B2B and Enterprise Product Ideas - January 2020
Happy New Year everyone. I’ve spent some time the past few months (like most VCs) looking at opportunities for new products and services within business applications and software infrastructure that serve SMBs, mid-market, and enterprise customers. A few of these ideas and themes are hyped; others are not.
This isn’t meant to be a list of predictions or glances into the future (most may end up being wrong!), but rather, products I’m personally looking out for and am excited about in B2B.
Business Applications
Collaborative Enterprise Search
Enterprise search could be viewed as document and file search across data silos and folder structures. It could also be viewed as a subset of business intelligence — querying insights and deriving analysis from structured data. Regardless, enterprise search should be a collaborative and shared activity across a company where query methods are more accessible to end business users. A user should be able to see a historical record of other questions asked or searches conducted inside her organization. Furthermore, the search software should create and update company knowledge stores based on relevant searches.
ThoughtSpot is interesting from a tech perspective, but it doesn’t have the collaborative feature set I envision. The Elasticsearch product (part of the ELK stack) is geared towards developers and is typically used on log data — pulling from something like Logstash.
HR Service Delivery Automation for Today’s Enterprise
Human resources teams get inundated with requests. I don’t understand my benefits package. Can I take these days as PTO? I’d guess many of the requests are fairly repetitive and point to relevant information in company wikis or systems. I’m curious to see how this function could get automated — similar to what Moveworks is doing to the IT help desk — and tailored towards the HR needs of the modern enterprise.
Think about the employee and workforce base of a company like Coca-Cola. It’s complicated with workers across offices and countries maintaining different layers of employment status (full-time, contract, etc.) Utmost, as a spinout of Workday, recently got funding to take on extended workforce management.
Process Analysis for Risk Functions
If there was a tech trend that captivated 2019, the superset of low-code, RPA, and operations automation was probably it. An adjacency to these technologies is “virtual process analysis,” where software captures a holistic picture of business processes and functions through computer vision and machine learning. FortressIQ is an example of a company that does this.
Automation functions clump up in the enterprise, primarily by departments. One area I’ve paid attention to, which tends not to get much love, is a company’s governance, risk, and compliance (GRC) org. Legal, compliance, procurement, IT security teams are interdependent and interwoven, and they stand as the source of many repetitive processes. Rather than having an external team of business process consultants come in and examine what happens, perhaps virtual process analysis technology could play a role.
Modern Dun & Bradstreet
Dun & Bradstreet is an 180 year-old company (yes, that’s right) focused on serving up commercial credit and identity data. Think of it almost like an Experian or Equifax for business (those product lines exist as well), where companies buy data noting the creditworthiness and identity profile of another company. Even with these solutions in the market, good credit risk data on external vendors or counter-parties is hard to come by. Use cases could include a commercial landlord screening retail tenants, a bank on-boarding a new client and working through compliance, or a B2B marketplace assessing a multitude of buyers and sellers on its platform.
There might be ways to efficiently bootstrap data collection on businesses, which could enable a startup to catch up with the incumbent D&B. For example, on the consumer side, Truework provides workflows for HR departments within companies, so they can grab reliable datasets from employers. They also focus on powering the individual with their own data. With a B2B-focused tool, one could provide workflow software to banks or payments processing software for free to real estate landlords. Middesk is the interesting startup in this space, but it looks like they’re primarily focused on business identity (i.e. is this a real company?) right now.
Talent Intelligence and Enablement
Salesforce created the system of record (the CRM) in the cloud for sales teams, the traditional revenue center of the enterprise. Now, there are companies hitting significant scale building sales intelligence or sales enablement tools — some that come to mind include People.ai* (reliable data cleansing), Seismic (content management), and Outreach (engagement platform). They take the software a step further.
The talent outreach and acquisition function is vital to a business, commanding an increasingly larger budget. If Greenhouse and Lever are analogous to Salesforce (building the system of record for talent), what will be the intelligence and enablement tools for recruiting?
Holistic Internationalization / Localization Solution
At this point, I’m a broken record on internationalization (building the infrastructure so a business is ready to expand its borders) and localization (tailoring content and services to local audiences in different countries). Some push back and say there have been no sizable outcomes or exits in the category to date, but the tide will turn as software shifts from a vertical phenomenon to a horizontal one.
New interesting businesses are being built within pockets of the larger framework — such as Transifex (localization workflow), Lilt (machine translation), Papaya (global payroll), and Deel (contractor payments and compliance), but there may be an opportunity in stitching together these solutions and providing a more complete experience for companies managing both internationalization and localization processes. I hate to phrase it this way, but think of “AWS for Going Global.”
IT Infrastructure
Scalable Logging Platform
Logging, one of the three pillars of observability, can become very expensive with complex applications tying together many microservices due to *persistent* storage costs. The cost of storage scales linearly with things like EC2 instances. Many of the recent successful cloud IPOs have focused on observability or application monitoring (which are different), proving that one can build big businesses within the categories. And with most IT infrastructure, new computing paradigms and application architectures bring forth new market leaders. Within IT, there’s a continual progression to cloud (or hybrid-cloud), along with services-oriented applications and event-driven architectures.
For the other legs of observability, Lightstep has a point of view on tracing and Chronosphere is building a next generation metrics platform given learnings from scaling Uber. What company can round out the trifecta? One contender in logging could be LogDNA. Cribl is another company to watch.
Next Generation Data Warehouse
Data warehouse companies engineered for scale continue to be big winners for investors with Snowflake, designed for cloud-native environments, as a recent example. A data warehouse that couples itself with Kafka (Confluent) or other event streaming platforms feels like an opportunity. It looks like Confluent has accepted streaming platforms won’t completely replace the the prominence of the database architecture today or in the near future given that they launched ksqlDB.
A long-term data store for events could work — for example, look at the Apache Iceberg project.
Anomaly Detection in Production Environments
Open observability solutions, like Prometheus and Grafana, pull together data from applications and production environments to give developers and DevOps a clearer view of what’s going on. While they provide structure and data, I believe the next step would be to deliver insights on that data, particularly the flagging of anomalous events. After that, you can turn to remediation of resources or compromised code.
I understand this a very hard machine learning problem, but if someone cracks the code (no pun intended), one could build an incredibly powerful SIEM (security information and event management) tool. Splunk is big, but increasingly dated. Exabeam is interesting.
Abstracted Compute
Nutanix’s stock took a hit in 2019 as it transitioned its business model to the cloud. For those unfamiliar, Nutanix provides “hyperconverged infrastructure” (read: abstracting compute and storage) for enterprises. There’s an opening in the market to provide abstractions on compute and storage not only suited for modern application and infrastructure environments, but that also fill the needs of today’s developer.
This thread from Cindy, which goes into more detail, was excellent. Wasmer looks compelling.
Data Management and Backup for Regulated Industries
Large swaths of the industrial economy are undergoing digital transformation. The effect will take hold exponentially over the next few decades. Many industrial markets, like aircraft production, energy sites, and defense contracting, are heavily regulated with companies under intense scrutiny.
These businesses' data assets are mission-critical not only to the companies themselves, but also for regulatory bodies concerned. Rubrik has built a large (and quickly growing!) data management and backup business geared for the enterprise, but a data backup and versioning solution for the long tail of these industries might work. It would differ with hyper-focused product marketing, bottoms-up adoption capabilities, and ease of ability to audit.
ML Ops
As software development proliferated for enterprises, DevOps gained steam as a function to shorten software development lifecycles. Over the last few years, cybersecurity practices in companies have operationalized with “DevSecOps,” which in turn, produces market opportunity. In DevSecOps, Palo Alto Networks acquired one of the leading players, Demisto, for $560M.
The next bucket to go in the organization will probably be data science and machine learning. Companies will hire more data scientists and machine learning engineers and actually deploy models into production (only a small handful of companies actually do ML well). Then, you need to test and monitor the models “in the wild.” For example, Fiddler Labs* enables companies to put trust and visibility around their AI practices. All these steps will require operational workflow and personnel.
*Denotes a Haystack portfolio company
If you’re building in or thinking about these products and spaces, feel free to get in touch with me at aashay[at]haystack.vc.