HIStalk Quotes
An Exclusive Interview with Robert Connely, President and CEO of Novo Innovations
posted 08/16/2006
I'd been warned by a couple of folks that Robert Connely of Novo Innovations is a high-bandwidth guy. That advice was about as useful as a tornado warning - just because you know it's coming doesn't mean you can prepare yourself adequately. I cautioned him upfront to take a breath every now and then so I could stay caught up. Thirty seconds later, I was already hopelessly behind, even though he was puttering along at about half his normal speed. He's a tidal wave of enthusiasm, technical tidbits, industry criticism, wickedly humorous asides, and juicy hints about the product he's built at Novo. I found myself so fascinated by what he was saying that I sometimes forgot to keep taking notes.
My impression is that what his tiny company has developed is big, with even more potential healthcare uses than those he mentions here. If Robert is right, legacy applications will be taking a backseat role to intelligent agents that will do incredibly useful real-time stuff with data that otherwise would simply get hidden away in a database, not much more useful to a human than the paper it replaced. Beyond the theoretical, the company has grown like crazy, has a perfect KLAS score (all green lights), and a vocally loyal customer base.
Thanks to Robert Connely for doing the HIStalk interview. If you dream of radical technology changes in the stodgy healthcare IT industry, this is a company to watch.
Note: this interview was planned, but not done, before Novo decided to sponsor HIStalk, including formulation of all questions except the last one, which originally read "Do you read HIStalk? While I appreciate their support, it had no effect on their selection to be interviewed or the questions I asked.
Some background, please, on yourself and Novo Innovations.
We started Novo in April of 2003. My partner, Alok Mathur, and I came together in 1999 at McKesson. He was the chief architect and I was the VP of web technology. In December of 1999, I pitched the idea of a portal we had built at Memorial Savannah to Graham King. We started the Web Solutions Office and did Horizon WP Physician Portal, Horizon WP Foundation, and Horizon WP Community Portal. The clinical group hated that we did this project, so they threw in WP - Web Powered - to differentiate from their stuff (laughs.)
We rolled it out to great acclaim, including a Wall Street Journal article. In fact, they used the same words Microsoft used in their Azyxxi article (laughs.) We went from $0 to $200 million in sales. We also invented Horizon Pre-Services Manager, now called Horizon Outreach.
We decided the next place we wanted to work was the rising ambulatory market. At McKesson, we got to work with a lot of top hospital systems, who apparently loved working with us. We saw the problem with EMR adoption and believed that no vendor or current technology could solve the problem. McKesson had bought Abaton for 100-plus million dollars as their ambulatory solution and there wasn’t a place for us. So, we left in April 2003. The top execs loved us, the rank and file loved us, but middle management hated us. We stepped on too many toes – we just wanted to get a good product out to customers. Anything outside of methodical planning is frowned upon in big companies.
We served out our post-employment restrictions by not thinking for six months or talking to employees or customers (laughs.) Then we built our first agents and the Novo grid. We told McKesson what we were doing, thinking we might come back someday, and started a pilot at Northside Hospital in Atlanta. They signed in December 2003 and they went live in January 2004 between Northside and Atlanta Women’s Specialist, AWS, a very busy eight-doctor OB/GYN practice. We were automatically delivering labs, radiology, pathology, transcriptions, surgical notes, discharge summaries and the like, and inserting it directly into an A4 Healthmatics EMR.
We think that software agents like we ours are the next generation of middleware. An agent does its job and can be deployed anywhere. You’re using agents when you do a Google search or run an automatic antivirus update. We invented agents that act like interface agents to be deployed to hospitals or to a physician practice. The grid is an object oriented system that can replicate an object to multiple agents and keep it in sync across all locations. That’s how we do a push transport. We’ve patented that. Northside went live on that piece in February 2004 and turned off all the faxes to AWS in March.
We’ve grown this far by word of mouth among people we know in the industry. You are our marketing beginning (laughs.) We’re in about 120 hospitals and we have a huge pipeline that amazes me. By the end of the year, we should have four of the top 20 hospital chains under contract and many community hospitals as well. Some people think that’s a lot for an eight-employee company, but we’re like vampires … we never sleep (laughs.)
What was it like being McKesson’s mad scientist?
It was kind of cool. I loved working at McKesson. I was in the best of all worlds, both entrepreneur and corporate. When customers were brought in to the corporate office for the dog and pony show, I was the guy to get them excited about the future. That was a great thing.
We always had resources that loved to work with us. We’d just talk with people on the side, telling them, “Hey, you leave at 5:00, just stick around until midnight and we can get this thing done,” and developers would work with us. We didn’t do a project plan or resource allocation. I’m sure McKesson hated it – they were big in CMM and Six Sigma. I was told, “We have enough innovation, we just need execution.” I agreed.
I was sort of a maverick, I guess. You get in a position where you piss off a lot of people just by being successful, like maybe what we were doing cast a bad light on everyone else. That’s why we left.
Do you secretly hope they’ll buy your company someday?
No. That will never happen. In fact, I was told that I’d never work for McKesson as long as the current management is in place. I guess creating a $200 million project for them made them mad for some reason (laughs.)
Our product competes directly with their business model. They’re a late market company. They let small companies like us innovate and then they buy the winner. When I was there, I bought the Kool-Aid that McKesson liked to innovate. Not only do they not like to, they can’t. They can’t throw out the investments they’ve already made. If Novo goes there, I’ll probably be gone from here, too.
How does the Novo grid work for connecting physician EMRs to hospital systems?
The grid is an object distribution system. There’s a small piece of software that includes a Tomcat application server and other services we’ve written. Our metaphor is the Martian rovers because, to a hospital, a physician practice might as well be on Mars. They can’t manage or do anything there. The grid node is the platform that includes the application server, communications modules, a tight security module, data storage and remote management components. Those are the nodes that go to the endpoints.
There’s another node that lives on the Internet called a Rendezvous. Nodes are programmed to check with the Rendezvous for a payload, like going to the post office to see if anything is in your mailbox, or to upload a payload for another node. What they exchange is interesting. We’ve patented a method of exchanging structured, secured objects between participants on the grid. It can be structured around a document, an HL7 message, a PHR, or almost anything. Once I create an object, I can determine where else I want that object to live. For example, I may want a hospital lab result object to live in a physician’s practice and the performing lab like a hospital. Or, maybe if the object is about a patient, when a prescription is added, I want to update records everywhere the patient’s record may exist.
The grid is the communications underpinning. The nodes are platforms for intelligent agents, which are the interfacing components. Agents are autonomous, can interact with their environment and run without attention. They can communicate with other agents. We program these agents for specific behaviors, basically teaching them. Databases become edge things. Agents react to and move data well before it gets buried in the database.
At Northside, they were mostly HL7 driven, attached to an interface engine port. We capture all of the stuff they would normally fax. The agent examines the data, parses it into Java objects, looks into a definition object to determine whether the physician wants that data, such as particular lab or patient type, for example. The agent will, if told, transform the message. For example, it can insert text for the performing location, lab director, and phone number into the HL7. It may have to truncate disconnected segments or move things around to standardize. We act as a secondary interface engine.
Once that’s done, the agent creates an object, adds the HL7 message, encrypts it, and sends it over the grid to the receiving doctor. The receiving agent decrypts it and decides what to do with it, like inserting it into the EMR by dropping it into a folder or posting it to a TCP socket. The agent can even convert hospital information into a PDF file that’s named according to the receiving practice’s standard convention, ready to put into a folder. We can convert it to XML or do just about anything else.
We have an interesting model. We’ve realized over dozens of installs that our real value is that we work with the EMR vendors. Hospitals don’t want to deal with a dozen EMR vendors in interfacing. That’s far beyond the standards thing that everyone thinks will solve world hunger (laughs.) We open up our intellectual property. Anything we’ve developed to connect to these systems is available to our customers. Once Allscripts was done, for instance, we can interface any other Allscripts customer. We run tests and make tuning changes to provide a specific use case for that EMR vendor in that community. It’s 80% repeatable and the last 20% is where we do our business.
Once the hospital connects to a system, all the EMRs are now covered. It’s the only interface they need. They just need to turn on the feed to us. At Lakeland in Michigan, they signed on Friday and the system was live the next Thursday. Their total staff time from start to finish was 14 hours. Our biggest complaint is that it’s too good to be true (laughs.)
What kind of workflow is created from what type of events?
Everything IHE can throw at it, that’s for damned sure (laughs.) We started with report distribution. At Northside, we’re working on prenatal record transfers. Those are faxed to the hospital time after time, but still 37% of the time, hospitals don’t have the current record. Amazing. The agent looks for a change in the record in the practice and synchronizes it with the agent in the hospital. No matter where the record was last updated, the birth center always has the most current copy. Any change to the record by anybody is automatically reflected at every other location. It’s information where it’s needed before it’s needed (laughs.) That’s our model.
Next is orders, which is a fairly simple thing to do other than the workflow of the hospital. We grab the order from the EMR, send it to the hospital and later receive the result, matching them up. MD to MD referrals are coming. If you don’t have an EMR, you can still communicate with MDs in the community, creating private referral groups on the grid.
I can erase ¾ of the lines in the IHE diagram with our technology. We’re clearly focused on the workflow aspects, not just the RHIO stuff.
You don’t seem to believe the RHIO and IHE approaches are the right ones.
They’re applying yesterday’s technology to tomorrow’s problems. We’re applying tomorrow’s technologies to today’s problems.
We’re on two distinct but converging paths with RHIOs. Most RHIOs are clearly business models, trying to solve governance, legal, and financial issues. All the efforts going on through IHE are simply experimental, designed to give RHIOs enough experience to solve their own challenges. We’re working on the next generation of technology that will completely change the RHIO concept. Instead of a billion dollar RHIO, let’s say it costs $30 a month per doc. It will change the way we think and do.
I don’t think the way we see RHIOs today will last long. Big things happen organically. The next generation of technology will suddenly come into play to solve problems and you’ll realize the technologies are perfect to solve RHIO challenges. For example, take a hospital lab result. It’s no effort to have the agent depersonalize it and send it to a CDC database or a biosurveillance repository. We think RHIOs will establish sustainable business value using these next-generation workflow systems, achieving their current goals as a byproduct of the initial workflow transactions.
Because they chummed up the water with all these dollars, they threw everyone off. Vendors are rebranding 10-year-old products as the next RHIO thing. These newer technologies change the cost and effort paradigms so much
How do you convince hospitals and physicians that they need more than just a portal, which itself is a new technology for many of them?
If you’re a doctor with an EMR, you don’t need a portal. In fact, you hate the concept. You want to see the data in your own system. You want data in your patient charts. Going and getting it doesn’t do anything. You’ll see docs saying that whichever hospital gets their data into their own EMR gets their business.
Hospitals lose business to reference labs. A portal can’t help you there. Your physicians are bringing on EMRs and you need another way to get data to them.
How do you handle the patient identifier issue?
That’s the most ridiculously overwrought issue in the world. If I have a national identifier, I still can’t use it exclusively. It could be transposed or stolen. You always need to reconcile the elements you have. If I get a 100% demographic match, I’ll put the information in a worklist or directly into the chart. If not, I’ll post it to an exception list. Once it’s reconciled, we’ll remember that for next time. A single identifier improves the reconciliation from 90% to 95%. It’s not needed for integration.
Do you see Stark changes increasing EMR deployment and therefore increasing the need for integration with the sponsoring hospital’s systems?
I’m not sure how hospitals buying EMRs for doctors will work. It’s never been very successful and it hasn’t been tried much except with small practices. We surveyed Northside and looked at how many practices had EMRs. We found that 40% of the big referrers already had an EMR, and that group made up a large percentage of Northside’s business. Where they’re talking about buying EMRs for doctors is the 1-2 doc practices that don’t give you a lot of business anyway. I don’t see a lot of hospitals doing that today.
Who are your competitors and how does your approach compare with theirs?
Our biggest competitor is customers doing it themselves, which dies down quickly when they realize how much expertise they need. Medicity, Healthvision, and interface engine vendors compete with us. We’re sort of the next generation of Quovadx and SeeBeyond, maybe Iatric. We don’t see a lot from the RHIO vendors because they’ve decided to chase that rabbit down another hole. Our customers say no one does it like us and we know nobody’s doing it for the price we offer.
What’s your opinion of RelayHealth?
There are a number of companies like them out there starting with patient-to-physician communication. I don’t know what McKesson is doing, maybe parlaying it into some grand RHIO scheme. All the EMR vendors already have that capability with their EMR product. I guess they’re going for practices that don’t have EMRs. I think they’re going for an approach that just connects their client to an EMR like secure e-mail.
How do you price the service?
Pay us a monthly rate for the technology, support, and implementation. It starts as low as $1250 per month for a hospital and grows incrementally as usage increases. We originally started at $5,000 per node licensing, but decided that would scale out of sight if you had a lot of customers. We leveled out, starting small and keeping it about the same as fax, maybe less. We don’t break anybody’s budget. We’re daring the market to come down to us. Forget the million-dollar deal unless you’re 15 hospitals (laughs.) It’s kind of a Trojan horse effect, trying to get it out there and get new agents built. Join our service and you get the same integration team that the biggest healthcare systems get.
You must be pretty happy with your KLAS scores. I don’t think I’ve ever seen a stoplight report in which every metric is all green. I noticed this quote from a customer: “This is really a cool tool. It is also inexpensive and very reasonable. The product is of high quality, and we have had no problems with it whatsoever.” Can you keep making customers this happy as you grow?
I don’t think our customers are happy just because we’re small and can deal with the workload. We’ve got a low bar to clear compared to what they’re used to (laughs.) We wouldn’t do it if our culture was built around how much money we can make or how methodical we can be. Our sole focus is the customer.
Novo is doing pretty well for just getting started and little in the way of marketing. What’s the next step?
I come to you (laughs.) I have a word of mouth marketing engine in our customers. One hospital actually flew their person to a prospect’s site to present on our behalf. That’s the kind of customers we have. That’s our sales and marketing. We get calls. We don’t solicit other than a little bit of e-mailing, other than this big investment in a little blog called HIStalk (laughs.)
We’ve got some very senior, qualified, and passionate people. We have personal contacts, reputation, and word of mouth. We don’t present at HIMSS. They turned us down this year, twice. We’re taking the guerrilla approach. I think I want to stick to that way. I don’t want to be associated with those other tools. I’ll sacrifice massive sales to control my growth and keep my passion. We’re looking ahead at a different way of approaching it. It’s like that Smokey and the Bandit movie where the Italian guy rips the rearview mirror off his Ferrari – what’s behind me isn’t important.
What’s the future look like?
These new middleware technologies will do the things people are trying to do directly with EMRs and with RHIOs. For example, one of the next agents we’re building is a patient agent. I as a person could tell my agent to work with other agents in my community. It might send information to my PHR or to someone involved in my care. We believe there’s an emerging model where people will instruct an agent on what to do with their data before it’s put into a database. The cost is negligible. I don’t hear anyone talking about that concept.
We didn’t start out by building agents and grids, we started out trying to solve problems with Northside Hospital, then tried to see if others had come up with something similar. We figured we weren’t smart enough to have figured this agent thing out on our own. Agent technologies have been around since 1991, originally called knowbots. DARPA’s been working on it for six years. They have a huge problem with information overloading military commanders. They decide to put intelligence out there, to let agents make decisions about data in real time. Europe’s also quite advanced in doing that.
I think we’re the first to bring it to US healthcare. We’re opportunists (laughs.) We realized we’d built an intelligent agent. We’re unique in how we use grids to allow them to communicate. I can’t get information if I don’t know it exists. The grid was done so an agent could monitor the state of an object and react to it, effectively knowing what happens elsewhere.
Couldn’t the same technology be used to add a far superior clinical decision support capability to current inpatient clinical systems without changing the underlying application?
I can collect information and scrub it down to decide what’s important. That kind of work should be done outside of the application. In the future, agents will do this work because it separates people from the data and turns data into knowledge. It’s incredible that we’re still using approaches that are 20 years old, like monolithic forms of clinical decision support.
Legacy products will be the edge pieces, doing their job so the admissions clerk has the same green screen to admit a patient. Knowledge workers will be using the new tools that agents will facilitate. Those vendors’ vision of being the center of the universe will quickly die out. There is no center of the universe. Agents are no different, except they can collaborate.
This is more like city architecture than a putting up a building. HIS systems were the pylons in the ground, although they won’t like being called that. There will be a new level of these agents that will do things in real time that have never been done. We’re getting to the ability to build an application in runtime as it’s needed, instead of application architects trying to guess the future and being wrong a lot of the time.
People will ultimately quit buying HIS systems. They’ll say, “I can already admit a patient on the green screen application. I want to do something new.” Why would anyone replace something to gain functionality that’s not even available in the new product? You need to build on top of what you have. My goal in life is to say, “Open up your eyes, people. If you use Google or NAV, you’re using this stuff today. There’s no reason to upgrade or replace.”
Novo might be like the 100 miles per gallon car that Detroit was rumored to have bought and buried. If you threaten the status quo, maybe someone will just buy and bury you.
Vendors don’t know we exist and we don’t depend on them. They can’t do what we do. Computing science is moving like a tidal wave. Can they kill Novo? Sure. We’re a gnat, but you can’t stop a tidal wave. It’s coming.
Some vendors have expressed interest, but most don’t know about us. That fascinates us. They’re not looking at us, they have their head in the sand and are worried about their next version and RHIOs. No one wants to buy a tool that integrates applications. They’re busy looking for the next EHR. People look at us as incidental. For the most part, we’re in the guerrilla war. We’re not self promoters. We try to be more fun guys.
We’ve never done a press release. We work with customers and they tell other hospitals. Once we’re in a city, the next customer is always the other hospital in that city (laughs.)
Who do you respect in the industry?
I don’t know that many people. Halamka’s been through the wars. I respect people that you don’t ever hear about. Dr. James Morrow of North Fulton Family Medicine won the HIMSS Davies in 2005. Here’s a doctor who’s actually on Brailer’s board and still sees 100 patients a week. He’s still plugged in to real medicine. George Conklin, CIO of CHRISTUS. He’s focused on how to make it work, not the ups and downs of technologies. Brailer’s a good guy, although it seems that he got out just as the going got tough. My hat’s off to the real people in the frontline trenches of the interoperability wars.
Why did you sponsor HIStalk?
You’re touching the people that we want to be in touch with. We don’t care if McKesson and other vendors know about us. The people you have in your readership community, we want to start getting the word out that there’s new stuff out there that changes your dynamic. It’s an inflection point moment. Maybe we can change this stuff. My VP of sales said, “This is the best blog on the Internet and this guy’s the most respected.” Your people would appreciate that we’re out there, doing it, making it happen for 2 ½ years before talking about it. You have an interesting place to tap into the market. You’ve got it, man.
comments (7)
1. Noodles Panini left...
08/16/2006 6:59 pm
This is fascinating! Thanks so much for publishing it. I've heard only
the best things about Novo, and I hope that Robert Connely and his team
scare the &*%$ out of the big vendors. He is certainly right about one
thing; this is the most respectable health care blog in existence today,
and you have a dedicated and unique audience (at least as far as I know).
Both you and Novo are in for a big payday sometime soon.
2. curious left...
08/17/2006 2:07 am
Same old SOA in a new bottle..
But still you have gone one step further from paper to having deliverable
agents.
Good luck.
3. X left...
08/17/2006 8:17 am
Excellent interview! I'd advise Robert to have his food tasted before he
eats it. if he keeps this up, the big vendors wil be after him.
4. money_matters left...
08/17/2006 9:44 am
WOW!!! Now there's a curve ball for the 300lbs. players out there. It’s
almost intimidating to know such big things are coming from these guys. I
may need to rally my company’s innovative juices just to stay in the game.
However, in all actuality, all I want to know is how do I get my hands on
some Novo stock so I can go ahead and quit my job. Great interview Novo!
5. Matthew Holt left...
08/17/2006 12:29 pm
This may sound fascinating, but I've read the interview and I don't
understand what's different between this and Teramedica, SOA, Azzyzzi
(however it's spelt) or any of the other new interface tools. On the other
hand if this is a cheap tool that replaces all interfaces with no
consulting needed, they won't need marketing, they'll just need
bodyguards--not so much from the vendors as from the consulting shops.
6. Dave Dillehunt left...
08/17/2006 3:05 pm
I've worked with Robert a couple of times (once with McKesson and once
before that). He's an incredible guy and a straight shooter. That was
part of McKesson's frustration with him, especially when he set the value
of the technology at one level, and they set it much much higher. But this
guy can make things happen. I'd continue to keep your eyes on his company
if I were you.
7. Dale Hunscher left...
08/21/2006 12:08 pm :: http://hunscher.typepad.com/futurehit/
His comment on patient identifiers being an overwrought issue: What a
breath of fresh air! If the ONCHIT people would just say something like
that out loud, I'd have great confidence in their efforts.
