Founded in 2002, Elzware is a natural language processing company that specialises in human to computer conversations. These systems are commonly called Chatbots but some people call this kind of technology Artificial Intelligence. We caught up with Phil Hall to find out more about Elzware, the wider implications of this tech, and how the Bath and Bristol tech scene has impressed him…

TechSPARK: What’s your background and how did you get into AI/Chatbots?
Phil Hall: AI for me has been about the human computer interface for as long as I can remember. When I first got my hands on hardware, a ZX Spectrum in 1982, I immediately tried to get it to talk to me. I created a very simple “hello how are you” kind of tree and went from there. None of that code is left, but the logic and desire was awakened.

The threshold for me was entering university in my early thirties (I wasn’t quite ready for higher education when I was in my early twenties). Studying anthropology at UCL was a great experience; come the final piece of work, which needed to be an Ethnography, I decided not to go to Costa Rica but to go to Meridian 59 which was the world’s first commercial 3D MMORPG.

What that experience did for me is set a benchmark on digital stupidity – I’m in a virtual bar (obviously) trying to get sense out of an Non Player Character (NPC) and it’s just rubbish at communicating. I wasn’t studying NPCs, I was looking for people’s sense of self, their (here’s the academic spin) articulation of subject and objectivity. In English, who were people online, who offline and how did those two things interface with each other. I was hooked!

“Natural language is coming to get us and the world is just stepping off a precipice into an interesting and somewhat murky new set of communication services”


This innovative/unusual experience and my history in engineering of various flavours got me into an international systems integration company and while there I worked in Email Response Management Systems for a while, then started working with really early what should be called ‘Chatbots’ though everyone and their mother has tried to brand the notion to something else.

I built systems, I trained people, I deal with cultural change management implications and all that good stuff, but I wanted more from this fledgling conversational market. Then a swerve in the road…

As the company I was working for absorbed another company things slowed … right … down … Like ‘carrying treacle through bushes’ was the local joke, so I decided to put my money where my mouth was and set up Elzware. That was in 2002.

Merly the Mersquid: an iPad-based construction, part of a larger
interactive fiction called 7 fables

The punchline from then still holds true now:

Elzware Ltd is a company set up with the specific objectives of opening up a conversational system marketplace in line with: available bandwidth, digital consumer devices and the general public’s desire to communicate with computers in natural language.

That’s the real driver now, natural language is coming to get us and the world is just stepping off a precipice into an interesting and somewhat murky new set of communication services.

TS: What are you working on at the moment?
PH: Right now I’m moving a really complex socio-cultural system proposal forwards locally with an NHS focus and also reviewing the games and VR sector to see how close we are to acceptance of a ‘Jeeves’ to look after us. I mean the butler rather than the branded search company.

Aside from that and because my world is a touch on the strange ticket and not something that people realise they can control, understand, influence, purchase, be-afraid-of, I’ve taken on a little public speaking. The South West Virtual Reality conference, more on that later, and an Analytics Conference of sorts.

“I believe that an open source, transparent and ethical system that can deliver conversation for any interested party against defined business requirements is what should be happening”


I’m also presenting a talk entitled ‘AI is dead. Long live AI’ in April where I’ll be looking to set the record and the future straight.

Elzware is also putting together a Git-fronted Linux back-ended Chatbot deployment system based on a responsive HTML5 website where conversation is the driver for conceptual steps in a person’s progress through info-tainment.

This is a Platform-As-A-Service play which will open up the doors for rank and file business people or other organisations to deliver a Siri/Cortana like experience to their users.

Bottom line, right now I believe that an open source, transparent and ethical system that can deliver conversation for any interested party against defined business requirements is what should be happening. It shouldn’t cost squillions of pounds and it shouldn’t be tying organisations into long-term service relationships.

TS: What is the future of Chatbots/AI in your opinion?
PH: The future of Chatbots has already been written: The media memes from Metropolis, through the Turing Test up to I Robot, Star Trek, 2001 A Space Odyssey, Red Dwarf pick any production of the last 60 years has AI interactions conversationally with computers as seamless and human like.

In reality voice recognition is patchy, text to speech can be highly bespoke and expensive and the focus is in part patent wars and in other part educational inertia that is leading professionals in AI to try and control a definitive layer between people and their digital realities.

What we are going to see in the near future is increasing amounts of ‘hard AI’ being applied to try and control the conversational layer – think of this as a Black Box.

We may see some beautifully built systems, think of these as ‘Glass Boxes’ but the history of Chatbots in customer service has not been particularly endearing to the public. Underneath many systems you need to look at those stakeholders that have a vested interest to not rock the boat; I’ve seen lots of that over the years but the route to solve that issue is a long one.

“The future of Chatbots has already been written: any production of the last 60 years has AI interactions conversationally with computers as seamless and human like”


What worries me is the future of Chatbots used as surveillance engines through the hidden application of Machine Intelligence of whatever flavour.

Building a relationship with a human through AI is fine if the ethics, privacy and value of that conversation is clearly explained and bought into by the human user, but we aren’t in that space. We are in a place where businesses are running rampant and the users, us, are either unknowingly or uncaringly delivering value to third parties in a fashion that has little historical precedent.

For my money there needs to be some hard conversations held publicly about how Wild West the current state of the internet is. At present ignorance is not bliss and while the pragmatic English person will wryly smile and suggest that in their lifetime AI will become self-aware and kill us all off (doh) that isn’t really a future that we can entertain apart from as entertainment of course.

“I’m actively working towards conversational AI being broken out from the hallowed halls and valleys, and bought into the hands of expert laypeople”


I’m hoping and actively working towards conversational AI being broken out from the hallowed halls and valleys, and bought into the hands of expert laypeople.

That’s a wish though, pragmatically the future is that Black Box AI will be increasingly pervasive and could easily put a barrier between humans and the internet that shouldn’t be there.

TS: Do you think education can be improved by Chatbot tech?
TS: Let’s look at the education sector and reflect on some of the prototype systems that Elzware has made.

We started with a thing called DesignBot back in 2006 which was worked up with a technologist called David Perry – he had the vision for a conversational educational system in the 1980s and we worked together through the trial of the system at Cabot City Technical College in Bristol.

This then spawned the building of the TeachBot English system, which was fully developed, and a skinnier prototype TeachBot Maths.

Meet TeachBot English: TeachBot English is a learning support tool for school and homework tasks, that gives personalised learning based on students pace and goals

These systems were built in 2008 to 2011 and the user group – KS3 and KS4 students – engaged well with the systems, opening up any-route-through teaching increased the simple exchange of information significantly and with a little ‘gamification’ the system was looking towards a regional rollout … until … highly political changes in the education market from Michael Gove put people’s back up and then the implications of the casino financial sector bet and lost all the world’s money and our gatekeeper organisation, the City Learning Centres was disbanded.

What was missing was a non-proprietary solution for education and also an appropriate hardware paradigm to deliver learning to students which was, let’s be clear, just not there in 2011.

Elzware teetered for a while on the brink of stepping away from the Chatbot market until two events occurred in close proximity.

The first was the uncovering of a viable open source engine and codebase that could deliver to Elzware’s aspirations and the second was a funder and academic partner with sufficient vision to look at blending AI with educational methods and theories.

The funder was the Paul Hamlyn Foundation and the academic institution was Ravensbourne College and over a period of 18 months a participatory lead design method was applied to a system that actively engaged with creative people looking to teach them a specific skill, WordPress, but looked more deeply into their drivers and capabilities.

It did this by identifying a user’s level of thinking, according to Blooms taxonomy of such, and then moved users through the educational steps by Kolb and Fry’s experiential learning cycle. This for me was a consolidation of Anthropological principles from my time at University and also the lessons learnt through TeachBot in the secondary level.

The Shift: a creative thought and education delivery service which has been co-designed by young people aged 16-24

So, to answer the question directly – Chatbot technology can and should improve education assuming that systems are built with sufficient insight, inspiration and in-cahoots with the users.

The alternative is an increasingly utility biased internet showing a Filter Bubble of information according to the GAFA (Google Apple Facebook Amazon) hegemony. This will restrict students gaining their own information and enlightenment in context to their query, not their consumer user-type.

The back end of digital systems will always point to Data Scientists, but the front end should not strip out the ability for people to explore in their own terms. Chatbots can support that endeavour.

the_shift_homepageTS: You are very keen to get Bristol and Bath as a major development area for AI? How could this happen?
PH: Well the situation with Elzware in the last few years has been that it’s been working out of the area, which was where the contracts dropped, but it’s far better to be able to work in the exploding Bristol and Bath digital community. The trigger point for me was the Health Hack Day, which really bought it home to me how much there is locally.

The practicalities of how to move forwards with a local AI centre of excellence internationally has many different tendrils and I’ll be moving forwards with these in the coming year as time, money and synchronicity allow.

The arrival of free and open-source solutions and the passion to drive them into production needs one key thing though – people.

I’m thinking that we need to approach generationally, by that I mean we look at the primary school kids, secondary, tertiary then into the professional coding market.

There are enough companies out there in the world that will deliver FAQ systems or live chat with transient humans, but a hybrid Glass and Black Box solution as an information manufacturing activity, that’s new and it’s the engaged people who will make it.

I’m looking in the coming months to engage with the code club, recently sourced from these very pages, to work with primary students in Chatbot building.

For the KS3 and KS4 students, Elzware has got some history with the development of the English Reading and Writing system TeachBot – but we’d like to come back in with this open source system and assist with the recent ICT GCSE revolution.

Underneath the open source conversational system is C/C++ for those that want to really push the envelope, but the pattern matching efforts themselves feeds into linguistics, logic, problem solving and a whole range of good skills for the 21st Century student.

Spiral thinking: Phil Hall worked with David Perry on the trial of a conversational
educational system at Cabot City Technical College in Bristol

I’m also keen to not forget the tertiary education layer as we’ve some brilliant internationally leading groups, units and courses at the Universities surrounding Bristol and Bath.

I’m pretty sure that the way to nail this down is to attack the problem directly. Get some seed funding together, bring a limited number of ideas to a development table that can fit the socio-cultural parameters and then build them out in an Agile fashion straight out to market.

“I’m also keen to not forget the tertiary education layer as we’ve some brilliant internationally leading groups, units and courses at the Universities surrounding Bristol and Bath”


I don’t have all the answers to this question, but I do believe I have a plan of attack and structure of the ‘dream team’ to take an innovative approach to AI and make Bristol and Bath the international hub for such.

TS: What was the subject of your talk at the South West Virtual Reality conference? Were they ready for you?
PH: The South West Virtual Reality conference, was a great day. Respect is due to Dan Page and the broader team at Opposable Games for making the first step with this.

One of the delegates said to me after, “… it feels like year 2 or 3, super professional and slick …” we should be proud in the local area to be taking the first step to setting the EU centre for VR development right here as part of our broader technological evolution.

That said, my presentation in the tech stream was bringing up issues only glossed over at the main stage. Don’t get me wrong, the speakers were brilliant and the topics covered all interesting – I’ve never seen such low drop-out of delegates at any gig to be frank. But where people were talking about how it would be good, is good even, when characters interact, no-one was talking about the methods to do this.

There were some rumbling from the final panel about the kind of fear that the notion of VR and AR brings to traditional media production methods, so maybe I was just a little farther out there than most people were ready for.

It really is a return to first principles for me as this allows me to focus on the NPCs that so frustrated me back in the 1990s. I talked about renaissance development methods under the title of Engineering + Anthropology + Conversational AI ( Ext.I. ) = NPC+

Bringing relevant experiences in my academic, professional and ‘making’ life and the context from my ‘Futurology’ (for want of a better term) I drove people towards the notion of Extended Intelligence, as opposed to Artificial Intelligence.

I pinned this down towards the human requirements of narrative and storytelling, creation and engagement. I suggested that the AI in games, the physical AI where environments and NPCs move and interact in a way that is meaningful to games mechanics is fine. But we need the conversation to be added.

When I was in Meridian 59 in 1997/98 the Chatbots/NPCs were reactive, dumb and stupid. Now in Skyrim in 2013 the Chatbots/NPCs are prettier, but they are still giving very limited options.

There is obvious logic in the GUI (graphical user interface) and most people in games want to know two things:

1. What do I need to take from point A to point B?
2. How can I kill this thing in a quicker or more entertaining way?

“In triple A games with a budget of millions there is a cultural change required to really allow conversation to take part in games”


On the other hand, in triple A games with a budget of millions there is a cultural change required to really allow conversation to take part in games.

I’ve looked quite deeply at the history of this conversational failure in the games world and it’s a subtle thing that has gone ‘wrong’. It’s part approach, part deliverable, part trajectory of higher education, there are a lot of parts!

TS: What most excites you about the South West?
PH: With the high-quality people around and about in Bath, Bristol and farther afield the joy is easy travel and good-quality communication lines.

Our region is really buzzy right now with the ever rampant music scene and our historical capability to build anything, everything whether buses, bikes, cars, planes – it gives me a great lift to know that everything is possible.

Some of the South West Mobile folks, for example, are putting a Hack Day together dealing with voice input and voice output.

There are some good tools out there and knowing Bristol there’ll be skills and people out there that’ll get stuck in.

I keep my fingers in a dozen or so of the local meetups and groups in Bristol and Bath.

The exciting bit is the depth and breadth of the region, the danger is spreading one’s brainstem to thinly and never getting builds done.

If we go to the side and look at the more ‘traditional’ tech community, the Engine Shed, SETsquared and a squad of people fighting for our Local Enterprise Zone to be kicked off big time.

That’s a brilliant opportunity for a hub to be built right by Bristol Temple Meads that can be the physical representation of the digital network.

TS: What other tech companies/startups impress you in the South West?
PH: First and foremost the Bristol Games Hub should get a huge shout out.

The major companies that are there, Opposable Games, Auroch Digital, and then the squad of freelancers and smaller companies are seriously professional and the quality of the builds is as good as you’ll find anywhere.

It’s easy to just look to build apps/games/systems because you can, rather than thinking about what those builds mean for people and how the synthesis of people’s digital and physical experiences can be enhanced for the good of the South West and inevitably UK PLC.

When I meet people that are keeping their eyes on the build rather than the patents or posturing, that really impresses me, and I’m always on the lookout for ‘dream team’ members whether they are individuals or organisations.

My vision here is for (a) Conversational system building to be moved into an area where it is a common skill (b) utilisation of Glass as well as Black AI in delivery of digital information (c) an elegant and available editor to drive Animations/Avatars/Information as needs be (d) make some money (e) push UK PLC’s digital skills base forwards starting right here in Bristol and Bath.

phil_hall_photo_elzwareThanks to Phil for taking the time to talk to us. You can find out what Phil’s up to next via his blog, and we’ll keep you updated you via the TechSPARK event calendar too.

Kulsoom Middleton