Insights
Recent thinking.
Articles, field notes, and perspectives from the Lanthos team on the AI work that matters most. We write what we believe, and revise when the evidence changes.
The archive, by category.
93 recent pieces
Logistics & Transportation
View Logistics & Transportation library
Autonomous trucking is a liability question
The technological feasibility of autonomous trucking often overshadows its fundamental challenge: establishing a clear, universally accepted framework for liability when the unexpected occurs. This question shapes adoption more profoundly than any engineering feat.
Read
The load board is becoming an agent market
The load board, once a simple digital marketplace for freight, is evolving into a more sophisticated, agent-driven ecosystem, demanding new strategies for participants.
Read
Your ETA model is a promise you did not price
The ubiquitous ETA represents a binding commitment. Rarely, however, is the full cost of consistently meeting that promise accurately factored into operational expenditure, leading to systemic underpricing and unforeseen liabilities.
ReadWarehouse robotics stopped being a hardware bet
The future of warehouse automation is no longer defined by the physical capabilities of robotic hardware, but by the sophistication of the software ecosystems that orchestrate diverse, adaptable fleets.
ReadTariff volatility is a routing problem now
Tariff volatility has transcended a purely economic concern; it is now a fundamental challenge in the physical and digital design of global supply chains.
ReadFreight fraud scales faster than freight
The digital acceleration of logistics, intended to streamline global movement, has inadvertently created new vectors for sophisticated fraud, outstripping the industry's traditional defences.
ReadThe driver shortage was a scheduling failure
The perceived crisis of driver scarcity is not a fundamental lack of individuals, but a symptom of systemic inefficiencies in how resources are planned, allocated, and managed.
ReadLast mile economics do not survive averages
The economics of last-mile delivery defy simple aggregation, demanding a granular understanding of costs and opportunities hidden within each unique delivery context.
ReadPort congestion is a data-sharing standoff
Persistent port congestion reveals a fundamental reluctance to share operational data, hindering collective efficiency across an interconnected global supply chain.
ReadCold chain telemetry is an evidence problem
True cold chain integrity demands more than just data collection. It requires a systemic understanding of how information is generated, validated, and integrated across fragmented supply chains.
ReadAsset & Wealth Management
View Asset & Wealth Management library
The great wealth transfer is an onboarding problem
The anticipated generational shift of capital demands a fundamental re-evaluation of how wealth management firms cultivate and integrate new client relationships, moving beyond mere transactional engagement.
Read
Private markets in retail wrappers change the liquidity math
The increasing availability of private market investments to retail investors fundamentally alters the traditional calculus of liquidity, necessitating a re-evaluation of product design, regulatory frameworks, and investor expectations across the asset and wealth management landscape.
Read
Fee compression is finally an operations question
For too long, fee compression in asset and wealth management has been viewed primarily as a revenue or investment challenge. The enduring reality is that it has always been, at its core, an operational one.
ReadYour advisor copilot needs a suitability record
For AI copilot tools in asset and wealth management, technical robustness alone is insufficient. Proving an advice record, rooted in client suitability, demands a different architectural imperative.
ReadModel portfolios are eating discretionary mandates
The shift from bespoke discretionary mandates to industrialised model portfolios fundamentally reshapes how wealth is managed, demanding a critical re-evaluation of established operational and client engagement paradigms.
ReadPersonalization at scale is a data lineage problem
True personalization at scale for asset and wealth managers remains an elusive ambition, not for want of data or algorithms, but due to fundamental inadequacies in understanding and governing data's journey and transformations.
ReadDirect indexing moved the work into tax
Direct indexing, often heralded as a technological leap for bespoke portfolio construction, fundamentally shifted the locus of operational complexity and value creation from investment management to tax optimisation.
ReadAlternatives reporting is still a PDF pipeline
For all its sophistication, much of the alternatives investment landscape still relies on antiquated data pipelines, leaving firms to manually extract insights from static documents. This bottleneck impedes efficiency and masks true portfolio dynamics.
ReadResearch alpha decays the moment it is machine-readable
The instant an insight becomes digitised and systematically consumable, its inherent value begins to erode. This fundamental truth demands a strategic re-evaluation of how research is generated and leveraged within asset management.
ReadThe next RIA rollup will be won on integration
The proliferation of RIA rollups has created an industry of acquirers, yet few have truly mastered the delicate art of organisational and technological fusion. The next generation of success hinges not on acquisition volume, but on profound integration.
ReadBanking & Capital Markets
View Banking & Capital Markets library
Deposit stickiness is now measured in minutes
The digital era has irrevocably altered the fundamental nature of bank deposits, transforming them from stable, long-term liabilities into volatile, near-instantaneous liquidity pools. Traditional models for managing this essential resource are now fundamentally obsolete.
Read
Real-time payments moved fraud upstream
The swiftness of real-time payments, a cornerstone of modern finance, inadvertently shifted fraud detection challenges upstream, demanding a fundamental re-evaluation of financial crime prevention paradigms.
Read
Your AML alerts are a labeling budget
Alerts in AML are not merely indications of potential illicit activity; they are a critical input for training and validating the next generation of financial crime detection systems.
ReadCredit models inherit the last cycle's assumptions
Credit models, often rooted in historical data and established paradigms, inherently carry forward assumptions from prior economic cycles, subtly influencing risk perception and capital allocation decisions today.
ReadStablecoin settlement is a treasury decision
Stablecoins are often viewed through the lens of digital asset strategy or regulatory compliance. Yet their true disruptive potential lies within the treasury function, necessitating a fundamental re-evaluation of liquidity management and settlement protocols.
ReadT+1 exposed the reconciliation debt
The swift shift to T+1 settlement has not merely compressed timelines; it has laid bare the extensive, often unacknowledged, debt accrued in operational reconciliation practices across banking and capital markets.
ReadThe trading desk copilot needs a surveillance trail
As AI copilot systems proliferate across trading desks, the imperative shifts beyond mere compliance; it demands an intelligent, continuous surveillance trail that understands context and intent, not just keystrokes.
ReadCore modernization fails on the reporting layer
Despite significant investment, core banking modernization often founders on the reporting layer, where outdated paradigms clash with new data architectures, hindering true agility and insight.
ReadBasel endgame is a data quality program
The impending Basel framework revisions, often framed as a capital calculation exercise, fundamentally expose and demand a radical transformation in how financial institutions manage and understand their data.
ReadEmbedded finance distributes your risk appetite
Embedded finance fundamentally reconfigures the traditional risk landscape for banks, requiring a strategic shift from direct ownership to intelligent distribution and calibration of risk appetite.
ReadInsurance
View Insurance library
A claim is a decision, not a task
Claims is rightly called the highest-leverage workflow in insurance to automate — straight-through processing, lower costs, faster settlement. But the framing treats a claim as an administrative hurdle, and a claim is not a task. It’s a decision about money, made under uncertainty, with two opposite ways to be wrong — and the scoreboard that justifies the automation can only see one of them.
Read
Insurance assumes a person
Putting agents into underwriting, claims and eligibility is where the money is, and the firms moving on it are reading their own economics correctly. But every instrument that sits underneath those functions was built around an assumption nobody wrote down — that a person decided. Remove the person and the machinery doesn't break. It just stops having a subject.
Read
Social inflation is outrunning your reserving cadence
The conventional cadence of actuarial reserving, calibrated for historical patterns, is increasingly misaligned with the accelerating, often non-linear forces of social inflation. This systemic lag demands a fundamental recalibration of both methodology and mindset.
Read
Parametric cover is a data contract
Parametric insurance, fundamentally, is a contract defined by data. Its true transformative potential lies in understanding and orchestrating the underlying data streams that trigger payouts, rather than merely adjusting risk calculations.
Read
Submission intake is where the underwriting margin leaks
The submission intake process, often seen as a mere administrative prelude, is in fact a critical juncture where inefficiencies compound, directly eroding underwriting margins before a risk is even assessed.
ReadClimate-driven repricing is a communications problem
The insurance industry's climate-driven repricing is not merely a technical challenge of actuarial models or capital allocation; it is fundamentally a failure to adequately communicate a profound societal shift.
ReadCyber underwriting needs telemetry, not questionnaires
The current reliance on static questionnaires for cyber underwriting yields an incomplete and often misleading risk profile. A shift towards dynamic, continuous telemetry offers a more accurate, actionable, and sustainable path forward.
ReadThe MGA boom is an outsourced control problem
The proliferation of Managing General Agents represents not merely an expansion of market opportunity, but a fundamental shift in control dynamics that insurers must acutely understand and strategically address.
ReadLife insurance underwriting quietly went accelerated
Underwriting in life insurance has undergone a subtle, profound shift. The quiet acceleration, driven by data and algorithmic advancements, demands a fundamental re-evaluation of established operational paradigms.
ReadClaims leakage is a routing decision, not a fraud one
Claims leakage is often misconstrued as a problem of fraudulent intent. Instead, it frequently arises from systemic inefficiencies in how claims are processed and routed, highlighting a critical operational challenge rather than solely a moral failing.
ReadDistribution is consolidating faster than carriers
The rapid consolidation of insurance distribution channels, driven by capital and technological shifts, is outstripping the pace of carrier integration, fundamentally reshaping power dynamics within the industry.
ReadYour loss run is the most valuable asset you do not use
For too long, the insurance industry has viewed its accumulated loss data merely as a record. This essay posits it as a strategic, generative asset, waiting to be unlocked.
ReadPrivate Equity
View Private Equity library
Dry powder is not the constraint. Exit paths are
The private equity industry often laments its substantial 'dry powder', viewing it as an operational challenge. Yet, the more profound impediment to value creation and deployment lies not in capital availability, but in the increasingly congested and unpredictable landscape of viable exit strategies.
Read
Continuation vehicles moved the valuation argument
Continuation vehicles have reshaped how private equity assets are valued, pushing beyond traditional liquidity discounts and introducing new dynamics for pricing, governance, and capital allocation across portfolios.
Read
AI diligence is mostly a data room quality test
True AI diligence in private equity is less about evaluating sophisticated models and more about assessing a target company's fundamental data maturity and readiness for an AI-powered future.
ReadValue creation teams are becoming platform teams
The operational capabilities within private equity firms are evolving beyond mere value addition, establishing shared infrastructure and expertise that fundamentally redefine the relationship between portfolio companies and their sponsors.
ReadRoll-ups break on the second integration, not the first
The conventional wisdom posits that integration challenges are front-loaded in private equity roll-ups. The deeper truth is that systemic failures often manifest only upon the second, more complex layer of integration, subtly undermining the entire thesis.
ReadPortfolio reporting is still assembled by hand
Despite advancements elsewhere, private equity portfolio reporting largely remains a bespoke, manual craft. This persistence of hand assembly undermines efficiency, constrains insight, and ultimately limits strategic agility.
ReadThe AI thesis needs a defensibility test
For private equity, an AI thesis requires a rigorous examination of enduring competitive advantage, moving beyond mere technological adoption to scrutinise true defensibility.
ReadPrivate credit underwriting inherits the sponsor's model
The rapid growth of private credit necessitates a critical examination of its underwriting practices, which often mirror, rather than independently assess, the financial sponsor's investment thesis.
ReadOperating partners are being asked to prove attribution
The increasing sophistication of capital deployment and portfolio management necessitates a clearer, more rigorous demonstration of value created by operating partners, moving beyond mere influence to quantifiable impact.
ReadFund administration is the quiet automation prize
The operational core of private equity, long overlooked as merely administrative overhead, presents a profound opportunity for strategic automation, yielding efficiencies and insights previously unattainable.
ReadReal Estate & Mortgage
View Real Estate & Mortgage library
Insurance cost is now a cap rate input
Escalating insurance costs are no longer a mere operational expense but a fundamental determinant of property valuation, profoundly reshaping acquisition strategies and portfolio resilience across real estate sectors.
Read
Office conversions are an underwriting fiction until financed
The envisioned transformation of vacant office blocks into vibrant residential or mixed-use spaces remains largely theoretical until the complex financial and operational realities are rigorously underwritten and funded.
Read
Mortgage origination cost per loan is a workflow verdict
Mortgage origination costs are often perceived as a fixed externality, yet they are a direct outcome of an organisation's operational design and strategic choices. This essay argues for a fundamental re-evaluation of workflow as the primary determinant.
ReadAppraisal modernization moved the bias question
Appraisal modernization aims to streamline valuation, yet its reliance on data and algorithms introduces new, often subtle, dimensions to the enduring challenge of bias in real estate.
ReadServicing transfers are where the compliance risk lives
The operational complexities of mortgage servicing transfers create a unique nexus of compliance vulnerabilities, often underestimated until regulatory scrutiny or consumer detriment arises. Leaders must recognise this inherent fragility.
ReadTitle work is document understanding with legal consequences
Title work, at its core, is a sophisticated exercise in document comprehension, where misinterpretation carries profound legal and financial ramifications for all parties involved.
ReadBuild-to-rent is an operations business wearing a real estate costume
The build-to-rent model, often perceived through a traditional real estate lens, fundamentally demands an operational mastery akin to complex service industries, challenging conventional investment and management approaches.
ReadProperty data is fragmented by design
The persistent fragmentation of property data is not a mere technical deficiency; it is a structural artefact, deliberately maintained through entrenched incentives and a complex ecosystem of power dynamics.
ReadRate lock behavior broke your pull-through model
Mortgage rate lock behavior, once a predictable variable, has shifted profoundly. This disruption has rendered traditional pull-through models inadequate, necessitating a fundamental reassessment of pipeline forecasting and risk management.
ReadData centers are the new industrial underwriting
Real estate investment in data centres requires a fundamental shift in analytical frameworks, moving from traditional property metrics to a deep understanding of digital infrastructure's intricate, interconnected, and rapidly evolving operational demands.
ReadCX Operations
View CX Operations library
Deflection is not a customer outcome
Contact centers report deflection as a win. It measures a contact that did not reach a human — not a problem that got solved. The two diverge in exactly the cases that cost the most.
Read
Average handle time is the wrong contract
AHT was a staffing input that quietly became a performance target. Once agents optimise against it, the operation buys speed with re-contacts and pays for them in a different budget line.
Read
The voice agent is a routing decision
Most voice AI programs are scoped as a channel replacement. The valuable version is narrower: an agent that decides, in the first thirty seconds, what kind of problem this is and who should own it.
Read
QA samples two percent and calls it quality
Traditional contact center quality assurance reviews a statistically meaningless sample, scores it against a rubric written for a different era, and produces a number nobody uses to make a decision.
Read
Escalation is the product
Every CX operation designs the happy path and improvises the escalation path. Customers form their durable opinion of you in the second one.
ReadConstruction Management
View Construction Management library
The submittal log is a decision queue
Submittals and RFIs are treated as document administration. They are a queue of decisions with schedule consequences, and they are almost never managed as one.
Read
Change orders are an evidence problem
Contractors lose change order disputes they were entitled to win, because the entitlement existed in the field and the evidence never made it into the record.
Read
Schedule risk is known weeks before the slip
The information that predicts a milestone slip is almost always present in the project weeks early. It is distributed across people who each see one piece and never assemble it.
ReadFoundation & Structural Repair
View Foundation & Structural Repair library
Foundation repair is an inspection-to-proposal business
In residential foundation, basement waterproofing, and concrete lifting, the economics are decided between the inspection and the signed proposal — not in the field crew productivity everyone measures.
Read
Multi-branch service networks fail at the second acquisition
Roll-ups in residential structural and specialty trades survive the first acquisition on goodwill. The second one exposes whether the operating model was ever transferable.
Read
The first sixty seconds decide the foundation job
In residential structural repair, the contractor who answers first books the inspection. Everything after that — crew skill, pricing, warranty — is competing for a slot that was already assigned.
Read
Drive time is the real capacity constraint
Branch leaders hire inspectors and crews to add capacity. A large share of the capacity they already own is sitting in a truck between two appointments that should never have been sequenced together.
Read
The proposal should leave with the inspector
The gap between an inspection and a priced, readable proposal is where foundation repair revenue leaks — and it is almost entirely a document assembly problem, not a judgment problem.
ReadAI governance & risk
7 pieces
An agent with a wallet
The case for agentic commerce is being made loudly: agents that don’t just recommend but pay, settling on rails engineered to be final. It’s framed as the natural endpoint of autonomy. But a payment is the least reversible thing an agent can do — and “irreversible by design” is a strange property to want in a system you’ve just handed your judgment to.
Read
The double-edged arms race
Everyone frames AI as an arms race, and the framing is only half right. The real danger isn’t that the same capability arms both sides — it’s what the race does to judgment.
Read
Shadow AI is a requirements document
The security teams raising the alarm about unapproved AI tools are right about the exposure, and the instinct to inventory, block and police is the same instinct that handled shadow IT correctly. But shadow IT was a procurement problem, and this is not. What employees are routing around now is not a purchasing process — it is a judgment your sanctioned system makes badly, and the standard response erases the only record you had of where.
Read
The deadline moved — your exposure didn't
Compliance teams spent eighteen months building toward a hard date under the EU AI Act, and when the timeline moved, relief was the honest reaction — the work was real and the cost of it was real. But the deadline was never the exposure. It was a schedule for confronting an exposure that sits in production right now, on the old calendar or the new one.
Read
AI browser agents inherit everything
The browser agent is the most immediately useful agentic AI anyone has shipped, and the productivity case for it is the most honest one in the category. But the security conversation around it is aimed one level too low. An agent working inside an authenticated session is not software you deployed — it is a credential you issued, to a colleague nobody onboarded.
Read
Prompt injection is not a bug
The security community is right about prompt injection, and the defensive work around it is serious engineering that reduces real risk. But it is being treated as a vulnerability, and a vulnerability implies a patch. The property that makes an agent exploitable is the same property that makes it worth having.
Read
Integration was the checkpoint
Standardising the way agents connect to internal systems was a genuine advance, and we would not want it undone — integration used to be the whole cost of an agent project. But the cost was doing a second job nobody assigned it. It was the only thing forcing anyone to ask what an agent should be able to reach, and AI agent identity is now the register almost no organisation keeps.
ReadEvaluation & measurement
5 pieces
Calibration data is the moat
The defensible position in enterprise AI was never the model. It’s the calibration — the accumulated record of what worked, under what conditions, after how many corrections — that a competitor can’t buy at any price. Most organizations are handing theirs to vendors without noticing it was the asset.
Read
Automation inherits the metric, not the goal
The 2026 wave of contact-center automation is real, and for most of the work it’s good. But automation doesn’t pursue your goal — it pursues your metric, relentlessly and at scale. And it removes the human agent who used to quietly bend a bad metric back toward what you actually wanted. Before you automate the contact center, be sure the number you’re about to optimize is the one you meant.
Read
How to build an eval that survives production
We argue constantly that you should measure the decision, not the deployment. Fair to ask in return: what does a good measurement actually look like? The constructive half — how we build an evaluation for a consequential decision.
Read
The value of AI isn’t a number
The move from promise to impact was the right correction — and it stopped one step short. Value isn’t a total a company can sum; it concentrates in a few consequential decisions.
Read
Grading the wrong exam
The number making the rounds on enterprise AI ROI is directionally true, and the executives who took it seriously and started asking harder questions did exactly the right thing. But it grades pilots as investments, when the honest function of a pilot is to produce information. By that standard, it counts learning as loss.
ReadInsurance AI operations
2 pieces
Brittle was a feature
The migration off brittle rule-based RPA onto AI automation that learns and self-heals is the right move, and we don’t dispute the upside. But in the migrations we watch, the thing that bites teams isn’t the part that failed to work. It’s the part that kept working when it shouldn’t have.
Read
You automated half a job
The roles being automated genuinely are automatable and the savings are genuinely real — we help clients capture them. But those desks were doing two jobs, and only one of them was ever written down. The second was a continuous, unmeasured correction of everything that arrived slightly wrong, and when the role goes, that work doesn't transfer anywhere. It stops.
ReadAI strategy & economics
4 pieces
Why “AI strategy” is usually the wrong question
Most organizations asking for an AI strategy don’t need one. They need a way of deciding — a small set of questions asked per decision, not a plan written once and outdated by spring. This is the frame the rest of our thinking expands from.
Read
Autonomy is a delegation problem
The agent revolution is real: systems are moving from advising to acting. But the framing — autonomy as doing more with less human friction — hides what actually changes. The moment a system acts instead of advising, it is making decisions. You haven’t bought a faster worker; you’ve delegated authority — and everything hard about agents lives there.
Read
AI model selection is a habit, not a decision
Reaching for the largest available model is a sound answer to a question you cannot yet answer any other way, and we would rarely talk a client out of it at the start. But the default outlives the uncertainty that justified it. What began as insurance quietly becomes the reason nobody ever measured how well the system works.
Read
The assumption nobody wrote down
Building the case on falling AI inference cost was the empirically supported call, and we would rarely have argued against it. What concerns us isn't the assumption. It's that it was never written down as one — so nobody has worked out what breaks if it stops holding, and nobody owns it.
ReadStay current
New thinking, delivered when it's ready — not on a schedule.
Lanthos publishes when we have something worth saying. If you would like to receive our pieces as they are released, write to us.
Get in touch