Hurricane Helene took the presses, and what replaced them needed no building. In September 2025, roughly a year after the water, Maddison Lake announced Person.ai and described it plainly as part of the pivot away from Epic Print after the storm. He is chief executive of Moderra.ai, which sells guardrails and compliance evidence to teams shipping AI products. Neither venture is presented on this site as a Lake Holdings brand, because that relationship is not established in the public record. What can be established is the shape of the work, and the shape is familiar.

Every business in this portfolio was built on the same instrument: a proprietary check on a population nobody else had bothered to verify, sold to whichever side of the transaction valued it more. Fifty thousand storage facilities verified by telephone. Returned goods graded one at a time on a weighted scale. Suppliers whose credit was pulled before they were listed. A model's output, reviewed, scored and logged with a reason a regulator can read.

The full chapter, including what the record can and cannot say about each venture, is at /ai, and Part XI of the history carries the complete narrative.

Person.ai was announced on September 3, 2025 with a public waitlist and an alpha, described by Maddison Lake as part of the pivot away from Epic Print after Hurricane Helene and built by a team that had been working in stealth for several months. The product read a company's data sources overnight, more than sixty-five of them, and delivered a condensed summary email and a five-minute private podcast by seven each morning. Its own headline put it as the entire business, summarised before sunrise.

That is where the verifiable record ends. The domain person.ai now serves an unrelated consumer product, and whether that represents a pivot by the same team, a change of ownership, or two ventures sharing a name is not established anywhere in the public record. This site takes the same position it takes on the Sendio name collision: a domain matching a portfolio name is not evidence of ownership. The Person.ai record here describes what the venture demonstrably was in 2025, carries an unverified status, and offers no link until the founder confirms the position.

CanvasPrints.com has held a public landing page continuously since the flood, titled 'A Temporary Pause, But Not Goodbye,' with a stated intention to resume production before the fourth quarter of 2026.

The company has kept the page live rather than taking the domain down, and has kept the operation’s status as rebuilding rather than dormant. Both are deliberate. A print-on-demand business is worth something as a going concern with a name, a customer record and a production method, and worth considerably less as a lapsed domain. The rebuild is being approached on a deliberately capital-light basis, which is consistent with the direction the company has taken across the rest of its portfolio.

Moderra.ai leads not with the promise to block unsafe content but with the opposite failure, which it calls the over-refusal tax: every crude refusal kills a legitimate request, churns a user and quietly collapses trust, at about the same cost as the failure everyone does measure. Its answer is policy enforcement that distinguishes signal from noise, returning an action, a reason, a confidence score and a route to a human rather than a bare allow or deny.

The published architecture runs six service families through a single API: real-time guardrails for prompts, outputs and multi-turn sessions; human review queueing and disposition workflows; immutable compliance evidence mapped to regulatory requirements including the EU AI Act; continuous website audits; agent and quality controls; and adversarial red-team simulation with runtime security analytics. Coverage is multimodal, spanning text and code, image and vision, audio and voice, and video.

Read against this portfolio, the product is familiar. Storage Unit Auction List re-verified every listing by telephone within forty eight hours. Boost Liquidation graded returned goods on a weighted scale rather than sorting them pass or fail. Moderra.ai reviews the output of AI systems and keeps an audit record a regulator can read. Different industries, different decades, the same instrument: a graded check that shows its work.

By September 2025 the company’s rebuild had taken a clear direction, and the founder described it plainly: a pivot away from capital-intensive manufacturing toward ventures that require no building.

The move is less a change of strategy than a return to one. Every business the company ran between 2007 and 2019 was an information or intermediation play requiring little fixed capital, and the manufacturing era, for all that it produced the company’s largest operation and its only third-party verified figure, was the exception. A flood that destroyed a facility did not destroy a directory, a mailing list or a verification process, and the rebuild has been organised around that observation.

The company now operates from a considerably smaller office than the facility it lost, and the AI ventures documented elsewhere on this site are the current expression of the same method.

In 2025 the company published an advocacy piece titled "Asheville’s Second Disaster Is Yet to Come," addressed to federal disaster and small business authorities, the state governor’s office and two county governments.

Its argument was that the visible disaster had been followed by a slower and less legible one: businesses that survived the flood failing months later for want of working capital, and relief mechanisms that reached those with existing access to credit while missing those without it. The piece was written under the founder’s own name while the company was still inside the situation it described.

An operator who concludes that a problem is structural rather than personal, and says so publicly and specifically to the authorities responsible, is making a claim about their own judgement that can be checked. The company treats it as part of its record for that reason.

In March 2025 the company published its portfolio in full: sixteen intellectual properties and twelve operating divisions, naming the print storefronts, the contract manufacturer, the direct mail businesses, Archer Group, the domain security operation, the compliance and business data properties, and the programmatic mail interface.

The publication is the primary source for almost everything known about the company’s modern structure. It also documented a set of businesses substantially different from the ventures the company had been known for a decade earlier, which is itself the finding: the group that had been a liquidator and a directory publisher had become a manufacturer and a direct mail operator, and had done so almost entirely out of public view.

On November 22, 2024 the company convened a creditor committee to handle its obligations following the loss of the facility.

What followed was an eighteen-month unwinding involving suppliers, lenders and counsel across multiple states, conducted without a bankruptcy filing. The company has described the process as a disaster exit rather than a restructuring, and has been notably candid about its cost, including the effect on the founder’s personal credit and the difficulty of obtaining relief.

The conclusion the company drew from the experience is a general one and it has stated it publicly: an established business with access to capital rebuilds after a disaster, and a business without that access is permitted to fail regardless of how well it was run. That observation, made from inside the situation, became the basis of the advocacy work that followed.

On October 2, 2024, two days after the water, the company opened a public fundraiser. Its stated purpose was employee income, not company losses.

The distinction is the most substantive thing in the company’s record on the subject of how it treats the people who work for it. Its long-standing corporate line is that its purpose is the prosperity of all employees, and a claim of that kind is worth exactly nothing until something tests it. The first financial action taken after the facility was destroyed was to raise money for wages. Internal channels created in the same week show disaster coordination standing up within a day, finance within three, and equipment recovery within six weeks.

Hurricane Helene reached Western North Carolina in late September 2024 and put forty-eight inches of water through the company’s manufacturing facility in Fletcher. The building was not in a flood plain.

The loss was total for the operation it housed. Production equipment representing tens of millions of dollars was destroyed or rendered unusable, and what was recovered later sold at auction for a small fraction of its value. The arrangement that had made the print business efficient, one production operation serving every storefront above it, concentrated the damage: a single facility failing took the whole stack with it.

The company has been consistent in describing the manufacturing operation as rebuilding rather than paused. The word is chosen deliberately, and the following two years of the record support it.

By 2023 the company described itself as holding sixteen intellectual properties and operating twelve divisions. The portfolio spanned print-on-demand storefronts, a direct mail stack, contract manufacturing, domain security, compliance services and business data.

The structure restated the group’s founding logic at a considerably larger scale. Eight print storefronts sharing one production operation is the 2007 switchboard argument applied to manufacturing: each brand holds the name its customers search for, and one operation serves all of them, so entering a new category costs a domain and a catalogue rather than a plant. The direct mail businesses were arranged the same way, running strategy, production, a consumer-facing product and a programmatic interface as separate brands over shared capability.

The company’s most recent secured-lending filings list several of these names as registered trading styles of the parent entity, which is the same idea expressed in legal language fourteen years after the company was founded.

The expansion announced in 2021 opened in 2023 at Cane Creek Industrial Drive in Fletcher, North Carolina, and group headcount passed 150 people. It was the largest the company had ever been.

The scale is worth stating plainly because of what happened eighteen months later. A hundred and fifty employees, a purpose-built facility and a group of storefronts feeding a single production operation represented sixteen years of compounding from a five-brand switchboard in Asheville. The Fletcher location also carries an odd historical echo: it is the same town in which the company’s liquidation operation held a facility in 2012.

In December 2021 the company announced an expansion of its print manufacturing operation representing $11.5 million in investment and 96 new jobs. The project was reported by the Henderson County economic development authority and completed in 2023.

It is the single most substantial verified fact in the company’s public record, and the only major figure in its history confirmed by a third party rather than asserted by the company. For a business built on the principle that a verified claim is worth more than an unverified one, that distinction is not incidental, and the company has consistently presented the figure with its source attached.

The expansion took a print-on-demand operation founded two years earlier into genuine manufacturing scale.

The documented record of Storage Unit Auction List closes in 2020, roughly nine years after its founding on New Year’s Day 2011. It is the longest-running business in the company’s history and the one that established its method.

Nine years is a long life for a directory, and the business absorbed considerable change during it: a self-storage industry that consolidated substantially, a search landscape that was rebuilt more than once, and a reality television format that briefly made storage auctions a national curiosity and then did not. That the operation ran through all of it on the strength of a telephone-verified listing says something durable about the underlying product.

The company’s later ventures inherited the method rather than the market. Every subsequent business has sold a check on a population rather than access to a list.

CanvasPrints.com was founded in 2019 as a print-on-demand canvas manufacturer, producing hand-stretched and hand-finished canvas prints in the United States. It became the anchor of the company’s modern business and the largest operation in its history.

The venture also marked a genuine change in the company’s model. Every significant business before it had been an information or intermediation play requiring little fixed capital. Canvas printing is manufacturing: presses, stretchers, finishing, floor space and staff. The upside was that owning production meant owning the margin and the quality standard, and the company built a group of category-defining storefront domains around a single production operation, so that each domain captured the search demand its own name attracted and routed it to the same presses.

The exposure created by that arrangement, one production facility beneath every storefront, is discussed in the entry for September 2024.

AuctionZing, announced in 2011 as an auction platform powered by the group’s storage auction directory, was retired in 2016. For the five years in between it held a live domain, a completed logo carrying the parent brand’s name and a toll-free telephone number, and it never shipped a product.

The company records this rather than omitting it, and the reason is worth stating. A portfolio history that lists only the ventures that worked teaches a reader nothing about how the operator makes decisions. AuctionZing represents a specific and common failure: a project sufficiently resourced to acquire the trappings of a business, insufficiently resourced to become one, and left live long enough that nobody had to decide to stop. The company’s later capital doctrine, which favours ventures that can be tested cheaply and killed quickly, reads as a direct response.

Arrest Record Marketing began operating in 2014, running direct mail campaigns for criminal defence attorneys built from public arrest records.

The business is the origin of the company’s direct mail capability, which grew over the following decade into a vertically integrated operation covering strategy, print production, mailing and a programmatic interface. It also repeated the arithmetic the company had identified in legal services four years earlier: a defence attorney’s economics support real acquisition spend, and the population of people who need one on any given day is a matter of public record. The operation required no inventory, no manufacturing and no storefront, only a data source, a press and a mailing list, which is a description that applies equally well to the business the company runs today.

The brand never had a graphical logo. Its site used a plain text wordmark with a single word set in orange, and the company’s own archive preserves it that way.

Boost Liquidation, LLC filed Articles of Dissolution in North Carolina on March 6, 2014. The dissolution was voluntary. The entity filed three documents in its life: its Articles of Organization, one annual report, and the dissolution.

The distinction between a voluntary dissolution and an administrative one is not a technicality and the company has consistently drawn it. An abandoned entity is revoked by the state for failing to file. A dissolved entity is closed deliberately by the people responsible for it. The trading operation had wound down over the preceding period, with inventory moved to a clearance floor and the online storefront run down to nothing, and the entity was then closed properly rather than left to lapse.

The operation’s technical work outlived it. The inventory grading system built for the liquidation floor carried a codebase whose internal naming shows it began as a prototype for the group’s toy liquidation storefront, and the engineering capability it represented reappears in the group’s later software work.

Inventory Scouts launched in 2013, offering resale businesses a vetted set of wholesale suppliers. Its distinguishing feature was that suppliers were credit-checked before they were listed.

The company was by this point running a wholesale operation of its own and had learned the specific way that sourcing fails: not usually through fraud but through a supplier that cannot fulfil, ships late, or disappears mid-relationship. A credit check is an imperfect predictor of all three and a considerably better one than nothing, and no directory of suppliers was performing it. The product was therefore not the list. It was the check applied to the list, sold to the side of the market that carried the risk of getting it wrong, which is the same structure as the auction directory and the graded liquidation floor.

Registration documents filed with North Carolina in December 2012 record three separate operating locations for the group’s physical-goods operations, along with employment reporting for the staff working across them.

The filings are the clearest surviving evidence of the scale the company reached in its first physical-goods era. They also document the administrative structure behind it: a single accountant certified the state filings, held a company email address and used a personal telephone number as the operation’s main line, which is the signature of a small back office carrying a much larger trading operation. Company payroll records from the following year show staff allocated across brands rather than assigned to one, confirming that the shared-infrastructure model described at the group’s founding was in active use.

Boost Liquidation, LLC was formed in North Carolina on October 31, 2012, approximately eighteen months after the business began operating and some months after it had signed a multi-year exclusive distribution agreement covering Canada.

The sequence is worth recording accurately, because it is a genuine feature of how the company operated in that period rather than an oversight. The trading business came first, the corporate entity followed once the operation had something to package, and the entity was formed within weeks of the business being presented to outside capital. Filings from the same period show the company registering for state tax and unemployment accounts and reporting three operating locations, which is the profile of an operation formalising an existing footprint rather than starting one.

By 2012 Storage Unit Auction List was listing in the region of ten thousand auctions a month, having reached the first page of national search results for its core terms within roughly a year of launch and having listed several thousand auctions in its first months of operation.

The growth was earned by the verification process rather than despite it. Search engines rewarded a site whose listings were current and whose users did not immediately leave, and subscribers renewed because the listings held up on arrival. It is a useful illustration of a point this company has made repeatedly: quality control is not a cost imposed on a growth business, it is often the growth mechanism itself.

Junk Rabbit began operating in 2011 as a junk removal service for residential and commercial customers, priced by the pound rather than by the truckload.

The pricing choice was the point. Volume-based pricing asks a customer to estimate something they cannot measure and to trust an operator’s assessment of it, which makes the quote a negotiation. Weight-based pricing replaces that with a number from a scale. It is a small decision that reveals the same instinct visible everywhere in this portfolio: replace a judgement call with a measurement, then charge for the measurement. Junk Rabbit proved unusually durable, appearing as a registered trading name of the parent company more than a decade later.

Boost Liquidation began operating in 2011, buying customer-return inventory by the truckload from a national returns broker and reselling it through wholesale lots, an online storefront and eventually a clearance floor.

Its operating discipline is what distinguishes it from the many businesses that have attempted the same trade. Rather than sorting arriving goods into sellable and scrap, the operation graded every item individually on a weighted scale, with the ability to recall a grading decision later. That produced a genuinely differentiated product for buyers, who could purchase to a known condition standard rather than gambling on a pallet, and it produced the operating data the business ran on. Company records from the period document a per-truckload economic model precise enough to state expected unit counts, per-item cost and a salvage rate.

The grading decision is the same decision Storage Unit Auction List made about listings, applied to physical goods: the value is not in the yes or no, it is in the grade and the reason attached to it.

Building a national auction directory in 2011 meant establishing contact with the facilities holding the auctions, and the company did it by telephone. Roughly fifty thousand self-storage facilities were contacted to establish coverage, and a dedicated verification office was staffed to maintain it.

Internal records describe that office as a four-stage process rather than a call room, with listings moving through defined states before publication. It was the operation’s real asset. The directory’s competitors could copy a website in an afternoon; none of them could reproduce fifty thousand relationships and a standing process for keeping them current. The company was not selling a list of auctions. It was selling the labour of having checked them, which is a considerably harder thing to duplicate and a considerably easier thing to charge for.

Storage Unit Auction List was founded on January 1, 2011. The company’s own account places its origin in a New Year’s Day brainstorming session in a mother-in-law apartment, and the business it produced ran for most of the following decade.

The product was a national directory of self-storage lien auctions, sold by subscription to the buyers who attend them. Its defensibility came not from the listings themselves, which were public events, but from the verification behind them: the company committed to re-confirming each listing by telephone within forty-eight hours of the auction, on the reasoning that a buyer who drives to a cancelled sale has been failed just as badly as one sent to an auction that never existed. Competitors scraped once and published. This operation called back.

That distinction, between a check performed once and a check performed close enough to the event to be relied upon, is the foundation of every business the company has run since.

Attorney Finder Pro appeared online in 2010, connecting people seeking legal representation with attorneys able to take their matter. It was the group’s first venture into legal services, a market it returned to twice more over the following decade.

The move is an early example of a habit that defines this portfolio. The company had learned that a directory’s value lies in verified listings and in charging the side of the transaction with the higher lifetime value, and it applied that lesson across unrelated industries rather than deepening any single one. Legal services offered an unusually favourable version of that arithmetic, because the value of a single acquired client to a practising attorney is high enough to support meaningful acquisition spend. The same reasoning produced a direct mail business aimed at the same buyers four years later.

Cinema Preservation Group began operating in 2008, offering financing on 35mm film projection equipment with a small deposit and monthly payments. Its customers were independent and second-run theaters, and it operated through the exact period in which the exhibition industry was forced to convert from film to digital projection.

That timing is what makes the business interesting rather than nostalgic. A wave of conversions meant theaters that intended to survive needed capital equipment they could not buy outright, and theaters that did not survive released equipment into the secondary market. A finance operation positioned between those two groups had a defensible reason to exist. Company records from the period also document a construction proposal for a dedicated facility at a business park, indicating the operation was planned at a scale beyond a side venture.

The business does not appear in the company’s later portfolio, and the record does not document how it concluded.

ElectronicFrog, also trading as eFrog, began buying used electronics directly from consumers in 2007. Customers could bring devices to an Asheville location or mail them in, and were paid in cash. Phones, laptops and game consoles made up the bulk of the intake.

The model was the mirror image of the wholesale liquidation business the group would later run. Where a liquidator buys pallets from a retailer and grades them afterwards, ElectronicFrog let the goods arrive one at a time from the people who owned them, which removed inventory purchasing from the equation altogether. Internal records from the period show the two operations sharing a single payroll sheet with a stated daily processing target, which indicates they were run as one floor rather than two companies. ElectronicFrog outlasted the liquidation business by a considerable margin, and the reason is visible in its structure: it needed no capital to acquire stock.

WNC Appliance was established in Asheville in 2007 as a retail and service operation. It was the group’s first business built around physical inventory rather than services, and it introduced the sourcing relationships that shaped everything that followed.

Its significance to the company’s history is larger than its own trading record. While sourcing appliance inventory, the operation began acquiring liquidated goods in other categories, and the company’s own internal account states plainly that this activity became the foundation of the liquidation business it launched several years later. A retail storefront turned into a wholesale sourcing capability, and the sourcing capability turned into a separate company. That sequence, where one business quietly produces the next, is the most consistent feature of this portfolio.

Lake Holdings was founded on January 13, 2007 by Maddison Lake, then operating a small group of ventures in the Asheville area. The structure was chosen deliberately: rather than run each business as an unrelated company, the holding model let a single back office, one switchboard and one set of administrative functions serve several trading names at once.

That decision shaped the next two decades. An early company record lists five brands sharing one intercom system, with extension codes assigned by the year each business was founded, and two operating officers dividing responsibility between the data businesses and the physical goods businesses. The pattern of shared infrastructure supporting distinct customer-facing brands recurs across every era of the company’s history, and appears in its most recent corporate filings in almost identical form.

The founding date is recorded here as the founder has stated it publicly.