the two years, starting from the commencement of the award. This is realistic because Thomas is not undertaking paid work at this time. He will prepare a detailed work plan ahead of the interview.

Bims will be “funded by digital-science.com” for two years. That will be on every report issue and all pages on the web site. After that, we will mention the catalyst grant for ten years on all the web pages. After that, we will mention it in the history section. Thomas will write monthly progress reports.

Much the product operates behind an interface for registered users, i.e. selectors, only. They receive a list of PubMed papers, linked through an e-mail, sorted by likelihood of inclusion. They make two binary selections on the papers. In the first screen they select the papers that are relevant to the report’s subject or are at least related to it. The results of this screen inform machine learning. In the second selection screen, users select from the selections of the first screen the papers that will go into the public report. The differences between the public and private selections are the “secret sauce” the selectors bring to their report. Finally, selectors have an optional sorting screen to rank papers. The report issue is sent as an e-mail to the selector. It also appears on the reports page.

The product is very robust. At this time, we think of it as “good enough”. Hardcore scientists tell us that it is great. However, if you work on topics with little jargon, it is a bit trickier. Thus, there is room for improvement.

We have 19 users. They curate 24 reports.

We do not intend to charge either selectors or readers. We see only two ways to raise funds. One is by sponsored advertising. The other is by running the same technology on a software-as-a-service business model for private reports. Both sources will remain limited. They ought to be enough to sustain the product provided that it operates in a low-cost way.

We want to keep the project as a non-profit entity. This is to shield it from market hazards and provide accessibility to all who may want to use it. We need to keep running it even if it generates no income. However, we realise that it is good to associate it with a for-profit corporation. That corporation will manage the asset when it finally gets substantial traction. We think that will be in about ten years time.

We don’t really know. We think that for current awareness, no product has significant market share. Academics really tend to work in their own ways.

We could fill the entire application with other products and companies that are in the current-awareness dimension of bims. They now all tell you that they use artificial intelligence because that is fancy. In reality, they aim to do some machine learning. However, these systems lack incentives for their users to be good machine teachers. With poor machine teaching, you get poor machine learning. Case in point, Gavin tried Abstream, F1000 Alerts, Google Scholar Alerts, Mendeley, Meta, PubMed alerts and Sparrho. He found all of them lacking. Generally, they get users' input like papers they wrote, citations they made, and papers they interacted with. Then they send out pointers to related papers. Users are overwhelmed with old papers of marginal relevance. That experience—more than any of Thomas’ conviction power—led him to try Thomas’ untested solution.

We only offer recent papers of the past week. We require weekly use. Only precise machine teaching will produce good machine learning. The communication of the results helps to make the extra effort worthwhile. One fine day, being a followed bims selector will be a valuable service item on an academic CV.

For market entrants, our technical infrastructure is not difficult to replicate. But any entrant who would copy us would face similar problems as we have to get users, with unclear prospects of profitability.

We do not know. We think interest in biomedical research is global. Bims has the potential to be used by people in all places with an Internet connection.

Our core market consists of researchers in the biomedical sciences. We have just recruited our first selector who is a journalist. Doctors, nurses, health information professionals, health policy makers, sufferers from chronic diseases … basically anybody who can benefit from regular scrutiny of PubMed is a potential customer.

Customer acquisition is our biggest challenge. We can’t fake it by inflating customer counts. We can’t demonstrate our product on the spot. It needs a few weeks of training data for it to reveal its magic. We ask our users to essentially take on a small unpaid job every week. There is no precedence for the type of task they are assumed to perform. Thus, the low uptake to date can entirely be blamed on us being incompetent marketers.

Some parts on the planned work will deal with features designed to help customer acquisition. However, we do not see a magic bullet that will get us to lift off. We will continue to toil on the ground through face-to-face meetings and online contact.

Customer retention is not a problem at all. Our most recent customer loss occurred on 27 October 2018.

We started on 4 February 2017. We use the server that runs NEP. We intend to run indefinitely.

In 2019, Thomas had a job that did not allow adequate time to work on bims. That job is has now finished. We gained 11 users. We lost none.

We do not.

We already had NEP. It validated the market for us, starting 21 years ago.

We expect continuing but initial slow growth. We hope to reach at least 100 users by the time the funding finishes. We do not need a large user base. Even if we only get 1000 users producing weekly reports read by, say, 100000 readers we will have made a significant impact on scholarly communication in the biomedical sciences. We will create a more level playing field between journals. Marginal ideas, that do not use the standard vocabulary that someone would search for, will be much more visible.