Two returns, underwritten in one diligence.
Life sciences and healthcare is the whole of the mandate, artificial intelligence is the accelerant inside it, and impact is scored on the same evidence as the money.
Return on Investment, squared.
Every company we back has to deliver two returns: a financial return and a measurable impact on patients and the health system.
The dual mandate
Access to care
Companies that change who can be reached, diagnosed and treated, at what distance, at what cost and after what wait.
Healthy life extended
Earlier detection, better-targeted treatment and the biology of aging, so that people gain years of healthy function.
System cost removed
Technology that takes cost out of a care pathway instead of adding a line to it.
Each company is underwritten on the one outcome it actually changes. Figure 2 shows how each outcome is measured.
Two underwritings, one standard.
ROI² holds both underwritings to one standard, including the impact side that most firms leave unmeasured.
Return on Investment
Underwritten the ordinary way and held to the ordinary standard: a company with real buyers at a real price, a commercial engine that can be built rather than hoped for, and a written exit thesis revisited every year whether or not the year went well.
- Named at
- Entry, in the investment memorandum
- Tracked
- On the reporting cadence, against the operating plan
- Owned by
- The investment committee
- Reported
- To investors, on schedule, including the bad quarters
Return on Impact
One measure per company, chosen at entry because it is the thing the company actually changes. It reports on the same schedule as the financials, with the same tolerance for a bad number.
- Named at
- Entry, in the same memorandum
- Tracked
- On the same cadence, by the same people
- Owned by
- The same investment committee, with no separate impact desk
- Reported
- Whether or not it flatters us
How each outcome is measured
Measured as reach
Whether a person who could not previously get the intervention can now get it, counted as people served in a defined population.
Measured as time and function
Whether the intervention adds time, function or independence, tracked against a named clinical or real-world endpoint and evidenced to the standard of a clinical claim.
Measured as cost taken out
Whether the intervention takes cost out of the system that pays for it. A payer or a provider has to be measurably better off, and this is the dimension a payer will test.
Each company names one measure at entry, in the dimension that fits it, with a baseline, an owner and a reporting cadence. It reports alongside revenue, gross margin and cash, and it is never blended into a composite score.
Impact returns are just as important as financial returns, so we measure them the same way.
Venture capital, private equity, and the passage between them.
Venture capital underwrites whether a thing can work. Private equity underwrites whether a thing that works can be optimized. Between them sits the passage in which a proven technology becomes a company, and that passage is the whole of our mandate.
Mid-market growth equity
Venture capital
Underwrites the science.
- Risk
- Technical
- Owns
- The proof
- Horizon
- Open-ended
LeverVenture
Underwrites the commercial inflection.
- Risk
- Execution
- Owns
- The build
- Horizon
- Defined
Private equity
Underwrites the optimization.
- Risk
- Financial
- Owns
- The margin
- Horizon
- Fixed
Figure 3 states the seat, not a claim about any single transaction. Companies arrive at the middle seat from either side, and the work in that seat is operating work.
One mandate, five places it shows up.
Life sciences and healthcare is the whole of the mandate. Inside it we look for the same shape every time: technical risk retired, commercial risk live, and a management team that would take an operator on the cap table over a board observer.
Life Sciences & Healthcare, with AI as the accelerant inside the mandate.
The mandate, stated onceSaving, extending and improving lives: from early detection and diagnostics through to personalized treatment across biotechnology, longevity, and physical and mental rehabilitation.
Artificial intelligence is not a sixth sector on this page and is not a sector in the fund. It is the layer that compresses the timeline inside each of the five below, and it is underwritten that way.
One mandate does not mean one sector. The strongest companies inside the mandate rarely sit in only one: a medical device that is also deep tech, a care platform that is also digital transformation, a clean tech or agtech company built around a health outcome, often with artificial intelligence as the layer that makes it work.
We underwrite the whole company as a single object, not only the half that fits a label. That reach across sectors, on top of the artificial intelligence layer, is part of our moat. About 46 percent of billion-dollar exits between 2023 and 2025 span two or more sectors, and a single-sector fund can diligence only the half inside its mandate.
Five domains
Biopharma & Therapeutics
Therapeutics with human data in hand, where the remaining work is manufacture, trial execution and the commercial organization that has to carry the asset.
Diagnostics & Precision Medicine
Tests that change what a clinician does next. Analytical validity is the easy half; the hard half is clinical utility, reimbursement and the ordering habit.
Devices & Robotics
Cleared or approved hardware short of a commercial organization, where scale-up, supplier qualification and the first reference accounts set the date.
Digital Health & Delivery
Software inside a regulated clinical workflow, where the buyer is an institution, the evidence bar is real, and distribution decides the winner.
Longevity & Neuro
Neurotechnology, brain health and the biology of aging, including physical and mental rehabilitation, where the science is furthest ahead of the business model.
Artificial intelligence runs under all five, not beside them.
Discovery, triage, imaging read, trial design, evidence generation, clinical workflow and the commercial motion itself. We underwrite it where it shortens a specific timeline in a specific company, and we discount it where it is a description of the software rather than a change in the economics.
Domain is a starting filter, never the thesis. Mandate fit is decided on stage, position and the conditions in section 06, in that order.
The middle seat, defined precisely.
Between venture capital and private equity is a description, not a definition. This is the definition: the stage, the position, the governance and the horizon we actually take.
Five definitions
The Fund I facts, and how verified accredited investors request the private placement memorandum, are on the Fund I page. Terms and structure are in the memorandum only. Nothing here is an offer of any security.
Seven conditions, and we do not average them.
Our criteria keep the ROI² alignment honest while holding rigorous financial discipline. They are stated here so a founder can disqualify us as quickly as we can disqualify them.
Seven conditions
All seven, not a majority. A company that misses one is not a smaller position; it is a decline with a reason written down.
The LeverRating. One scale, six dimensions, a stated verdict.
Every opportunity is scored on the same house framework and written up the same way, so two deals a year apart are comparable and a decline is as legible as an approval. The framework declines. That is the point of having one.
Worked example
An illustrative composite
Not a portfolio company
The red tick marks 60, the floor of the conditional band. Below it the house answer is a pass or major conditions, and it is written down as such.
| Dimension | Weight | Score | Weighted | Scale |
|---|---|---|---|---|
| Team | 25 | 60 | 15.0 | |
| Market | 20 | 80 | 16.0 | |
| Product | 20 | 80 | 16.0 | |
| Traction | 15 | 70 | 10.5 | |
| Financial | 10 | 60 | 6.0 | |
| Thesis fit | 10 | 75 | 7.5 | |
| Composite | 100 | 71.0 |
| Score | Rating | House answer |
|---|---|---|
| 90 – 100 | Exceptional | Strong proceed |
| 75 – 89 | Strong | Proceed |
| 60 – 74 | Moderate | Conditional proceed |
| Below 60 | Weak | Pass or major conditions |
A regulated asset gets its own rulers.
For a device, a diagnostic or a therapeutic, the six dimensions expand to ten so that regulatory position, clinical evidence, reimbursement and freedom to operate each carry their own weight instead of being blurred into a single product score. Both rulers are run, and a rating that only holds under one of them is reported as unresolved.
The number is the summary, not the work.
Behind every rating is a written diligence file: the claim, the primary source that settles it, and the open items that would change the answer. Names are never published, and neither are the files.
Subject A is an illustrative composite built to show the instrument. It is not a company, a portfolio position or a live opportunity, and the scores are not a rating of anything real.
What each dimension is actually measuring.
The composite is the summary, not the work. This is what sits behind each of the six numbers, and why the team dimension carries the most weight of any of them.
Weight and meaning
The heaviest weight on the card, and deliberately so. Who has owned the revenue line before, who owns the plan between board meetings, and whether the people who actually close revenue are in the room. A strong asset behind a thin team is the most common decline we write.
The reachable market rather than the addressable one: who the buyer is, what they already spend, what they would have to stop doing, and how long their purchasing cycle really runs.
Evidence, differentiation and the position the product holds once a competitor answers it. For a regulated asset this dimension is the one that splits into four.
Repeatable commercial motion, not a pilot list. We read retention, reorder and the cost of the second sale far more closely than the first.
Unit economics, the runway the plan actually needs, and whether the next round is a financing or a rescue. Discipline here is scored even when the number is small.
Mandate fit on stage, position and the ROI² condition. A company can score well on the other five and still be somebody else’s deal, and we write that down as the reason.
The weights are fixed before a company is scored, not tuned afterward to produce a number somebody wanted. Where a regulated asset is involved, Product expands into regulatory position, clinical evidence, reimbursement and freedom to operate, each carrying its own weight, and both rulers are run.
Our proprietary methodology, sitting behind the diligence team.
One file per company, one evidence rule, one scorecard and one memorandum format. It is the reason a decline written this quarter is legible next to an approval written last year, and the reason the work does not have to start again at closing.
What LeverAI is
LeverAI is our proprietary diligence methodology and the technology that carries it. It sits as the backbone of the diligence team: it holds the file, enforces the evidence rule, applies the same scorecard to every company, and drafts the memorandum out of the record rather than out of anybody’s recollection.
What it is not
No investment is approved by a model. The committee decides, the partners sign the memorandum, and a rating that only holds under one ruler is reported as unresolved. LeverAI is how the work is kept comparable and auditable, not a substitute for doing it.
The number is the summary. The file is the work.
Four layers decide an investment. They do not stop there. What the same instrument carries after the money moves is set out in § 10, execution support.
The instrument does not switch off at the wire.
Most diligence is archived the day it is used. Ours is not. The same file, the same evidence rule and the same scorecard become the operating record of the company, so the work that decided the investment is the work that runs it.
The file becomes the operating record
Nothing is archived at close. The claims, the sources and the reservations carry straight into the board pack, so a plan is argued twelve months later against the same evidence that underwrote it rather than against whatever has since been remembered.
Open items leave closing with an owner
Every item diligence could not settle is carried out of the investment committee with a name, a date and the evidence that would close it. The ones that need an operator rather than an answer are the ones LeverSolutions staffs first.
The scorecard is re-run, not retired
The same six dimensions are scored again on the reporting cadence. A dimension that moves down is the earliest signal the platform gets, and it arrives well before the number does.
What the instrument carries
Diligence ends at the wire. The instrument does not.
One path, and everybody walks the same one.
The sequence below is the whole process. It exists so that two companies evaluated a year apart were held to the same standard, and so that a decline is legible.
Six stages
Closing is the start of the work.
The reason to publish this section is that every firm says it. Saying what specifically arrives, on what cadence and under whose name, is the part that can be checked.
The first hundred days, already written
The operating plan is agreed before terms, not after closing. On day one it becomes the working document: what gets built, who builds it, and which part LeverSolutions carries.
The wrap, against a named scope
Commercialization, operations and supply chain, expert teams and efficient capital use. Each is scoped in writing with a named owner, so the company can hold us to it.
A cadence that survives a bad quarter
Monthly numbers, a quarterly board, an annual rewrite of the exit thesis, and the impact measure on the same schedule. The cadence does not loosen when the news is worse.
Global innovation, commercialized into the United States.
We invest in companies that operate in the United States, and we source the innovation globally into that market. For a company built in Basel or Tel Aviv, the American entry is usually the largest commercial decision it will ever make, and almost never the one it is staffed for.
Sourced globally, operated here
Where the science was done does not decide where the company gets built. Our offices in Basel and Tel Aviv exist so that the sourcing is not limited to what walks into Cleveland or Miami, and the operating seat is held in the market that has to buy the product.
Regulatory and commercial are one decision
A product cleared in one jurisdiction is an unfinished product in this one. Regulatory position, reimbursement pathway and the first reference accounts are underwritten together, because a company that wins two of the three still has nothing to sell.
The passage is staffed, not described
Entity and quality system, supplier qualification, market access, and a commercial hire who has actually sold into American institutions. LeverSolutions carries the named lines. The company carries the product.
Into the US market
Science travels across borders. The company that sells it has to be built in this market.
Geography is part of the mandate, not a preference: United States operating, with innovation sourced globally into that market. It is underwritten at entry alongside the sector fit, and the entry plan is written before terms rather than after them.
What this thesis rules out.
The exclusions are the load-bearing part. They are what makes the rest of the page a commitment rather than a preference.
We do not underwrite unproven science
If the remaining question is whether the biology works, that is a venture question and somebody else is better at it. We enter after the proof and before the engine.
We do not take passive positions
A check with a board observer seat attached is not our product. If a company does not want operators inside the work, we are the wrong partner and we say so early.
We do not treat AI as a sector
Artificial intelligence is the accelerant inside the mandate, never a sixth sector. We underwrite it where it shortens a specific timeline in a specific company.
We do not publish what is not ours
Diligence files, company names and investor relationships are held at the partnership. A logo wall is not evidence and we do not keep one.
A thesis that excludes nothing is not a thesis.