The document explores challenges and success factors for AI startups aiming for scalability. Through interviews with AI startup founders, it develops a framework to classify AI startups based on their value driver and offering. The main findings are:
1) AI startups improve scalability by applying lean startup and customer development principles to systematically develop their business model and products in a market-oriented way.
2) Startups need to establish organizational structures, standardize processes, and specialize their technology to develop productized businesses.
3) International expansion can help scale startups but requires fully productized solutions; initial expansion occurs through clients and partners while eventual local presence is needed for customization.
4. 4
Expectations
Time
AI HYPE
HYPE CYCLE FOR EMERGING TECHNOLOGIES
(GARTNER, 2016)
X
AI
today
disillusionment
Plateau will be reached in 2-5 years (Enterprise AI maturity)
peak of inflated expectations
plateau of productivity
1
6. AI startups
Startup
growth
QI: How do
startup-
growth
principles
apply to AI
startups ?
Productization
Q2: How can AI
startups develop
into a productized
business?
Inter -
nationalization
Q3: What role
does
internationalizatio
n play in AI
startups?
6
RESEARCH AREAS
2
7. SYNTHESIS – AI HYPE CAN KILL THE
FOUNDATION
Startup
sales
growth
Expectations
Time
Hype Cycle (Gartner, 2016) vs. Lifecycle of an
Entrepreneurial firm (Picken, 2017)
1. Defining business model & analyzing the market1
2. Setting a foundation, productizing & specializing2
Expectations Sales growth
THEORETIC CONCEPTS
7
AI
Startup
sales
growth
Expectations
Time
AI in the Hype Cycle vs. Future AI startup growth
scenarios
today (2019)
Possible growth
scenarios**
**Unknown trajectory post-AI-
hype
AI STARTUP GROWTH IN A HYPED ENVIRONMENT
Expectations Sales growth*
*During AI
hype
1) Theoretic concepts include: Lean Startup Theory, Customer Development Model
2) Theoretic concepts include: Software productization, Service productization, Hedgehog concept, Transitioning stage in Lifecycle
3) Theoretic concepts include: OLI framework, lean global startup
3. Scaling & internationalisation3
2
8. 8
Method Semi-structured interviews
Approach Theory testing Theory building
Participants 7 experts 9 AI startups
Geogr.
Coverage
Globally Globally
Analysis
results
Developed categories Tested, refined categories
Outcome
High-level framework
High-level success factors
Low-level framework
Challenges
Success factors
RESEARCH DESIGN
Phase 1 Phase 2
3
9. Research
objective
EXPLORE CHALLENGES AND SUCCESS FACTORS FOR AI STARTUPS THAT AIM FOR SCALABILITY
Research
questions
Research
areas
Theoretic
concepts
Analysis
categories
1. APPROACH
TO BUSINESS
MODEL
DEVELOPMENT
2. MARKET-
ORIENTATION
3. SETTING &
MAINTAINING
DIRECTION
4.
ORGANIZATIO
N BUILDING
5. STANDARD-
IZATION
6.
INTERNATION
ALIZATION &
PRODUCTIZAT
ION
I. How do
startup-
growth
principles
apply to AI
startups
1. Startup
growth
1.1 Lean startup x
1.2 Customer
Development x x
1.3
Transitioning
stage in
lifecycle
x x
II. How can AI
startups
develop into a
productized
business?
2.
Productizat
ion
2.1 Software
productization x
2.2 Service
productization x x
2.3 Hedgehog
concept x
III. What role
does
internationaliz
ation play in
AI startups;
3. Inter-
nationalizat
ion
3.1
Internationaliza
tion in tech
x
3.2 Entry modes
3
10. Commercial-
ization
Research
Advisory
Standardized
products
TYPE II: AI consulting (limited
scalability)
TYPE I: Diffused, no clear focus (not
scalable)
TYPE IV: Scaling a product (high
scalability)
TYPE III: Focused tech innovation
(creating IP)
Value
driver
Customer-specific
solutions
Offering
AI STARTUPS CLASSIFICATION
FRAMEWORK
10
4
11. Commercial-
ization
Research
Advisory
Standardized
products
TYPE II: AI consulting (limited
scalability)
TYPE I: Diffused, no clear focus (not
scalable)
TYPE IV: Scaling a product (high
scalability)
TYPE III: Focused tech innovation
(creating IP)
Value
driver
Customer-specific
solutions
Offering
AI STARTUPS CLASSIFICATION
FRAMEWORK – APPLIED
11
S8 (’08)
S6 (’12)
S9 (’14)
S5 (’16)
S1 (’17)
S7 (’18)
S2 (’17)
S3 (’14)
4
12. Commercial-
ization
Research
Advisory
Standardized
products
TYPE II: AI consulting (limited
scalability)
TYPE I: Diffused, no clear focus (not
scalable)
TYPE IV: Scaling a product (high
scalability)
TYPE III: Focused tech innovation
(creating IP)
Value
driver
Customer-specific
solutions
Offering
RECOMMENDATIONS
12
Hypothesis-driven, market-
oriented, maintain direction,
organizational structures,
standardization
Hype-driven short-
term growth; no clear
direction, no
standardized
processes.
4
13. 13
SUMMARY
Question Finding
1. How do startup-
growth principles apply
to AI startups?
AI startups improve scalability, if they apply a
systematic approach to business model development (Lean Startup),
market-oriented product development (Customer Development),
and set & maintain a direction (Lifecycle – Transitioning stage)
2. How can AI startups
develop into a
productized business?
Organization building: structures, delegation (Lifecycle – Transitioning
stage)
Standardization, modularization for repeatability and scalability
(productization)
Specialization within data-types and data-processes (Hedgehog concept)
3. What role does
internationalization play
in AI startups, and how
does internationalization
link to the productization
of the business?
Market-seeking for scale – only viable when fully productized, to gain
ownership-advantage (OLI)
Initial expansion via clients and partners (Lean Global Startup)
Ultimately local presence due to need for customization (Software Startup)
4
Editor's Notes
In the following 15 min I will provide you an overview on th scalability of AI startups
my presentation will cover an introduction to the topic, my research questions and literature findings, the research design, findings, and recommendations. A summary is provided for discussion.
AI major transformative force of this century;
Founder of coursera put it, AI is the new electricity; it has the same level of impact on industries, as electricity had centuries ago.
Th reflected in fast growing revenues, the AI industry revenues are growing 43% p.a.
Also, in just one year, AI startup funding tripled.
But what happens to the many emerging startups down the line?
The hype cycle of emerging technologies provides a view.
public expectations towards a new tech get hyped through media.
Businesses often lack knowledge on the tech and therefore become disappointed when they realise the promises of the new tech are not fulfilled easily. Only at a later stage will expectations and capabilities align.
AI is currently at the peak of inflated expectations. While AI customers have a strong desire to use the tech, they primarily need strategic advise on where to start.
AI in enterprise is expected to reach a mature stage, meaning it will be more common and known, in 5 years from now.
So, what does all of this mean for AI startups; we know that only few startups survive, and even fewer achieve, what is called scalability, the ability to grow revenues faster than costs. Therefore, this research explored the challenges and success factors for AI startups that aim to scale.
Break down into research questions, …., I explored 3 key areas that help break down the reserch objective and inform the research questions.
Gaps
How it applies to AI startups
Various growth curves – diverting from Picken
Productisation – what activities?
Software startups – how to
Internationalization
Lean startup by Ries
Cust dev by Blank
Lifecycle by Picken
Software productization by Artz et al
Service productization by Ritala et al
Hedgehog concept by Collins
Internationalization
Motives: Bailetti, Onnetti, Blank
Entry modes
OLI: Dunning
Burgel and Murray: customization
Rasmussen and Tanev : lean global startup, partnering
But what will happen after the AI hype?
Without investing in business model development, standardization, processes and organization building, the startup is likely to experience a long-term decline. Falling into the trap of "hype-driven business", such an AI startup experiences success on the back of customers' lack of knowledge. Revenues are growing as long as enterprises require AI advisory and custom-made solutions. However, once the majority of enterprises have reached a mature stage in applying AI, these startups will struggle to scale. In the long-run, such an AI startup is unlikely to compete with providers that are focused on specific markets or domains and offer standardized products in a cost- and time-efficient manner.
1-high level, how the theory applies to AI startups. For example, focus was here on how to transform from being consulting focused towards creating scale, Approach to business model development
2- test and develop theory, ie input data and data processes to specialise market focus
Setting & maintaining direction
Organization building
Market-orientation
Processes, standardization & modularization
Internationalization
Lean startup by Ries
Cust dev by Blank
Lifecycle by Picken
Software productization by Artz et al
Service productization by Ritala et al
Hedgehog concept by Collins
Internationalization
Motives: Bailetti, Onnetti, Blank
Entry modes
OLI: Dunning
Burgel and Murray: customization
Rasmussen and Tanev : lean global startup, partnering
8 out of 9 AI startups have one foot in AI consulting (type 1); this highlights the high market demand for advisory and custom-made solutions.
Majority is cluttered between type 1 and type 2, indicating that they are gradually moving from 1 to 2
1 Startup is pure type 2, with a clear product (MVP), emphasized how difficult it is to stay focused
2 AI startups are at risk of getting trapped in the AI Hype, they have developed a product vision but are heavily investing resources in delivering customer-specific solutions and advisory
The lean startup theory …Experts suggest AI startups lack hypothesis-driven approach when they get pulled into various directions due to the hyped environment. Those AI startups that apply a systematic, hypothesis-driven approach, scaled sooner than the rest of the sample. Take example type 2 and 4
product to scale, address the specific needs of a clearly defined target market T2: value, data, and receptiveness. trade-off blue-chip customers and standardizable product. AI consulting cannot replace … getting diffused.
set and maintain direction: narrow down on key customers, product differentiation, and profitable project delivery. ... hyped environment. …consulting, they are at risk of losing focus. …agility and flexibility vs direction
organization building: technical teams along with their products, modules, or industries. Delegation, increased effectiveness and efficiency. industry experts to manage customer engagements. …external funding–the funded AI startups have effective structures and processes in place
Standardization for repeatability and scalability. customers’ input data. …specialize. Process, re-use models …AI consulting - service productization...’ unrealistic expectations,
Internationalization: market seeking, .. AI startups are different from common tech startups when it comes to internationalization. Even though they initially seek channel partners just as other tech startups, they ultimately setup a local presence due to the high level of customization required for AI products… less productized , less efficiently internationally.-ownership-advantage- fully productized.
Lean startup by Ries
Cust dev by Blank
Lifecycle by Picken
Software productization by Artz et al
Service productization by Ritala et al
Hedgehog concept by Collins
Internationalization
Motives: Bailetti, Onnetti, Blank
Entry modes
OLI: Dunning
Burgel and Murray: customization
Rasmussen and Tanev : lean global startup, partnering