Insights | Insights

The Back Office AI Opportunity

Patrick Mc Govern

Patrick McGovern

June 30, 2024
Back Office AI 6 30 24

The Back Office AI Opportunity

The buzz around application layer use cases of artificial intelligence in both horizontal and vertical contexts has never been louder. With companies increasingly accepting that AI will be part of their operations sooner rather than later, one big question that is still being answered is where within the mid-market and the enterprise will AI-native applications gain an initial foothold? Over the last 18 months, we have seen early examples of the adoption of vertical AI applications within medium and large companies and this trend will only continue. These are applications that rely on AI to be very good at tasks such as templatized document generation and review, data retrieval, and checklist-style workflows. But there is still skepticism about off-loading large portions of customer interactions to AI, and AI agent quality tends to vary significantly by the particular task.

One go-to-market approach that we are excited about, and which we see as a great way for vertical AI to break into the enterprise, is to use back office functions as a wedge (with an eye towards eventually moving towards mid/front office use cases). Back office functions are typically viewed as a cost center and the corresponding KPIs for VPs, business unit heads, and C-level staff around back office functions tend to veer towards cost cutting, outsourcing, and consolidation. Moreover, the back office is typically the domain of a business’s “chores” and since these are not client-facing, revenue generative roles, there tends to be a higher tolerance for imperfect work product than you might see elsewhere in an organization. Back office roles also often hinge on customer service offerings or reporting requirements which are regulatorily mandated, so companies will look to offer them at the lowest feasible cost basis. These dynamics lead us to believe that the back office is an area where there is significant near-term potential for the adoption of AI by mid-market and enterprise software buyers.

Financial Services

Financial services is the industry which the term back office is most typically associated with. And due to the highly-regulated nature of commercial and investment banking, there are ample opportunities within these organizations for AI to augment the human capital that keeps the bank’s operations compliant. Banks already have large teams monitoring transaction data for suspicious activity and filing federally mandated Suspicious Activity Reports (“SARs”) when they find evidence of things like nominee account owners or structuring of deposits to avoid making disclosures around sources of funds. Banks also have regulatory requirements around Know-Your-Customer (“KYC”) that require doing basic diligence on their depositors and managing the corresponding records, as well as burdens around proactive anti-money laundering monitoring programs. KYC & AML are both costly problems for banks to solve, and many will look to shift much of their back office KYC/AML efforts towards technology solutions vs. in-house staff, once the underlying software can show accuracy levels at parity with the existing workforce.

Many of the most regulated banks are also national or global institutions, leaving plenty of room for contract expansion as a piece of software is rolled out across different business units, geographies, etc. The corresponding revenues that only a few, large financial services customers can generate for an early-stage vertical AI business going after this opportunity also fits well into the EGVS thesis framework we published earlier this year.


Many medical providers are required by payors to provide extensive documentation in order to comply with value-based care payment regimes, and significant internal headcount is dedicated to gathering reams of patient and treatment data which is then compiled and submitted to government agencies and insurance carriers for reimbursement. Given where AI and AI agents are at today, and the slow pace of innovation among the incumbent EHRs, we see opportunity throughout the medical space for vertical AI solutions within the back office (i.e., referral management, prior authorization submittals, inventory management, etc.). We are also seeing AI being rolled out not just on the provider side, but payors are also increasingly using AI to manage the oceans of documents that are tasked with reviewing and processing from their provider base. We see compelling vertical AI use cases on both the payor and provider side and imagine an ecosystem will form around each of both of these potential software customer categories.

Mortgage Origination and Servicing

Many companies have tried over the last 20 years to tackle the opportunity presented by the heavy document management burden inherent to mortgage industry workflows. Mortgage origination and mortgage servicing are two examples of mortgage industry functions which rely on checklists, armies of back office workers, and highly-routinized processes. Prior to the last 24 months, the technology was not there yet to compete with the existing way of doing business but with the advent of LLMs this has changed and I would expect the next generation of players looking to automate mortgage origination and servicing to have far more success than we have seen in the past. Today, mortgage servicers still have large armies of workers who are manually processing documents based on a variety of checklists - I struggle to imagine this will still be the standard operating procedure in even 10 years for this key piece of the mortgage sector. Origination also presents opportunities to use AI to streamline the origination process, allowing originators to issue far more mortgages with the same sized teams.

Franchise Management

Many sectors of the US economy are operated using franchisee/franchisor business models. This includes almost all fast food restaurants, most quick service restaurants, and many hardware stores, gyms, automotive repair chains, real estate agencies, and home services businesses. Today, the relationship between the corporate parent company and the downstream franchisee locations is codified in a franchise operating agreement, but guaranteeing adherence to the many rules laid out therein can be challenging and is currently managed by teams of overburdened franchise compliance and operations professionals. We see a compelling opportunity to use AI in order to enable compliance and operations teams to manage many more locations and do so much more proactively using a AI-native franchise management platform.


The energy sector is one of the most regulated industries in all of the United States. We see opportunity across the large private and quasi-public enterprises who are responsible for today’s electrical power generation to sell them vertical AI to ease things like regulatory reporting burdens, permit management, and routine maintenance of their grid networks and physical infrastructure. While all of these use cases would benefit from layering on AI, the report generation required to be complaint with state and federal agency regulations around things like annual/monthly operations, abnormal incidents, temporary plant shutdowns, wildlife impact, and more serve as natural entry points for someone looking to build an AI-native command and control center for energy production facilities.


The operators of airlines and railroad networks are required by groups like the NTSB, FAA, FRA, etc. to provide extensive pre- and post-trip documentation in an effort to ensure these critical modes of transit remain able to operate safely. Moreover, they also must maintain paperwork attesting to the proper inspection, maintenance, and repair of their respective fleets, all of which must be drafted by someone after the inspection occurs solely for documentation purposes. This kind of formulaic document drafting is one of AI’s strong suits. These transit operators also have back office support staff which oversee things like aircraft/train maintenance schedules, route optimization, fuel planning, etc. By training AI on the existing workflows and data of these back office transit professionals, there is an opportunity for AI to double-check (and potentially) augment the fleet planning and maintenance protocols currently overseen by large teams of specialists.

The Back Office Expansion Path

We believe that by starting in the back office, vertical AI applications will be positioned to: i) capture immense amounts of their customers’ data to fine-tune models and develop new product lines, and ii) build trust between the enterprise and a given software vendor, which that vendor can leverage when rolling out additional modules that may touch on higher-value, customer-facing interactions. We also believe the ACVs for these vertical AI back office software products will be much higher than what we have seen back office solutions command in the past, as these will be augmenting human capital in a way that allows them to charge much higher contract values as a result of how they augment human capital and provide services on par with those of existing employees.

While many of these back office uses cases are large enough to sustain venture scale businesses on their own, the ability to enter an enterprise as a back office AI tool, leverage the company data flowing through your platform, and expand into a full stack AI solution for your given category makes them that much more exciting from an investment standpoint. If you are a founder building vertical AI for the back office, I would love to speak with you.

If you liked “The Back Office AI Opportunity” and want to read more content from the Bowery Capital Team, check out other relevant posts from the Bowery Capital Blog.