the wearable to consumer translation: giving your dog a voyce
TRANSCRIPT
Confiden1al and Proprietary – Not for distribu1on 2 2
Topics
o Background o Decision points o Implementa8on o Considera8ons
Confiden1al and Proprietary – Not for distribu1on 3 3
Who We Are
i4C Innova8ons Inc. is a wholly owned subsidiary of
Intersec1ons Inc. (NASDAQ: INTX), a leading provider of
iden8ty risk management, data monitoring and other consumer subscrip8on-‐based services.
Leveraging Intersec8ons’ years of experience monitoring disparate
complex data points and presen8ng useful informa8on in simple to understand format to consumers, i4C Innova1ons was created to transform the use of data in the veterinary, pharma and nutri8on animal world.
I4C’s Voyce monitors and collects pet health data through advanced and
unique technology. It then translates and dispenses
ac8onable health informa8on, educa8onal
content and other management tools for a passionate market of pet owners and veterinarians.
1996 2011 2014
Data and Insights
Confiden1al and Proprietary – Not for distribu1on 4 4
What is Voyce
Voyce is an comprehensive plaOorm. It delivers ac1onable informa1on to pet owners and veterinarians alike, in easily understood formats.
Customized Content: The en8re Voyce experience is tailored to the pet owner and pet. Recommended ar8cles, weekly newslePers and reminders are based on the dogs personal informa8on and the owner’s interest.
TRENDS & INSIGHTS Detailed health trends keep you and your vet connected between visits
RESOURCE CENTERConstantly updated content from top experts...tailored to you and your dog
CUSTOM NOTIFICATIONS Timely communications on wellness indicators and advice to help you be a better pet parent
REMINDERS & UPDATES Keep up with appointments, vaccinations and more
SYMPTOM CHECKER Determine when to seek veterinary attention
NOTES Keep track of changes in daily routines, medications, and more
Confiden1al and Proprietary – Not for distribu1on 5 5
Value Chain
Monitoring and engagement with clients • Enhances preventa8ve care op8ons to pa8ents
• Increases engagement with pa8ents through visits and bePer rela8onships
• Creates addi8onal revenue streams, increased visits & monitoring solu8ons
Educa1on and insights Owners and Pets • Insights into key wellness indicators
• Educa8on and early illness detec8on and management
• Cost savings through preven8ve care
PET OWNERS HOSPITALS / VETS MARKET
Plan Compliance & Studies • Powers accurate consumer and animal studies for new product research
• Offers an accurate baseline, parsed by breed and age
• Measures efficacy of certain drugs
Confiden1al and Proprietary – Not for distribu1on 6 6
Building the Voyce PlaOorm Looked at key partnerships and adop8on of standards that allowed us to “build the airplane while flying” …. 1. APIs as a strategy 2. AWS cloud managed 3. To-‐be-‐determined SaaS providers 4. A combina8on of open source and proprietary
applica8on development 5. Large integrated development partners
Given unknowns, MuleSob was aPrac8ve as it provided a low-‐risk path to interoperability and allowed us to mature along the way (security, scalability, etc.)
Confiden1al and Proprietary – Not for distribu1on 7 7
Key Requirements
1. Large number of devices connec8ng over HTTP
2. Changing data structures and formats
3. Mul8ple – Configurable, transac8onal data flow steps 4. Mul8-‐target data streams 5. Integra8on with S3 and SQS 6. Integra8on into exis8ng device management systems Core ESB 7. Single vendor providing an Enterprise Integra8on Role
Confiden1al and Proprietary – Not for distribu1on 8 8
The Data – Weekly
Nega1ve influences to Customer Experience: o Freshness of informa8on (performance) o Resiliency to retain data (data gaps) o Content and services driven from data (data in context)
Types # Data points Size
1 Check-‐in 42 2 KB
2 Content 67,400 5,900 KB
3 Configure 42 2 KB
1 Band 67,484 5,904 KB
1,000 Bands 6.7 M 5.5 GB
10,000 Bands 675 M 55 GB
50,000 Bands 3.3 B 275 GB
Confiden1al and Proprietary – Not for distribu1on 12 12
Device Core Paberns
Device data
capture
Device data load
o Device Signature Valida8on o Message Valida8on o Route Message o Persist message o Add to Queue (could be internal or
external) o No tolerance for errors or
excep8ons
Confiden1al and Proprietary – Not for distribu1on 13 13
Device Data Capture flow
Device message valida8on
Device message load to Queue
Device Management
Legacy Integra8on
Confiden1al and Proprietary – Not for distribu1on 14 14
Device Core Paberns
Device data
capture
Device data load
o Offload Queue o Correlate/Validate o Post to Core Services/Core DB o Trigger Aggrega8ons and
No8fica8ons o Error Handling
Confiden1al and Proprietary – Not for distribu1on 15 15
Device Data Load – (Flow) Offload Queue
Co-‐relate the messages
Post to Core DB
Confiden1al and Proprietary – Not for distribu1on 16 16
Device Data Load – (Flow contd) Trigger
Aggrega8ons and No8fica8ons
Error Handling and Compensa8ng Control
Confiden1al and Proprietary – Not for distribu1on 17 17
Core Paberns
Create Business En88es
Retrieve Business En88es
Update Business En88es
Delete/Cancel Business En88es
o Log business flow start o CRUD business en88es in internal
and external systems o Create response o Log Response/Errors (compensa8ng
control) o Send response back
Confiden1al and Proprietary – Not for distribu1on 18 18
Core Business Flow
API Router
Invite Caregiver Flow
Confiden1al and Proprietary – Not for distribu1on 19 19
Performance
Tuning and Results • Combine Device Capture and Device load job in to one node and tune threads • Use local queues for ini8al processing. On non-‐recoverable error use SQS as
reprocessing queue • Performance Tuning along with other code recommenda8on from MuleSob
consul8ng services, we were able to produce 150 requests per second, simula8ng about 25,000 devices
20
Considera1ons
o Consider an architecture review before deployment o Use an external tool for log monitoring e.g., SPLUNK o Use logs to evaluate device performance and trouble shoo8ng
Create these evalua8on rules early o Load test early, load test oben o Consider a con8nuous integra8on tool e.g., Jenkins. Use the
Console API to integrate and deploy packages o Clustering is not always required for IOT use cases. In error cases,
data should be persisted on the device and resend o Decouple Device interac8ons and Data Loads by providing low
latency high performing services for Device connec8vity to offload data quickly