Research: Innodata

Innodata

What they do:

• data engineering
• solve complex data challenges that companies face when they build and maintain artificial intelligence (AI) systems and analytics platforms
• seems like they basically manipulate large data sets so it is more usable
• 27.6mm shares outstanding at $1.95 for $56mm market cap
• they have $11mm of cash and $5mm of debt
• revenues were flat yoy on Sept quarter
• about 35% gross margins
• they made about 3c EPS in September quarter
• they were operating cash flow positive $4mm of operating cash flow in 9m

Employees:

• have 3000+ employees
• deep data domain expertise in various fields, including law, sciences, health, finance, and technology
• process data in over 25 languages
• work from our global operations centers in India, Israel, Sri Lanka and the Philippines
• expert staff provides an attractive alternative to the crowdsourced labor pools of competitors
• well-suited for high-context data, such as legal contract classification, medical images, medical records, and scientific and legal literature
• hybrid approach to produce large-scale, highly accurate data – AI and human expertise to tackle

AI SaaS Platform

• have AI-augmented SaaS platforms
•                 automates complex data annotation and data transformation tasks
• combines advanced dataflow, deep learning (a branch of AI), and purpose-built applications used by human experts
• enables us to perform data annotation and data transformation at higher efficiency
• dataflow technology enables us to configure workflows for specific data annotation and transformation setting and refining our accuracy thresholds and quality assurance parameters
• our AI is suite of domain-specific and task-specific microservices each of which performs a discrete data-related task
• AI microservices have a range of capabilities for data annotation and data transformation including deep sequence labelling, categorization, segmentation and sequence-to-sequence mapping
• when review required, dataflow automatically routes data to an appropriate human expert
• Synodex intelligent data platform – competes with Risk Righter, EMSI, Parameds and a few BPO companies
• Agility intelligent data platform competes with Meltwater, Cision, Kantar, Infomart and West Corporation

Problem they seem to be trying to solve:

• For AI-based algorithms to perform accurately, they need to be trained on large amounts of high-quality data
• projects fail, stall or perform inadequately because data sciences teams are unable to perform the complex and resource-intensive data preparation tasks necessary to properly train, tune, and operationalize AI models
• preparing high-quality data takes up 80% of the time for most AI and machine learning projects
• 19% of companies responding stated that lack of data or data quality issues was a main bottleneck holding back further AI adoption

New Market Opp

• re-designed our solutions and product portfolio in order to address the needs of enterprises across verticals for data annotation and data transformation
• aim to dramatically expand our addressable market
• historical core market for providing data services to information companies is relatively small (estimated by us to be approximately $250 million and to not show growth over the next several years), the market for AI and machine learning-relevant data preparation solutions is estimated to grow from $1.5 billion in 2019 to $3.5 billion by the end of 2024
• overall enterprise AI spend that is projected to reach $53.06 billion by 2026, registering a CAGR of 35.4%
• intend to shift our revenue mix from “services” to “solutions” and “SaaS products”

What is data preparation?

• data annotation (which is estimated to take up 25% of the time)
  • image and video annotation services and platforms may be used to annotate, or label, objects or people in images/video for facial recognition systems
  • text annotation services and platforms may be used to convert raw text data into richly tagged, AI training data
  • provide image/video data annotation and text annotation as full solutions, in which we provide all required technology, infrastructure and expert resources
  • provide data annotation for healthcare, compliance, scientific, financial and legal markets
• data transformation (which includes data identification, aggregation, cleansing and augmentation and is estimated to take 55% of the time)
  • data transformation solutions for high-accuracy data identification, aggregation, cleansing, augmentation and extraction
  • enables data to be extracted from websites, as well as internal data stores; converted from disparate formats including PDF; enriched with the necessary semantics, metadata and linking; and classified in accordance with an ontology or knowledge graph
  • consumed via API
• TAM: AI and machine learning-relevant data preparation solutions is estimated to grow from $1.5 billion in 2019 to $3.5 billion by the end of 2024
• global services and technology company focused on data transformation, enrichment, and management

Q2 Results/CC

• Synodex and Agility: expecting that we grow both of these businesses this year
• Synodex
  • have built the technology and systems to extract complex medical data from unstructured medical records
  • first half revenues increased 28% over last year, with revenues this quarter increasing 31% year-over-year.
  • enjoys strong 60% plus incremental margins, and nearly all of its revenue is recurring in nature
• Agility
  • SaaS Platform
  • providing full PR workflow platform
  • small player in the overall $3 billion global PR workflow market
  • ended the quarter with a year-to-date net retention of 87%, just a few points shy of our internal target of 90%
  • booked about $750,000 of new business with a small direct sales staff
  • from Q1 2019 to Q4 2019, we practically tripled bookings per sales executive.
  • given improved productivity, economics now support rapidly scaling the sales force
• on AI market:
  • late last year, we discovered a whole other market, practically made-to-order for us, just in its formative stages with significant growth expected in the next several years. This market is the AI data preparation and annotation market
  • Data sciences teams that want to build AI models need to train those models with large quantities of very high-quality data. But they express continued frustration that creating high-quality data is a task for which they are ill equipped
  • started marketing and selling AI data prep and annotation services in Q4 of last year
  • have closed 15 new customers. We have another 16 customers in late-stage pipeline that are expected to close in the second half.
  • majority of these deals will produce recurring managed services revenue at our target margins.
  • one of our recent data annotation wins is with a prominent big tech company
• did a very important release just in the fourth quarter. And the result of the work that we’ve done, the release that we’ve done is enabling us to compete with the 2 largest companies that dominate this market.
• product has been validated by Atlas and ranked highly in, I think, 7 out of 9 areas. It’s ranked as the top product
• saw a couple of important acquisitions take place last year in this space. One of which was -I want to say it was 5x revenue, their valuation
• Forecast:
  • new data annotation market in combination with a forecasted expansion from one of our largest traditional market clients will nevertheless enable us to show sequentially improving revenues
  • just the first half of 2020, we have already booked 71% of our full year 2019 bookings
  • presently forecasting beating 2019 bookings by 32%,
  • expect that the result will be a net savings of approximately $2 million in 2020 and $2.6 million in 2021

Tough to buy this after the big move.