A new research paper on the structural problem in home-based care M&A, the model we have built to address it, and what changes for the 35,000 agencies the industry has not historically served.
By Jarrett Bauer, Founder and CEO, Montauk AI.
The paper draws on nearly two decades of operating and advising in home-based care, combined with the AI infrastructure Montauk AI has built for the structural realities of the sector.
Goal
To introduce the research paper from Montauk AI on AI-native investment banking in home-based care, and to walk through what it means for founders, investors, and operators in the sector.
Key Takeaways
- The home-based care market is structurally underserved. Roughly 35,000 independently owned agencies generate hundreds of billions in enterprise value, yet investment banks of any consequence engage above $450 million. Most operators have never had a real advisory option.
- AI-native investment banking is a model where the M&A workflow is built around what software does well, with people doing the work that requires judgment, relationships, negotiation, and accountability. We have built this model at Montauk AI.
- The paper tests the model against four preregistered research questions covering valuation agreement, predictive validity of algorithmic buyer matching, time-to-close efficiency, and pre-transaction value creation.
- Continuous exit readiness, organized through Operate, Optimize, Exit, is the operating concept the paper introduces. Agencies do not prepare for a sale episodically. They build the discipline that determines exit outcomes as continuous practice.
- The architectural argument extends beyond home-based care. The sector is where the model is most testable, given its fragmentation, standardized financial reporting, and active consolidation cycle.
- The model has operated at institutional scale. Montauk AI served as exclusive placement agent in the sale of GrandCare Home Health Services to Pennant Group (NASDAQ: PNTG), a transaction that placed AI-native execution opposite a publicly traded strategic acquirer.
What this article covers
- The structural problem in home-based care M&A
- Why now
- The model and how it works
- The four research questions
- Operate. Optimize. Exit. and time to value
- Proof in practice: GrandCare and Pennant Group
- What this means for the home-based care industry
- Implications beyond healthcare
- FAQ
The paper
The full text of AI-Native Investment Banking in Home-Based Care, including methods, the four research questions, illustrative cases, and limitations, is available at montaukai.com.
Download our AI-Native Investment Banking Research Paper at montaukai.com.
The structural problem
Roughly 35,000 home-based care agencies operate in the United States today. Most are independently owned by people who built them over decades. Together they generate hundreds of billions in enterprise value across home health, hospice, home care, palliative, and post-acute services. By the measures that matter for healthcare, patient volume, capital flow, demographic tailwind, this is one of the most consequential markets in the country.
It is also a market without institutional advisory infrastructure.
Investment banks of any consequence engage above $450 million in enterprise value. Agencies below that threshold have historically had three options. Hire a generalist business broker. Represent themselves opposite institutional buyers who do this for a living. Or stay out of the market entirely.
None of those options reliably produce a strong outcome. Among letters of intent that do get signed in this segment, a substantial share never reach close, and most of the failures trace back to operational or clinical data that did not survive diligence.
That is the gap.
Why now
Three forces are converging in home-based care that have not converged before.
The first is demographic. Ten thousand Americans turn 65 every day, and preference for aging in place has shifted demand decisively away from institutional settings. The sector’s growth trajectory is structural, not cyclical.
The second is capital. Private equity dry powder in healthcare is at record levels, and PE-backed transaction volume in home-based care has accelerated year over year since 2015. Capital is hunting for deployment, which compresses the time founders have between deciding to engage and watching a process develop around them whether or not they participated.
The third is generational. The founders who built these agencies are reaching retirement age in significant numbers. Many built their businesses through decades when the sector looked nothing like it does today. The generational exit cycle is happening now, not in five years.
These forces are not arguments for selling now. They are arguments for being ready now. Operators who reach 2027 or 2028 still on the sidelines of preparation will face a different market than the one available to those who have used the next twelve to twenty-four months to build readiness.
The model
AI-native investment banking is the model we have built at Montauk AI.
The M&A workflow is organized around what software does well. Financial data is ingested and normalized in hours. Buyer-seller fit is scored in parallel across a curated database on strategic, geographic, financial, and regulatory dimensions. EBITDA add-backs are surfaced systematically, months before a process begins.
Execution runs continuously rather than episodically. The platform runs across dozens of engagements at once without the quality drop-off that comes from spreading any one team thinner.
People do the parts that require judgment, relationships, negotiation, and accountability.
Analysts validate every AI valuation before it reaches a client. Match recommendations are reviewed before any outreach is initiated. All externally delivered materials pass through human review. Every client-facing call and negotiation is managed by a banker.
The checkpoints are deliberate. The architecture is built to hold accountability in place.
The paper tests whether organizing the work this way produces measurably different outcomes.
AI-native vs. AI-enhanced
| Dimension | AI-Enhanced | AI-Native |
|---|---|---|
| Financial analysis | Models built manually with some automation | Financials ingested and normalized in hours; QoE adjustments generated systematically |
| Buyer outreach | Sequential, through personal networks | Parallel, across a curated database with algorithmic scoring |
| Pre-transaction work | Asset is marketed as-is | EBITDA uplift and operational improvements identified before going to market |
| Engagement model | Episodic: engagement, process, end | Continuous exit readiness |
| Scale | One team, limited concurrent deals | Platform runs across dozens of clients simultaneously |
| Access | Institutional-grade above $450M EV | Institutional-grade across the size spectrum, including founder-led mid-market |
Two things to be clear about. AI-native is not autonomous. The hybrid model is the design, not a transition state on the way to something else.
And the question the paper takes up is not whether AI tools help in M&A. They help. The question is whether organizing the entire workflow around them produces outcomes that other models cannot reach.
The four research questions
Four questions structure the paper. Each maps to a measurable dimension of the model.
| RQ | Focus | What it tests |
|---|---|---|
| 1 | AI–analyst agreement | Whether AI-generated valuations of home-based care agencies agree with human-analyst estimates within a 15% mean absolute percentage error |
| 2 | Match predictive validity | Whether the algorithmic buyer-seller match score predicts deal progression from initial response through LOI to close |
| 3 | Process efficiency | Whether AI-native execution is associated with shorter time-to-close relative to the nine to twelve month industry benchmark |
| 4 | Value creation | Whether pre-transaction operational optimization increases exit multiples relative to agencies sold as-is |
Each question is preregistered. Primary analyses are conducted or independently replicated by a statistician unaffiliated with the platform.
De-identified data supporting the analyses are deposited in a public repository on acceptance.
The conflict of interest is named openly: the platform studied is built by the author.
So are the limits of the design. RQ3 and RQ4 use observational comparisons against industry benchmarks and within-subject pre/post change, which is appropriate for evaluating performance on live transactions but not sufficient for strict causal inference.
The paper reports descriptively where causal claims would not be warranted.
Operate. Optimize. Exit.
The compressed timeline is the result most readers will focus on. The framework that surrounds it matters more.
Traditional investment banks optimize the sale process. Engagement begins when a transaction is contemplated. It ends when the transaction closes. The work happens during a defined window.
AI-native investment banks optimize the outcome. Engagement begins months or years before a transaction. At Montauk AI, the work is organized in three phases, and we engage with operators across all three.
Operate. The financial and operating infrastructure of a serious business. FP&A foundation, KPI dashboards, monthly close cadence, board-ready financial reporting. This is continuous work, and it is where we start with operators who are eighteen to twenty-four months from a transaction.
Optimize. Engineering enterprise value before going to market. EBITDA uplift initiatives, workforce utilization improvements, clinical and star-rating gains, revenue cycle and payer optimization, technology enablement.
This is the work traditional banks do not do because it falls outside the engagement window. For our clients, it is often where the largest share of exit value is created.
Our operating leverage playbook for home-based care walks through the specific levers in detail.
Exit. The transaction itself. A diligence-ready data room, algorithmic buyer targeting, CIM and comps strategy, AI-accelerated execution.
The framework changes what readiness means.
Operators do not prepare for a sale in the weeks before a teaser goes out. They build the financial and operating discipline of a sponsor-backed mid-market platform as continuous practice.
The result is more leverage at the moment of transaction, and the freedom to time the transaction to market conditions rather than personal circumstances.
Time to value
The compressed timeline matters because of what it changes for the owner.
A nine-to-twelve-month process condensed to three-to-five months means liquidity is reached sooner, operational disruption is shorter, and the transaction can be timed to market conditions rather than to the gestation of the process itself.
Combined with pre-transaction work in the Optimize phase, the gap between an as-is sale and an AI-native sale is more than speed.
It is the value built before the process, the certainty during it, and the time returned to the owner after.
Our companion piece, the 12-month seller readiness plan, walks through what the readiness side of this looks like in detail for founders considering a sale in the next twelve to twenty-four months.
Proof in practice: GrandCare and Pennant Group
The model has been tested in market. Montauk AI served as exclusive placement agent in the sale of GrandCare Home Health Services to Pennant Group (NASDAQ: PNTG), one of the most active public-company consolidators in home-based care.
That role is institutional by definition. Exclusive placement agent to a public-company acquirer means the workflow, the materials, the buyer interactions, and the diligence discipline have to meet the standard of a NASDAQ-listed strategic operating under public-market scrutiny.
There is no separate, less-rigorous bar for the founder-led mid-market on this kind of engagement. The same standard applies.
The work itself matched the model.
AI-driven market intelligence identified and benchmarked over 150 regional operators across the relevant geographies.
GrandCare was repositioned from a regional home health agency into a platform acquisition, with emphasis on its 5-Star clinical quality and 6,000+ patient census across four counties.
The framing matched what a public acquirer underwriting for accretive growth was actually looking for.
One transaction does not generalize, and the paper is careful to treat it as illustrative rather than representative.
What it does establish is that AI-native execution can deliver an outcome that meets the standard of a public-company buyer.
The workflow, the matching engine, and the diligence discipline behind the GrandCare outcome are the same ones available across the size spectrum of operators we work with, not only at the top of the market.
What has been an astounding buyer’s market will shift to sellers as new tools are introduced and traditional information asymmetry is eliminated.
What this means for the home-based care industry
The practical change for founders is access.
Institutional-grade execution becomes reachable across the size spectrum, not only above $450 million.
Valuation, buyer interest, and exit timing become visible in ways they have not been before.
The operating discipline of a sponsor-backed platform becomes available to agencies that could never have afforded one.
For investors, the same architecture produces a structured view of a market that has been opaque for as long as it has existed.
Cleaner signal on which agencies are positioned for premium outcomes. Faster, more predictable processes for building platforms.
In hospice in particular, we have written that buyer fit matters more than buyer volume for outcomes; algorithmic matching is built around that observation.
The broader effect is the slow erosion of information asymmetry.
For the last decade, that asymmetry has favored buyers.
As exit certainty becomes knowable before a process begins, the dynamic shifts.
Not in any one transaction. Not overnight. But the direction is clear, and the operators we work with are starting to feel it.
Beyond healthcare
The architectural argument extends past healthcare.
The gains from AI in professional services come less from adding automation to existing workflows than from rebuilding the workflow around what software does well in the first place.
Whether that principle holds in legal services, audit, consulting, and other relationship-driven industries is a question the paper raises but does not answer.
Home-based care is the natural first test.
Its fragmentation, standardized financial reporting, and active consolidation cycle fit the model especially well.
Other professional services will look different.
That is the next research program, not this one.
About Montauk AI
Montauk AI is a home-based care investment bank.
We work with founder-led and mid-market operators across home health, hospice, home care, palliative, and post-acute care, across the full Operate, Optimize, Exit lifecycle.
In Operate, we build the FP&A foundation, KPI infrastructure, monthly close cadence, and board-ready reporting that defines a serious business.
In Optimize, we engineer enterprise value through EBITDA uplift, workforce utilization, clinical and star-rating improvements, payer optimization, and technology enablement.
In Exit, we run the transaction itself through algorithmic buyer matching, CIM and comps strategy, and AI-accelerated execution.
The model is the product of nearly two decades of operating and advising in the sector, combined with AI infrastructure built for the structural realities of home-based care M&A: standardized financial data, branch-level operating metrics, state-specific regulatory environments, and a buyer universe that does not move efficiently through personal networks alone.
Considering an exit in the next twelve to twenty-four months?
The readiness work begins now.
If you operate in home health, hospice, home care, palliative, or post-acute and are thinking about what the next decade looks like for your business, we would welcome the conversation.
Reach Jarrett Bauer at jbauer@montauk.ai, or learn how we engage across Operate, Optimize, Exit at montaukai.com.
FAQ
What is AI-native investment banking?
A model where the M&A workflow is organized around what software does well, ingestion, parallel matching, systematic financial work, continuous execution, with people doing the work that requires judgment, relationships, and negotiation.
The architectural choice is what separates it from automation layered on top of an existing process.
Who is the paper for?
Founders and owners of home-based care agencies thinking about an exit.
Investors and operating partners in the sector.
Policymakers thinking about consolidation.
And practitioners across professional services interested in whether AI-native architectures apply to relationship-driven industries beyond healthcare.
What does the paper claim?
That an AI-native model can compress home-based care transactions from nine to twelve months down to three to five, extend institutional-grade execution to agencies traditional banks do not serve, and create enterprise value before a transaction through pre-transaction operational work.
The claims are tested across four preregistered research questions, with primary analyses replicated by an independent statistician.
What is continuous exit readiness?
The operating principle that financial reporting, KPI infrastructure, leadership depth, concentration profile, and compliance posture are built as continuous discipline rather than as a sprint a few weeks before going to market.
Operators in this state have more leverage when they transact, and can time the transaction to market conditions instead of personal circumstances.
What is the Operate. Optimize. Exit. framework?
The three phases that organize the work.
Operate is the financial and operating infrastructure of a serious business: FP&A, KPI dashboards, monthly close, board-ready reporting.
Optimize is the work of engineering enterprise value before going to market: EBITDA uplift, workforce utilization, clinical and star-rating improvement, payer mix.
Exit is the sale process itself, run through the AI-native platform.
How does Montauk AI handle the conflict of interest in the paper?
The platform studied in the paper is built by the author. The paper names this directly.
Three mitigations are in place: preregistration of every research question and analytical approach before data analysis begins, primary analyses conducted or replicated by an independent statistician, and public deposit of de-identified data supporting the analyses on acceptance.
The limitations and the appropriate scope of causal inference are discussed in detail in the paper.
What is the GrandCare and Pennant Group case study?
In the sale of GrandCare Home Health Services to Pennant Group (NASDAQ: PNTG), Montauk AI served as exclusive placement agent.
AI-driven market intelligence identified and benchmarked over 150 regional operators across the relevant geographies.
GrandCare was positioned as a platform acquisition emphasizing 5-Star clinical quality and 6,000+ patient census across four counties.
The transaction is one of the illustrative cases in the paper, included to show the model operating opposite a publicly traded strategic acquirer.
What is a home-based care investment bank?
An investment bank purpose-built for home health, hospice, home care, palliative, and post-acute care.
The category exists because home-based care has structural realities that general healthcare investment banks are not built around: standardized financial data with sub-segment specifics, branch-level operating metrics, state-by-state regulatory environments, payer mix complexity, and a buyer universe that does not move efficiently through personal networks alone.
Montauk AI is the home-based care investment bank serving founder-led and mid-market operators in this market.
Is there an M&A advisor for home health, hospice, and home care agencies under $450 million?
Historically, no.
Investment banks of any consequence engage above $450 million in enterprise value, which leaves the vast majority of home health, hospice, and home care operators without an institutional option.
Montauk AI was built to close that gap.
We engage with founder-led and mid-market operators across the size spectrum, from agencies preparing for a future sale eighteen to twenty-four months out to operators in active transactions today.
How long does a home-based care transaction take with Montauk AI?
Industry benchmarks for sub-$450 million healthcare deals run nine to twelve months.
The research paper reports time-to-close on AI-native transactions in the three to five month range, with the caveat that the comparison is observational rather than experimental.
Beyond raw timing, the framework allows operators to compress the time between deciding to transact and reaching liquidity, with less operational disruption and the ability to time the process to market conditions rather than to a banker’s calendar.
How do home-based care agencies sell to a strategic or PE buyer?
The process typically runs four stages: readiness, marketing, diligence, and close.
AI-native execution restructures the work inside each stage.
Financials are normalized and adjusted EBITDA bridges are built systematically.
A curated buyer database is scored for strategic, geographic, financial, and regulatory fit.
CIM and comps strategy are developed against actual buyer underwriting models.
Diligence materials are built before they are requested.
Both strategic acquirers (such as Pennant Group in the GrandCare transaction) and financial sponsors are addressed through the same workflow, with positioning adjusted to each buyer type.
Download our AI-Native Investment Banking Research Paper at montaukai.com.