software for domain experts - a new business...

Event sponsored by:

SOFTWARE FOR DOMAIN EXPERTS - A NEW BUSINESS MODEL?

October 29, 2015, The Cube, Athens

helping free thinkers work better

Special conference report

How to solve the climate, refugees and terrorism?

A new business model for Greece

Getting the most from 'low code'

Why experts could have better software

Getting investment for a software for experts project

Messaging Financial

aplonCashMessaging

aplonCash Treasury

project Bespoke

Messaging Libraries

toupays wal&t Robusowlalesiarbrile arwofts

our yt tarsump j&onementatiplmi niime cenanntaimur oy oc

e.mti T FISWfor t porSupand 022)SO20I(X MT FISW

aplonCashMessaging

date totoou yow

ze imni& st oT,MT

ASEPand

Management

yghlHi icfieftree vticfeef

paorate orpcSHCAonlap

ated automsents mypa

egntiylisea

Treasury Management System(TMS)

ost cand ent ifor temsysyurryastre

.onatitiniients mypaSH desiovpr an

ctroniecelated bean cat thonutiols

ng tisexithiwated r

project Bespoke development

prope cslfule handle Wnciomfrent opmeldevdel,onementatiplmito&sonatioperng goion

tporups . Our ghlhisiteam n ied llkis hecnTiF , nSeriF

hcTeensurI ,sonutiols us

t ojecprontiep

&yerryvidel

yghlvnSer &

e thng ius

022)(rpang,dilibuagesesm

ormatinsftrang,datiialv.enctesisrpe

egng,isr

and onormati

ntiyliseaorate orpc s

ement anagmbaorate orpc

ng tisexithiwated rng fyiuni,stemsys

our ylalof ement ts.ouncacnkba

hcTeensurI ,sonutiols use lobim&ebWt atesl

.esogilhnotec

e thng ius

Ls tennompoCtmenPayn ondoL W3 S ed tniU,T1E

poomentcmypa.www

Ou

88 ,rtuoes Ccn64 Prid.tL roBm,gdoKin

omconents. | @ entcmypa

r Ou solutions & libraries

notmpro dgsbritghiKnd,aoR

pomentc 2071 44 +:.Tel|

Poareolutions

edg ,

538172 2071

owering

This report is from the first Software for Domain Experts conference held at the The Cube, 8 Kleisovis Str, Athens, on October 29, 2015.To see the conference agenda, and download vidoes and presentations from the conference, visit www.bit.ly/Ath15SFDE

Software for Domain Experts - a new business model?

3

Report written by Karl Jeffery, Editor, Software for Domain Experts, Tel 44 208 150 [email protected]

Conference producersDimitris Lyras and Karl Jeffery

Conference ChairDimitris Lyras

Report LayoutLaura Jones, Very Vermilion Ltd

Cover Art Alexandra McKenzie

Software for Domain Experts Ltdc/o Future Energy Publishing39-41 North Road, LondonN7 9DP, UK

Softwarefordomainexperts.com

Could it be possible to build much better soft-ware for domain experts – and could there bemany business opportunities along the wayfor organisations which employ domain ex-perts, domain experts, software companieslarge and small, and investors?

This is the topic we explored in our Athensconference at The Cube, October 29, 2015,“Software for Domain Experts”.

By ‘domain experts’ we mean someone whois an expert in a particular field or domain,and has a job role where they use this expert-ise as part of an organisational process. So,they could be in anything from fish farmingto running a government department.

Our starting theory is that many of these peo-ple could do far more with software, givencurrent computing capabilities, than they cur-rently do, and modern enterprise (organisa-tional) software does not serve them as wellas it could.

Much enterprise software has been built toprocess transactions or manage ‘workflows’.This is excellent if it is built for users who domanage transactions or have work which fol-lows the same steps every time, for examplemanaging a billing or purchasing process, orapproving mortgage requests in a bank.

But most expert roles are very “messy”, inthat people don’t work in a standardised wayevery time which you can copy into a flow-chart and embed in ‘workflow’ software, orthey gather and analyse data in a very partic-ular way which is usually not compatible withenterprise software.

The ideal software for such an expert mightbe built in a very different way to standard en-terprise software. Its main focus could be pro-viding the expert with information relevant towhat they currently need to do, and providingthem with relevant experiences – for examplea story from an expert who was in a similarsituation, talking about what they did.

The software still needs to make sure theirwork fits with the business processes, but per-haps in a less restrictive, or more goal-centricway, than current enterprise software usuallyallows.This requires software to be made in a very

different way to how software is currentlymade. At the conference we identified severalfactors which might help.

To provide relevant information, the softwarewould need to be built with sophisticatedmodelling, which understands what the expertis currently doing, what they need to know,and what organisational data would servethem best.

To provide 'relevant experiences', perhaps youwould record many people speaking aboutwhat they have done and why in video form.They can talk about what decision they madeand why they made it. This can be much eas-ier for someone else to learn from, and makea judgement about, compared to reading theperson's written text.

The software might be built using ‘low code’software technology, which aims to automatemuch of the process of writing code, and en-ables you to create code which is more robustand requires less testing – thus reducing thecost of creating software and enabling soft-ware to be adapted easily once built.

An enormous amount of modelling would berequired – building models of the businessprocess, building models of the informationan expert requires, building models of howthe software would work together.

All of this could create big business opportu-nities for Greece. Building and maintainingsuch software could be a task better suited tosmall start-up companies, which would eachserve a narrow sector, and so many new com-panies could be created.

Greece could also become a centre of excel-lence for researching and providing the nec-essary modelling skills, which no-oneanywhere in the world has yet figured out howto do.

And meanwhile consider that so many of theworld’s biggest problems might be solved byexperts supported by better software as partof a looser business process. Climate change,counter-terrorism, refugees, cybersecurity,maintaining water and electricity supplies indeveloping countries, providing universalhigh quality education, not to mention corpo-rate challenges such as running complex or-

ganisations, managing complex infrastructure,avoiding accidents, and making sure yourcompany doesn’t run out of cash.

The ‘internet of things’ movement – addingbillions of sensors – will also generate muchmore data – and generate much more require-ment for non- transactional expert work tomake use of it.

Our conference in Athens on October 29th ex-plored all these issues.

Our opening speaker, George Karaplis, a for-mer CFO of Greek telecom company OTE,explained what he saw as the business casearound ‘low code’ and how to build a businesswith it.

Our second speaker, Roger Schank of SocraticArts, one of the world's top experts on artifi-cial intelligence, explained why and how toinclude stories as part of a goal model, allow-ing you to make informed decisions from youravailable options regarding anything from se-lecting your next CEO to the surgeon youwould like to operate on you.

Our third speaker, Aristos Doxiadis, a well-known Greek economist and investor in soft-ware start-ups, gave valuable advice on howto get a start-up software company thriving inthe current environment, in the software fordomain experts space.

This was followed by three speakers from ac-ademia - Vassilis Zafeiris of Athens Univer-sity of Economics and Business (AUEB),Kostas Kontogiannis of the National Techni-cal University of Athens (NTUA), and NikitasAssimakopoulos of the University of Piraeus,who talked about academic work to improvemodelling and develop more flexible enter-prise software structures.

Finally, we had three case studies of Greek (orpartly Greek) software companies in thisspace: Payment Components, which makes‘low code’ tools to make it easier to build fi-nancial transaction software, Sensorflare,which makes tools to make it easier to connect‘internet of things’ devices, and B2B Wave,which makes it easier for small business own-ers to build online stores.

Software for Domain Experts – a new business model?Our first Software for Domain Experts conference in Athens, Greece, on October 29 explored a better way todevelop software for domain experts, and how the business models could work

Special report, Software for Domain Experts - a new business model? October 29 2015




4 Special report, Software for Domain Experts - a new business model? October 29 2015

The introduction of low code tools couldmake a massive change in the way software isbuilt, said Dimitris Lyras, director of shippingcompany Lyras Shipping and software com-pany Ulysses Systems, also chairman and co-producer of the conference, in his openingremarks.

Mr Lyras' company, Ulysses Systems, buildssoftware for shipping and transactions. These“are built on conventional software architec-tures of entity, relationship databases," hesaid.

The software “hasn’t been easy to maintain,"Mr Lyras said. But also it is hard to imagineany enterprise software built using conven-tional means being easy to maintain, espe-cially if it is doing something significant.

As a software company owner, “I know thisfrom paying for it,” Mr Lyras said. “It’s some-thing I’ve looked into deeply."

If software is expensive to build and maintain,it means that many start-up companies strug-gle to get good software made, and manycompanies are reluctant to change softwareonce they have built or purchased it.

All of this could change with the advent of‘low code’ software tools, which promise tobe able to automate the task of building soft-ware. This should make it possible to buildsoftware much faster and cheaper, and thesoftware will be more reliable immediately, soless expensive testing will be required, MrLyras said.

Building such software could be a massiveopportunity for Greek software companiesand software engineers. There are many peo-ple in Greece qualified to work in softwarewho are currently unemployed or underem-ployed.

Mr Lyras gave one business idea, a tool whichcan be used to make business predictions, butalso keep track of what assumptions weremade in the process.

For example, he once asked a financial expertto build a model estimating future demand forshipping services. The most obvious softwareto use for such a task is Microsoft Excel, butExcel does not make it easy to see what as-sumptions were made in building the model,and makes every figure look too exact.

Dimitris Lyras – what low code tools couldmeanThe introduction of low code tools could massively change the financial proposition of building andmaintaining software, said Dimitris Lyras, director of shipping company Lyras Shipping and softwarecompany Ulysses Systems


5



Perhaps we are under-estimating the impactwhich ‘low code’ technology could have on thesoftware industry, said George Karaplis, SeniorAssociate with IG Icon Investments.

Mr Karaplis is a Senior Associate and Invest-ment adviser to I.G. Investments, a private in-vestment and consulting firm. He has served asChairman and board member of Netia (Poland), General Manager in Group Suez’swater division and Chief Financial Officer andvice president of international investments forthe Hellenic Telecom Organization. He workedfor 15 years in Canada and the US in variousmanagement positions.

In recent years there has been something of anacceleration in the use of ‘low code’ tools,which enable the rapid building of applicationswith minimum hard coding, he said.

Forrester Research recently published a studysaying they are mostly used in customer facingapplications at the moment.

Roughly speaking, you build applications bymaking a diagram of your process in your soft-ware, and then the ‘low code’ tool creates thecode automatically.

An analysis by Capgemini found that the aver-age time to build a ‘function point’ in low codeis 2.5 hours, compared to 10.6 hours for Javaand 15.5 hours for C#.

Using ‘low code’ also means that tools can bedeveloped and changed faster, he said. “This isvery important in environments where marketsand customer requirements are often chang-ing.”

Mr Karaplis is an investor in a company whichbuilds software for African markets, which willwork over low bandwidth mobile and satellitecommunications.

The company started using low code tools 3months ago. “The development that took placeis incredible,” he said. “We were fighting for ayear to build an app, and all of a sudden in onemonth it happens.”

Mr Karaplis envisages that software developerscould also develop templates and libraries,which could be re-used in different softwareapplications, speeding up programming time inthe same way.

Consider a shoe manufacturer who has custombuilt sales software for managing transactionswith retail outlets, and the manufacturer startsselling in a new territory where it is customaryfor manufacturers to sell to distributors, whothen resell to retail outlets.

Instead of the software requiring extensivemanual coding work to incorporate the distrib-utor, it would be possible to just upload tem-plates of ‘low code’ and the change in thesoftware would be implemented, he suggested.

In another example, Mr Karaplis was workingfor a large company (as an employee) and wasasked to arrange IT support for the company’s45 subsidiaries around the world. He signed alarge contract with a supplier to provide it, andthe company spent 6 months “going to coun-tries and supposedly building things,” he said.

But then Mr Karaplis found that all the sub-sidiary companies had been developing theirown IT projects on the side, which was feedingdata into the main corporate IT system. The lo-cally built software was proving much more ca-pable at coping with local requirements,including tax and regulations.

But in future, companies might use ‘low code’to build different tools for their subsidiary com-panies much faster, which all feeds into a largersystem.

Low code tools could also make it easier to fa-cilitate data sharing between departments of alarge company.

Low code tools might also help change com-pany structures, so you can have small teamsdeveloping and using their own software, hesaid.

Small company business models

Low code could be very valuable for smallsoftware companies. “It’s a relatively inexpen-sive way to develop a platform,” he said.

A possible business model is that domain ex-perts can work for, or set up, a small companyand put together a software tool to do some-thing useful, perhaps working together withprogrammers. “They take their brains, they putit in a model,” he said.

However it can be difficult for small softwarecompanies to sell to large company customers.It may be worthwhile creating an umbrella or-ganisation with a number of small softwarecompanies serving a certain sector (such asbanking). This means you appear to the buyeras a company selling a whole portfolio of prod-ucts, he suggested.

When selling to big companies, you often facea ‘herd mentality’ problem where buyers onlywant to buy software someone else is using.

Another challenge is the organisation’s dreadedIT department, who will often be the peoplewho make the final decision whether the com-pany will use your software. “IT departmentscan be your friends or your enemies,” he said.

The best way to make friends with the IT de-partment is to develop software tools whichwill improve the company’s relationship withoutside customers, because that way the impactof the software will be felt much faster throughthe whole company. Or you can provide toolswhich enable the IT department to respond torequests from the company’s sales force ormarketing staff.

Selling to IT departments can be a longprocess. “They have to look at the application,test it, then they have to go and get money.You’re talking 6 to 9 months a year,” he said.

And also bear in mind that for the company totake benefit from your product, they may needto stop using something they are currentlyusing – and this could lead you to make ene-mies of people who are invested in the currentsoftware, he said.

Greece

When asked how Greece could do more to sup-port its software industry, Mr Karaplis said thatthere could be more co-operation between in-dustry and universities. “We don’t have thathere,” he said.

Greece also makes a very low amount of in-vestment in research, ranked 154 in the world,he said. Uganda spends a similar amount of itsGDP on research as Greece, 0.6 per cent, hesaid.

George Karaplis - low code - an inflexionpoint?The advent of ‘low code’ technology could be an ‘inflexion point’ – a point of major change – in the wholeuniverse of software, said George Karaplis, Senior Associate with IG Icon Investments


Roger Schank – making better expert softwareThere are much better ways to build software which can make a useful contribution to experts’ lives, saidRoger Schank, CEO of Socratic Arts. One way is to start with story telling



There are much better ways to build softwarewhich can make a useful contribution to ex-perts’ lives, said Roger Schank, CEO of So-cratic Arts, and perhaps a good place to start isstory telling.

Mr Schank’s company Socratic Arts developsperformance improvement solutions for com-panies. He is also CEO of XTOL Corp, a com-pany which offers technology career programs.He is Professor Emeritus of Computer Science,Psychology and Education at NorthwesternUniversity, Illinois (USA), specialising in arti-ficial intelligence, learning theory, and cogni-tive psychology, and a former Chief EducationOfficer of Carnegie Mellon University’s Sili-con Valley campus.

Mr Schank presented a video of Dimitris Lyras,the Chairman of the conference, telling a storyfrom several decades earlier of how his ship-ping company had a ship delayed in a port inthe Caribbean.

The crew went fishing for barracudas, whichwere fried, roasted, boiled and eaten onboard.Unfortunately the barracudas were poisonousbecause of the plankton they eat. The poisoncan be lethal. None of the crew died but somehad to go to hospital onshore and be sent home,and the ship left the port without an electricianonboard.

This story is irrelevant to nearly everybody, MrSchank said.

But if you were running a ship which was en-tering the Caribbean, the ship was going to bethere for a while, and someone takes out a fish-ing rod, it would be a good story to know.

Meanwhile modern computer software, and thesystems we use to run our organisations, isnowhere near being able to provide this infor-mation, or this story, at the right time. Peopleare usually asked to go on a course or read amanual to learn what to do.

It is not hard to find similar examples. MrSchank presented a video of a US solider whohad been asked to help set up a school inAfghanistan, working with someone locally.The soldier can follow orders but has no ideahow to do it. But there might be someone elsein the army who has done it before. The organ-isational challenge for the US Army should be

connecting this soldier with stories from thesoldier who has done it before, Mr Schank said.

A further example was of a ship going throughthe Suez Canal which had a fire in the boilerroom. The chief engineer started to shut downthe engine, as per the instructions in the man-ual.

The captain told him that was a crazy thing todo. The ship would be hit by the ship just be-hind it in the canal. If it entered port, the au-thorities might demand a bribe or try to stealthe cargo.

The point here is that we are still dependent onhaving the right person standing behind us atthe right time to provide us with relevant ad-vice. Software ought to be able to bring us theright story at the right time, but in 2015, it can’t,Mr Schank said.

Yet as people, all of us are experts at this.“That's what we do,” Mr Schank said. “We saysomething, and someone says, 'that's what hap-pened to me'”.

So bringing up the right story at the right timecould be considered a reasonable definition ofwhat ‘human intelligence’ actually means.

Artificial intelligence

Mr Schank is recognised as a world expert inthe field of ‘artificial intelligence’. Yet he hasbecome very disillusioned by the field, he said.

Consider IBM’s “Watson”, a computer systemdeveloped by IBM to answer questions on theUS quiz show ‘Jeopardy’, and won a $1mprize, competing against former winners. Wat-son is claimed to be an example that ‘artificialintelligence’ could now be better than humanintelligence.

But there is also an argument that Watson isnothing more than a search engine, Mr Schanksaid.

As an example, one of the questions in theJeopardy show was, "It was the anatomicaloddity of U.S. Gymnast George Eyser, whowon a gold medal on the parallel bars in 1904."Watson gave the answer ‘a leg’ which wascounted as an incorrect answer.

What Watson probably did, Mr Schank said, islook at the Wikipedia page for George Eyser,which said “Eyser competed with a woodenprosthesis for a left leg”. Watson could under-stand that the answer had something to do witha leg, but all of its artificial intelligence powerwas not able to connect the phrases “anatomi-cal oddity” and “wooden prosthesis”.

It is important not to confuse statistical analysisand pattern generation, which computers cando better and faster than human, with thedeeper pattern generation which humans cando, and this is probably what ‘intelligence’ is,he said.

You should also be wary of technology com-panies making marketing claims that theirproducts are ‘intelligent’ when they are nothingof the sort, he said.

One computer training company claims to offerthe ‘future of study’ by automatically creatingstudy aids using content, he said. On closer in-spection the software seem to just find contenton Wikipedia relevant to what you want tostudy, serves up some text with some wordsdeleted, and asks students to fill in the blanks,he said.

As an Artificial Intelligence researcher, “Ithought we were trying to figure out how themind works and making cool stuff,” he said.“This is not cool stuff or figuring out how themind works.”

There seems to be a growing belief in societythat expertise is something you can get fromtext books, not something you gain from expe-rience, Mr Schank said.

But you don’t need a text book to “do what I’mdoing right now, having an idea and talkingabout it,” he said.

Expertise is really shared through stories.“When people are telling stories, they are hav-ing an emotional experience,” he said. “Emo-tions are the real stuff. That’s when you talk tosomeone, you know how they're feeling be-cause you see it in their eyes. This is what lifeis about. Watson doesn't have any emotions.Watson isn't excited about anything.”

6



7


Video for assessing people

Mr Schank demonstrated software built by hiscompany Socratic Arts for assessing people,which could work just as well for finding pro-fessionals (for example a doctor), finding a lifepartner (a date), or finding an employee foryour company.

People who would like to be considered as can-didates (as your doctor, date or employee)record videos about themselves answeringquestions about important topics, such as howthey dealt with a tricky situation (a doctor) orwhy they want to be a programmer (a potentialemployee).

Watching someone on a video, we can usuallymake a judgement about them in a few sec-onds. And whether our judgement is good orbad, it will probably be the same judgement wewould have made after a 30 minute face to faceinterview, which is far more time consuming.

There is still light computer processing in-volved – the computer can spot patterns in thetype of rankings different videos get, and usethis to (for example) bring you someone verydifferent to the last person you saw, or verysimilar.

Compare this to other software tools on themarket which try to use computer algorithmsto suggest you a life partner who would be agood ‘match’, but have only text answers to goon, such as ‘I like to laugh a lot’.

In today’s society, the best way we can find aprofessional to help us (for example a psy-

chotherapist or doctor), is by looking at onlineprofiles people have written about themselveson websites, which are not usually very helpful.“My premise here is that this kind of thingcould change the world if and only if we getthe stories,” he said. “We need people sitting infront of a camera and saying what they thinkthat's relevant.”

Mr Schank told a story about his father, who“never accomplished much in his life.”

Mr Schank was a graduate of Columbia LawSchool, and was on a work experience place-ment together with a graduate of the less eliteNew York Law School. “I made sure to givehim all the junk work,” Mr Schank’s fatherwould say.

The graduate of New York Law School wenton to be Chief Justice of the State of New York.

You can see why a father might want to tell thisstory to his son if his son was being a bit cocky,as a sign that hard work and humility mighttake you further than arrogance. It is also astory which would be good to hear at the righttime.

“We don't always have the right guy to tell yousomething at the right moment, but you could,”he said.

Too much complexity

Another problem with most enterprise softwareis that it is too complex, he said. Consider thatmost organisation education software products

allow you to upload as many PowerPoints,videos and documents as you want.

But the best education software might do noneof this – it would just give students one task todo, and once a (human) teacher agrees it iscompleted, the teacher will put you onto thenext task, and provide assistance as required.

This software might not be commercially suc-cessful though. “Teachers prefer talking,” hesaid.

Also the educational establishment might notlike this idea, because educational profession-als often don’t understand the concept of goals,he said.

“But goals are the driving force for absolutelyall knowledge. There’s nothing we do whichdoesn't have a goal. Goal is driving every singleaspect of your day.”

But in the educational world, the emphasis isusually on ‘creating courses’ or trying to covertopics, he said.

Similarly in the Artificial Intelligence world,many companies talked about trying to build‘general problem solvers’.

Mr Schank always argued that this was impos-sible. “People who are good at solving prob-lems in one area are not good at fixingproblems in another area. The idea that theremust be such a thing as general problemsolvers, is simply wrong,” he said.



Aristos Doxiadis of OpenFund –a successfulstart-upOne of the challenges of setting up a new business for expert users is finding a market which is the right sizeand the right level of specialism, said Aristos Doxiadis of OpenFund

When setting up a start-up in the ‘software fordomain experts’ space, one challenge is target-ing the right size market, said Aristos Doxiadis,Partner of OpenFund, a company based inAthens which invests seed capital in softwarestart-ups.

If your market is too small, such as pharmaciesin Greece, you might not raise enough revenueto satisfy investors, he said. “If you solve aproblem that applies to all pharmacies in Eu-rope then it’s different,” he said.”But conversely, if your market is too general,then it might be too easy for someone else tocopy if you succeed. If your target industry is alarge industry where many people work in thesame way, “then it’s likely that the workflowcan be standardised and monitored with genericsystems,” he says.

One advantage of specialist software is that do-main experts are more willing (or comfortable)paying for tools which solve specific chal-lenges in their business, than they are payingfor more general productivity tools, which theycan often get for free in the cloud.To be successful, you probably need to havedeep domain knowledge within your softwarecompany. “I have never seen a start-up tryingto solve problems like the ones I outlined, whohave not had at least one of the team with directexperience of what they are trying to do,” hesaid.

“You need to have experience as a user, or avendor of services or products to the user for along time. You need deep domain knowledgeYou need to understand the crucial issues andthe secondary issues. You need to be providingtools which address the most important thingsthese people do every day, not nice to have fea-tures.”

If you develop a specialist business serving do-main experts, “it is easier to defend your busi-ness from bigger competitors,” he says.“Domain experts understand you will probablyserve better than a big generic competitor.”

Also, as a specialist software company, “youcan become a leader in that field without hav-ing to spend huge amounts on marketing andsales,” he says.

Examples of software for experts

Mr Doxiadis characterises ‘experts’ (in the con-text of making software for them) as peoplewho have a “very messy workflow.” Or inother words, “they don’t work in a standardisedway every time, which you can copy into aflowchart, or they don’t do things in the sameorder. Or if they do things in the same order, itisn’t suitable for the conventional way of en-tering data.”

Experts typically “gather data and analyse datain a very particular way that is usually not com-patible with enterprise software,” he said. Also,“what they want to optimise is not obvious toan outsider, and not obvious to someone notdoing the job with them,” he said.

Mr Doxiadis would characterise software beingfor ‘domain experts’ if it was designed to servea particular industry (sometimes called an ‘in-dustry vertical’).

He suggested four examples of ‘software fordomain experts’.

In aquaculture (fish farming), when employeesfeed fish in tanks, it is critical that they recorda number of observations, including at whatlevel of the tank the fish are swimming in. Ifthey are swimming close to the surface they areprobably not being fed enough food.

Beekeepers have to record data when they openbeehives, but are wearing heavy gloves at thetime, so need a computer interface which canbe used with gloves on.

Vineyard operators sometimes plant roses be-cause they are an early indicator of some typesof fungi, which may develop on the rose beforeaffecting the vines. The data about fungi can begathered to make plans to remove the fungi.

Jewellers might make a bracelet with preciousstones, where each one has a different weightand shape. “It is not easy using any sort of En-terprise Resource Planning (ERP) software toclassify, tabulate the material they buy or thecombinations they use,” he said. The measure-ments of the stones come with tolerances (errorbars) which need to be included with the data.The data also needs to record when the stonesare purchased and then they are included in abracelet.

OpenFund has not invested in any of the aboveexamples, but it has invested in three compa-nies which could count as “Software for Do-main Experts”, Discoveroom, Workable andIntellicent, he said.

Discoveroom is a system for owners of smallrental properties to keep track of their reserva-tions using smart phones, for property ownerswho do not have a dedicated office.

Many lodging owners are not working full timein their properties. They might be farmers, havewinter jobs as lawyers, or run a café. With Dis-coveroom, any time the owner receives a re-quest, he can see if the room is available on hisphone and tabulate the booking. The user inter-face “is the best in the world,” he said.

Workable is a tool for small companies, whohave no HR department, to manage the staffhiring process. For small companies, hiring “isa pretty complicated process,” he said. “Youhave to put out the ad, get out the CVs, com-pare the CVs, talk to people in the company soyou see the same people, exchange notes”.

“They found ways to make this very userfriendly and they are doing extremely well.”

Workable also makes it easier to compare CVs,even if they are in different formats, such as aLinkedIn CV, a specially made pdf file, or fromfilling in a form on a job board.

The software can tabulate key data about can-didates, such as how many years’ experiencethey have, what type of jobs they were in.“Workable found a way to parse the data in thevarious formats of the CV to make it easier tocompare,” he said.

This is a specific function which can be verymessy, and so hard to manage on non-spe-cialised software, he said.

OpenFund has also invested in Incelligent, acompany which makes software for mobilephone network operators to optimise the energysupplied to the network.

It can make predictions of how much transmit-ter power will be required, based on history andfuture events.

This enables operators to (for example) de-crease the power to the transmitters at times of

8



low demand, and have higher power availableat times of high demand (for example if thereis a football match happening nearby).

“They gather data from diverse sources, aboutthe weather, or sports events, as well as fromthe network itself, and develop a model for pre-dictive analytics,” he said.

Getting investment

OpenFund typically invests between Euro100,000 and Euro 750,000 in a company. It isone of four funds which are active in Athens atthe moment “working more or less the sameway,” he said.

So far OpenFund has seen more than 3,000ideas sent by e-mail, and had face to face dis-cussions on about 300 of them. Some of theseideas could count as “Software for Domain Ex-perts”, he said.

Different investors are looking to make invest-ments of different sizes. For example, some Sil-icon Valley investors will only invest inbusiness they think could be worth a billiondollars, known in the jargon as a ‘Unicorn’.

“They will not invest in businesses where themaximum likely valuation is $50m or $100m.They think, they will have to spend a lot ofmoney in the later stages of growth of the com-pany,” he said.

It may might make sense to look for peoplewho understand the specific industry sector, forexample people or companies in the telecom,farming or aerospace industry, who havemoney to invest.

You might find ‘Angel investors’ who can inject the first €100,000 or €500,000. “If youneed more than a million in funding you willprobably not solve the problem with angels.”

If you need larger amounts, there are venturecapitalists outside Silicon Valley who are inter-ested in ideas which could be valued at $100m.“There are VCs like that in Europe. You can getfunding in the range of a few $m,” he said.

If you can get loyal clients for your businessafter a “few million dollars” investment youcan be self-sustaining.

Investors will usually be looking for an exitpath, because investors are unlikely to be sat-isfied with dividends alone, he said. Thismeans there must ultimately be someone will-ing to acquire the company for “tens of mil-lions of dollars” as a trade sale.

However, “you don't necessarily need to builda business that is attractive to investors,” hesaid. “You can build a business which can em-ploy 10-20 engineers and build a good businessout of that and everybody will be happy.”

“We in VC tend to call these businesses'lifestyle' businesses.This is not a derogatory term, it is a perfectlylegitimate way to do business. If you have soft-ware ideas which are not very scalable, youshould still try them.”

Greece

On the question of Greece, Mr Doxiadis saidthat software engineers in Athens are a third ofthe cost in London, or a fifth of the cost of Sil-icon Valley.

An advantage to setting up a company inGreece is that teams tend to stay together forlonger. “One of the problems Silicon Valley hasis that it’s an open market. The very best engi-neers are shopping around all the time andmoving. In a mid-size company, they don'thave loyalty.”

A disadvantage of Greece is that “the graduatesof good Greek universities have got no practisein software engineering, although they are verybright and well trained in theory,” he said.

Graduates haven’t had experience producingsoftware modules within a certain time frame,or in how to develop software so that it is easyfor someone else to take over the work, he said.

Consequently there is a shortage of experi-enced software engineers in Athens, meaningpeople with over 5 years of experience. Start-up companies typically look for the brightestpeople in schools such as Athens University ofEconomics and Business (AUEB) and teachthem the practical skills on the job.

9




SOA, activity models and system modelsSpeakers from Athens University of Economics and Business (AUEB), National Technical University of Athens(NTUA) and the University of Piraeus presented research to come up with better ways to build enterprisesoftware, how to build activity models and how to build system models

Vassilis Zafeiris, from the Department of Infor-matics with Athens University of Economics andbusiness, explained "service orientated architec-ture (SOA) and how it might help make softwarefor domain experts.

The 'Service Oriented Architecture' approachesthe development of a system in terms of 'services'.These are defined as a piece of self-containedsoftware functionality that supports a specificbusiness capability.

A service has a broader scope than a softwarecomponent but narrower than a software applica-tion for the same problem domain. Examples of business capabilities that can be im-plemented as services are register a new customer,submit a new order, lookup the completion statusof an order, perform a bank deposit, he said.

Service composition is the essence of the SOA. Itallows the reuse of existing services, usually pro-vided by different organizational units, and theircoordination in order to support enterprise-widebusiness processes.

There are two basic approaches to service com-position: orchestration and choreography.

Service orchestration adopts the conductor/or-chestra metaphor where a central coordinator hasthe control of the composite service execution.

Service choreography represents a decentralizedapproach to service composition. The executionof the composite service is not controlled by acentral process. Each participant service is awareof its role in the composite service and executesas a response to events published by the other peerparticipants.

Services must be loosely coupled, that is a servicemust be designed to have the least possible de-pendencies to implementation, platform, deploy-ment, interface details. Loose coupling leads toservices that are characterized by fault toleranceand potential for independent evolution fromother services.

It is clear that loose coupling has a price that is in-creased system complexity. Increased complexity,results, in its turn, to higher development and/ormaintenance cost.

In “mainstream” SOA the emphasis is on the sup-port of enterprise-wide processes. The adoptionand evolution of SOA is based on centralized gov-ernance lead by a team of managers, domain ex-perts and software architects.

Building models for any activity

Kostas Kontogiannis, Associate Professor, De-partment of Electrical & Computer Engineering,National Technical University of Athens(NTUA), presented a method for building modelsfor any activity, which can be used as a basis forbuilding software for people who work in that do-main.

You can build a ‘metamodel’ for your activity,which includes different viewpoints, basic rela-tionships, processes and information systems.This could be put together by a domain expertrather than a software person. For example you can describe how a hotel reser-vation system works.

These models can be transformed to different“platforms” (so the same hotel reservation systemcan be used for different hotels).

There is a special language developed to describemodels. It shows how the various objects arestructured. The model can be built up of layers.

The jargon is to have ‘platform independent mod-els’ (PIM) and ‘platform specific models’ (PSM).Ultimately all PIMs become PSMs.

As part of the model, you can describe a structureof a system – for example say that a company hasdepartments, and departments have employees.You could have a model of an invoice.

To illustrate how it works, consider that comput-ers include ‘compilers’ which convert the code(written by the programmer) into something acomputer can run (basically 1s and 0s).

The ‘model’ approach takes this one level back –the programmer builds the model – the computergenerates the code, which is fed to the compilerand turned into something the computer can run.

“Think of model based engineering as raising thelevel of abstraction one level up,” he said. “Weare not writing programs, we are writing models.”

There could be many business opportunities cre-ating software for smaller enterprises based onthese models, he said.

Perhaps too may software start-up companies arefocussing on building apps. “My vision wouldn'tbe to build a little mobile application that doessomething very cool and sold 1m copies which

makes money - but ten days later - there will besomebody else that will develop another applica-tion. It’s a very short-lived market.”

Professor Nikitas Assimakopoulos

Professor Nikitas Assimakopoulos, from the Uni-versity of Piraeus, School of Information andCommunication Technologies, presented hiswork to develop a ‘systematic methodology’which can be used to understand systems in anyorganisation, with case studies presented for ahospital and a bank.

Mr Assimakopoulos’ view is that any organisa-tional environment can be modelled using ‘sys-temics’. You can include variable factors by using“variable system modelling.”

For example, you could model a bank by splittingit into five layers of systems. The first is the cen-tral operating divisions. The second is how thesedivisions are co-ordinated. The third system is thecontrols and audit. The fourth is research and mar-keting (including all external communications).The fifth is executives.

Control is very important in any system. “Thecontrol procedures are completely different to theexecution procedures,” he said.

Some banks have different control procedures forauthorising loans, for different branches.

The systems could be used to reduce the amountof documentation any project requires.

You can also develop systems to handle how agroup should make a decision. “Otherwise every-body says what he likes, what sort of knowledgehe has about the particular procedure,” he said.

Mr Assimakopoulos also certifies professionals,to say they are able to resolve project manage-ment problems, he said.

You can develop technology for all aspects of acompany’s systems.

Mr Assimakopoulos presented a business modelcanvas for budget airlines, normal airlines, chem-ical corporations and banks.

Mr Assimakopoulos has been involved in estab-lishing the “Hellenic society for systemic stud-ies”.

10




Payment Components, Sensorflare and B2BWavePayment Components, Sensorflare and B2B Wave are three Greek (or partly Greek) software companiesdeveloping software and ‘low-code’ tools for experts

Elias Gagas, Chief Digital Officer (CDO), Pay-ment Components, presented his company whichmakes tools / libraries, to be used as buildingblocks, for those developing software for financialtransactions. Payment Components also builds itsown FinTech applications.

The communications building block is a systemfor payments between banks called SWIFT whichhas been operated (as a company) since the mid-1970s. SWIFT includes communications sys-tems, security systems, and the model of whatmessages should be exchanged for each transac-tion. A global standard (ISO20022) has alsoemerged and a subset of it has become a Europeanstandard, called SEPA.

This standard system “creates an environmentwhich is pretty ripe for innovation,” Mr Gagassaid. For example you can create systems whichcan handle billions of transactions, or very com-plex transactions and automate the process ofsending & receiving them.

Having a mixture of standard software infrastruc-

ture, standard communication protocols, standardsoftware interfaces (APIs) and standard languagesmakes it much quicker to build new softwaretools and invent new things, he said. All of these things together could be called ‘lowcode’ – ways to make programming much easierand faster, he said.

Payment Components has a headquarters, salesand business development hub in London, butsoftware development is done in Athens. Clientsinclude JP Morgan, Computer Associates, Citi-group and Credit Suisse.

It develops messaging ‘libraries’, which are toolsto make it easier to incorporate SWIFT, SEPA andother communications standards into software.

Its ‘SWIFT MT’ libraries can handle cumbersomesoftware activities associated with SWIFT trans-actions (parsing, building, and transformingSWIFT messages).

It also provides a ‘Treasury Management System(TMS)’ solution called aplonCash, which makes

it easy for companies to add a SEPA and SWIFTpayments automation functionality to their treas-uries. This effectively creates a payments controlcenter which allows process automation, cost ef-ficiencies & accurate cashflow forecasting.

If you have multiple bank accounts, the softwarecould work out which is the optimum bank ac-count to make a certain payment (for example toavoid unauthorised overdraft fees) and make thepayment from there.

You can have a consolidated view of all yourtransactions across different bank accounts. Thissoftware is used by Greek shipping company Hel-lenic Seaways and Greek insurance company IN-TERAMERICAN.

Looking at standards in general, you can say thatthe internet was a result of standards, mobile tele-coms was a result of standards. “And on top ofthat we stated building things and things again andagain,” he said.

The automotive industry was also gradually de-

11



veloped using standardised components, he said.This has allowed new entrants to use existing au-tomotive ‘low code’ components – such as elec-tronics and chassis to disrupt and innovate.

This trend has been exemplified by Tesla, whichstarted building its cars by putting together otherpeoples’ components. A similar trend is happen-ing in the financial sector and combined withrapid innovation, brought by the likes of Bitcoin,P2P lending & social media, leads to an ecosys-tem in the brinks of a paradigm shift.

Sensorflare

Ioannis Chatzigiannakis, co-founder Sensorflare,talked about his company, which makes it easierto gather data from sensors and other internet en-abled devices, and shorten the time taken to buildsoftware.

Sensorflare’s business model is to create softwarewhich can automatically communicate with arange of different devices. This software can thenbe incorporate into other companies’ software, tomake their life a lot easier.

So Sensorflare will manage the hardware drivers(software to communicate with the sensors) andapplication programming interfaces (tools to con-nect the sensor software with other software).

Many people have predicted a rapid growth in theuse of internet enabled devices over the comingyears. “This is a technology we have heard a lotabout, but we still have to build it,” he said.

Last year there are estimated to have been 2 bil-lion internet enabled devices installed, includingin buildings, transport and other infrastructure.

By connecting sensors to the internet, we can“merge the physical and the digital, and combinethis with processes,” he said. “We get a new wayof doing business and a new way of working,leading to a really big cultural change.”

The business models get most interesting whenthey are part of people’s processes. For example,a vendor could fit sensors in an office which au-tomatically switch off light and heat when roomsare empty, and be paid based on the savings.

In the automotive world, you can pay an annualfee to have your car performance monitored andanalysed, with data from devices on the car auto-matically transmitted to a monitoring centre.

In the security world, a security company couldprovide a service to store data continuallystreamed from video cameras. If you need towatch some of the video to investigate a possiblecrime, you can automatically pull up video from

a specific time. Otherwise the video is eventuallydeleted.

The ‘ecosystem’ model could work in houses,where you have a range of devices and you cancontrol them all with your phone.

Another business model is to continually monitorthe environment (for example noise).

To build a business in this sector, you need goodhardware, then you need stable software, a clearbusiness concept and marketing strategy, MrChatzigiannakis said. “This is a combination ofthings that essentially make internet of thingsproducts really hard.”

It isn’t hard to come up with new ideas - for ex-ample someone posted on Kickstart an idea for amicrobrewery you could control with a smartphone. But does anyone want this?

Once you have your product idea, you need tobuild a large amount of software, to handle thesensors, collect data, analyse it and feed it intobusiness processes – not to mention storing thedata and ensuring security. You also need to de-cide how much you want to build yourself.

In terms of security, it is common for data fromdevices to be transmitted completely unencrypted.So if you are using a ‘Smart Weight’ device which

SYS T EMS

Many systems have come and gone from mainstream software houses

Domain knowledge and persistence has been the main ingredient of the success of the Ulysses Document Management Module

Software is no different than any other business in this respect

Greece is the cradle of the Shipping expertise and we are honoured to access this vast source of totally focused expertise

Designed by Seafarers for Seafarers

98 Rosebery Avenue, London, EC1R4TL, United Kingdom [email protected], www.ulysses-systems.com 23 Agiou Spyridonos Str., 185 35 Piraeus, Greece, Tel.: +30 210 4190800

Ulysses DOCUMENT MANAGEMENT since 1998

since 1998


Software is no differe


since 1998


nt than any other business in this respectSoftware is no differe


since 1998


nt than any other business in this respect


since 1998


nt than any other business in this respect








98 Rosebery Avenue, London, EC1R4TL, United Kingdom

98 Rosebery Avenue, London, EC1R4TL, United Kingdom

98 Rosebery Avenue, London, EC1R4TL, United Kingdom [email protected], www.ulysses-systems.com

[email protected], www.ulysses-systems.com

esigDor Sers feafarS

[email protected], www.ulysses-systems.com 23 Agiou Spyridonos Str., 185 35 Piraeus, Greece, Tel.: +30 210 4190800

yned besigerseafaror S

23 Agiou Spyridonos Str., 185 35 Piraeus, Greece, Tel.: +30 210 4190800

23 Agiou Spyridonos Str., 185 35 Piraeus, Greece, Tel.: +30 210 4190800

12



passes your daily weight onto your smart phone,someone walking past your house could hack inand see how much you weigh. But the benefits ofkeeping data simple could outweigh the possiblesecurity implications.

The aim is to make it as easy as possible to accessthe data with different devices. This might be eas-iest to do if you build the screen interface inHTML5.

Sometimes it is better to process data locally – forexample using a Raspberry PI computer or smallPC.

Different sorts of data need to be handled differ-ently. For example, if you are monitoring temper-ature, you might want to just provide an averagetemperature reading once an hour. If you are mon-itoring for fire, you might monitor continually, butonly communicate an alert when it is necessary. For sensors in one house, you could have 25 de-vices, generating 60 data readings a minute, hesaid.

The hardware drivers can be difficult. Some em-bedded hardware devices are programmed in acompletely different way to microcontrollers usedin other industries, he said. “You need really ex-perienced programmers to deliver a productwhich is really reliable,” he said.

B2B Wave

Alex Seimanidis, co-founder, B2B Wave, pre-sented his startup, founded in October 2014,which builds B2B e-commerce software forwholesalers, distributors and manufacturers.Mr Seimanidis had the idea for the business afterreceiving many similar requests to build a B2B e-commerce portal, while working as a freelancesoftware programmer.

The company currently has four customers, withthree using the same software (which is availablefor purchase) and one having a more customisedsolution.

The technology aims to reduce error and admin-istrative overload associated with managing or-ders, so business owners have more time toconcentrate on selling and building their busi-nesses, and less time answering phone calls ande-mails about orders.

The users can place products online, manage or-ders, and make suggestions to users about prod-ucts they might like.

Users pay for the software with an annual sub-scription fee.

“We are confident we can build a good product alot of small business would like to use,” he said.

B2B Wave aims to communicate with its cus-tomers on a weekly basis. Customers can seework in progress, provide feedback and suggestfunctionality additions.

The company aims to only introduce new featureswhich can be easily understood. “If we explain toa person what we are going to do and the persondoesn't understand in 2-3 minutes, then we say,forget it,” he said.

New feature requests are prioritised. For example,when Greece changed its VAT rates, incorporatingthat in the software was a high priority task.

B2B Wave also aims to build using the minimumpossible code. “That has a positive effect for usand our customers,” he said. “We can understandwhere we are going.”

The software integrates with tools from other software companies, including a tool to generatee-mails and a tool to send e-mails.

13


Speakers and Delegates

What did you enjoy most about the event?

“

“

As an IT professional I appreciated the agendaof the event (model driven software implementation). Speakers presentations werevery interesting, I enjoyed mostly the conversations and knowledge transfer. Keep up the excellent effort!Maria Siboni, IT & SAP Consultant

The Document "THEPRINCIPLES OF SOFTWARE FOR DOMAIN EXPERTS" is excellent (text andsketches).

“The variety of thetopics of the speakers.

14

”

”

”


SpeakersDimitris Lyras, Director, Ulysses Systems and Lyras ShippingStavros Messinis, Founder, The CubeGeorge Karaplis, Senior Associate, IG ICON Investments Roger Schank, CEO, Socratic ArtsAristos Doxiadis, The Open Fund, PartnerVasileios Paraskevopoulos, Graduate, NTUAKostas Kontogiannis, Associate Professor, NTUANikitas Assimakopoulos, Professor, Depart-ment of Informatics University of PiraeusElias Gagas, Chief Digital Officer (CDO),Payment ComponentsIoannis Chatzigiannakis, CEO, Co-founder,SensorflareAlex Seimanidis, Co-founder, B2B Wave

Delegates from Industry / Domain Experts / GovernmentChristos Priftis, EOPYYPanagiotis Papaioannou, Systems Analyst, EYDAPMaria Siboni, IT & SAP Consultant, FreelanceTasos Makris, Information Systems Director,Gourdomichalis Maritime S.A.Lakovos Diamantopoulos, Director, Halkidon Shipping Corporation

Agapi Iglesi, Technical & HSQE Coordinator,Halkidon Shipping CorporationVassilis Zafeiris, Teacher of Informatics, Ministry of Education and Religious AffairsAchilleas Choursoglou, IT Manager, Olympic Shipping and Management SAParaskevi Filippi, EHSQ Executive, S&B Industrial minerals

Delegates From Software IndustryMaria Calafatis, Community Curator, The Cube AthensPanos Nomikos, Business Development Director, Danaos Management ConsultantsS.A.Karl Jeffery, Editor, Digital Energy JournalLeonidas Katelaris, iNet WalkPanagiota Stagkouraki, Startup CommunitySpecialist, Microsoft HellasGeorge Pittas, OBS TechnologiesPopi Lyrintzis, MD, SQLearn Ltd.Chris Vosnidis, CEO, SQLearn LtdPavlos Kalenderoglou, Product DevelopmentDirector, Threenitas S.A.Efthimios Bakas, CEO, Threenitas S.A.Konstantinos Dimopoulos, Graphic Designer,Ulysses SystemsAngela Lyras, Content Analysis Manager,Ulysses Systems

Anastasia Vrakas, Commercial Director,Ulysses SystemsSpyros Goumas, Ulysses Systems

Delegates From Academia And StudentsEugenia PapargiriadouGrigoris Psomas, Akmi Metropolitan CollegeElena Lazaridou, AMCBarbara Evaggelou, AMC CollegeAsimina Papadimou, AMC CollegeVirginia Pollezel, student, AMC CollegeStamatina Kyrkmani, Business AdministrationStudent, AMC Metropolitan CollegePaola Mihaylova, Student, AMC MetropolitanCollegeChristina Spyronikou, BA (HONS) Marketing, Amc metropolitan collegeGiannis Karapiperis, AMC StudentN Diamantidis, AUEBLeandros Papadomanolakis, Student, Department of Informatics, University of PiraeusIlia Gouma, Student, University of PeiraeusNektarios Marmaras, university of PiraeusRallis Antoniadis, PhD Cand., University ofPireausSimos Xenitelis


Deliver less stressful environments Unify di�erent IoT devices and Deliver advanced tele-monitoring services.

Achieve absolutely minimum response times for time-critical products.

Sensor�are is an award winning platform, that helps you exploit

the power of your IoT Resources

www.sensor�are.com | info@sensor�are.comContact us to �nd out how we can empower your IoT product

Deliver high-availability solutions without thinking about the IoT InfrastructureNo need to circulate and store tons of useless data. Process on the edge and distill data that really matter. Deliver solutions that keep infrastructure costs at minimum.

Unify infrastructuresCombine legacy infrastructure with Innovative solutions. Simplify complex environments and maximize available infrastructure.

Integrate GloballyAct Locally

We work for the Future of Internet

Run anywhere - Recon�gure easily - Interact Intelligently

ess on the edge and distill daocrPo cirNo need t

Deliver high-availability solutions without thinking about the IoT Infrastructure

ess on the edge and distill daons of useless dae tore and sttculao cir


ertteally mat rta thaess on the edge and distill data. ons of useless da


. er



er solutions tha

Unify infrastructures

chiev

elivD

AUnify di�

astruc

Unify infrastructures

ely minimum r

t keep infrer solutions tha

e absolutchieves and DT devict Io oT devicener di� �er

t minimum.

esponse times f

osts ae cturastruc

ely minimum rancer advelives and D

Deliver less stressful environments

t minimum.

itical prcr-or time mes f for timeing seror-moniteleed tanc


.tsoducitical pr. esvicing ser


omple

w

Simplify combine legacC

the pothae is an a�arensorS

ts and maximiz

esour

onmenvirx enomplee with Iturastrucy infrombine legac

T Rour Io oT Rer of ywxploitou et helps ytha

d winning plaarwe is an a

ailable infr

es

ve ats and maximize solutionstivavnnoe with I

cesour

m, ortfd winning pla

. eturastrucailable infr. e solutions

o �nd out hot us ttaconC�ar.sensorwww

Run an

e can empow wo �nd out hoo@sensinffo@sensor | om .ce�ar

on�gurecR-e wheryRun an

T our Io oT prer ywe can empoom.ce�aro@sensor

n I-e easily on�gur

toducT pr

tlyelligentnt Iacert tly

software for domain experts - a new business...

Documents