Ashcroft & Thorne

Applied Intelligence

Something fundamental has changed in how complex work gets done — and most organizations haven't caught up. We have.

Est. MMXXVI
IntelligenceA Field PublicationVol. I · MMXXVIAshcroft & Thorne

Intelligence

Sector assessments and field observations from the practice of applied intelligence. Not case studies — patterns we have observed, problems worth understanding, and the narrow places where leverage actually lies.

11Briefs
MMXXVI
In this issue
  1. 01Enterprise AI ReadinessPre-Acquisition AI Assessment: What Private Equity Misses
  2. 02Geospatial IntelligenceComputed Place Intelligence: Beyond the Comparable
  3. 03Legacy SystemsWhen to Bridge and When to Rebuild
  4. 04Telecommunications InfrastructureThe Catalog and the Carrier: Reading Frontier Telecom Assets
  5. 05Emerging Market OperationsOffline-First Architecture Is Not a Compromise
  6. 06Philanthropic & Development CapitalReading the Funder Landscape: Where Capital Actually Waits
  7. 07Healthcare DataInteroperability Is Not a Technology Problem
  8. 08Fraud & Anomaly DetectionPattern Recognition at the Margin
  9. 09Supply ChainProvenance Tracking: From Compliance Burden to Competitive Moat
  10. 10Climate & CarbonVerification Is the Bottleneck in Carbon Markets
  11. 11Defense & IntelligenceData Architecture as Strategic Advantage
Brief No. 01 / 11Enterprise AI Readiness

The Brief

Pre-Acquisition AI Assessment: What Private Equity Misses

Most PE firms evaluate AI readiness by asking whether a target company "uses AI." The more revealing question is whether the company's data architecture can support it. Legacy SaaS platforms built on deprecated frameworks and fragmented databases require foundational middleware before any intelligent layer becomes viable. The gap between "AI-ready" marketing and actual infrastructure readiness is where valuations break down — and where the real opportunity lies.

Brief No. 02 / 11Geospatial Intelligence
Aerial view of agricultural patterns

The Brief

Computed Place Intelligence: Beyond the Comparable

Real estate due diligence has relied on comparables for a century. This approach ignores everything that makes a place actually livable: viewshed quality, flood exposure at granular resolution, wildfire probability, soil composition, proximity to environmental hazards. Every one of these factors is now computable from freely available federal datasets. The firms that integrate geospatial intelligence into property assessment will redefine what "due diligence" means.

Brief No. 03 / 11Legacy Systems

The Brief

When to Bridge and When to Rebuild

There is a persistent temptation to rebuild legacy systems from scratch. In most cases, this consumes eighteen months and delivers an inferior version of what already existed.

The alternative — an intelligent bridge layer that translates between deprecated infrastructure and modern AI interfaces — preserves institutional knowledge while enabling new capabilities. If the underlying data is sound, bridge it. If the data model itself is the problem, rebuild the data model and bridge everything else.

Bridge infrastructure substrate
Brief No. 04 / 11Telecommunications Infrastructure
Telecommunications tower against open sky

The Brief

The Catalog and the Carrier: Reading Frontier Telecom Assets

Telecommunications assessments in frontier markets default to two registers — the regulator's inventory and the operator's investor deck. Both describe the network as the country wishes it were. Neither describes what carries traffic on a given Tuesday in a given province.

The relevant intelligence sits in a third register: provincial-level resilience, cross-border peering arrangements, the actual depth of fiber redundancy under the only road that matters, the difference between a tower's nameplate capacity and its usable capacity at 8 PM local time.

The work of a frontier telecom assessment is, in this sense, less about cataloguing what exists than about quantifying the gap between what is reported and what is carried. That gap is the country's true digital baseline.

Brief No. 05 / 11Emerging Market Operations

The Brief

Offline-First Architecture Is Not a Compromise

Building for environments with intermittent connectivity is not about graceful degradation — it is about treating local-first operation as the primary mode and synchronization as a background luxury. The architecture patterns required — conflict-free replicated data types, local-first databases, progressive sync queues — produce better software everywhere, not just in markets where bandwidth is scarce.

Brief No. 06 / 11Philanthropic & Development Capital

The Brief

Reading the Funder Landscape: Where Capital Actually Waits

The list is the smallest part of it. The pathways are the landscape.

Most searches for funding begin and end in the same register: the open call, the published tender, the grants database filtered by country and theme. These describe the funding landscape as it advertises itself — the capital that has announced it is looking.

The capital that matters most rarely announces itself that way. For an organization with a track record — awards, fellowships, a funder or two who already believe in it — the decisive money moves through a different register entirely: the warm introduction, the prize nomination sourced through a network, the foundation that accepts no unsolicited proposals but acts as a connector to the ones that do, the design window of a fund whose architecture is still being written. None of it appears in a database. All of it is mappable.

The work of reading a funder landscape is, in this sense, less about cataloguing the calls that happen to be open than about quantifying the distance between what is listed and what actually moves — and then preparing the brief, the application, and the response to the standard each particular door requires. The list is the smallest part of it. The pathways are the landscape.

Brief No. 07 / 11Healthcare Data

The Brief

Interoperability Is Not a Technology Problem

Siloed data creates billing leverage. Until that economic fact changes, the standards will continue to lose.

Healthcare systems have spent two decades and billions of dollars attempting to solve data interoperability through technology standards. The problem persists because it was never primarily technological. It is an incentive problem: siloed data creates billing leverage.

Applied intelligence in healthcare means building systems that align data sharing with financial incentive — structured intelligence that makes a patient's complete picture valuable to every provider in the chain.

Brief No. 08 / 11Fraud & Anomaly Detection
Wave interference patterns

The Brief

Pattern Recognition at the Margin

Rule-based fraud detection catches the fraud you already understand. The consequential fraud operates in the space between rules. Applied intelligence systems that learn the shape of normal transactions and flag structural anomalies consistently outperform rule engines by an order of magnitude.

The challenge is not detection accuracy. It is institutional willingness to act on probabilistic signals rather than deterministic flags — a governance question masquerading as a technical one.

Brief No. 09 / 11Supply Chain

The Brief

Provenance Tracking: From Compliance Burden to Competitive Moat

Supply chain provenance requirements are typically treated as compliance costs. The firms that reframe provenance as a data asset discover something unexpected: the infrastructure required to prove where something came from is the same infrastructure that optimizes how it gets there.

Compliance becomes intelligence. The audit trail becomes the supply chain brain.

Brief No. 10 / 11Climate & Carbon
Forest canopy from above

The Brief

Verification Is the Bottleneck in Carbon Markets

The voluntary carbon credit market's fundamental constraint is not supply or demand but verification — confirming that a claimed offset actually occurred, persisted, and was not double-counted. Satellite imagery, sensor networks, and on-chain registries each solve one piece. The opportunity is in the synthesis layer: systems that ingest heterogeneous verification data and produce auditable confidence scores for every credit in circulation.

Brief No. 11 / 11Defense & Intelligence

The Closing Brief

Data Architecture as Strategic Advantage

The organizations that will define the next decade of defense capability are not the ones with the most data. They are the ones whose architecture allows a question asked at 9 AM to be answered — with provenance, confidence intervals, and source attribution — by 9:15.

The bottleneck in modern intelligence work is not collection. It is synthesis: pulling signal across schemas, formats, classification levels, and source reliabilities fast enough to inform a decision while it is still a decision rather than a post-mortem. Synthesis is an architecture problem before it is an AI problem, and pretending otherwise produces brittle systems that hallucinate confidently in operational contexts.

We are not sector specialists. We are methodology specialists who have worked across many sectors.

Inquiries are read by a partner and answered within two business days. There is no formal pitch process. We prefer a paragraph to a deck.

Begin an Inquiry Careers
Approach A Practitioner's Note MMXXVI Ashcroft & Thorne

Approach

How we think about complex problems, what we build around them, and the manner in which we begin.

§   Foreword

Ashcroft & Thorne is an applied intelligence consultancy. We work with institutions on questions where the terrain is unfamiliar or in flux — regulatory environments, emerging markets, sectors in transition, problems that cannot be answered by analogy to what has worked before. Our role is to map that topography thoroughly, then help clients act within it.

What has changed, in recent years, is the instrument. Applied intelligence — the disciplined deployment of modern AI systems for substantive research and analytical work — now allows careful work to be done at a pace and a breadth that would not have been possible a short while ago.

The work is, and has always been, preparation. The tools that support it have changed. The standard has not.

§ Methodology

We decompose complex problems into auditable components. Every engagement begins with the same discipline: map the informational topography — the regulatory structures, market dynamics, stakeholder networks, and technical constraints that define the landscape — then identify the constituent questions and determine what evidence would answer each one. This process is recursive: an initial broad scan surfaces the contours, those findings are measured against the problem as stated, and the same instruments are retrained at higher resolution on the areas that matter most. Each pass sharpens the focus.

Our methodology runs parallel investigative streams across regulatory, market, technical, and stakeholder dimensions simultaneously. Where a conventional research process moves sequentially — one analyst finishing a section before the next begins — ours operates in parallel, cross-referencing findings between streams as they develop. The result is not simply faster. It is structurally different: contradictions surface early, gaps reveal themselves against adjacent evidence, and the final synthesis reflects the full landscape rather than the sequential accumulation of isolated findings.

What this architecture produces is not a finished report. It is the most thoroughly researched starting point your team has ever worked from — every claim sourced, every citation linked, every data point traceable to its origin. The analyst who receives this work spends their time on judgment, interpretation, and the institutional knowledge that no system can replicate, rather than on the weeks of retrieval and assembly that ordinarily precede it. We do not replace the people who do this work. We return them to the part of it that actually requires them.

Ninety percent of what passes for knowledge work is retrieval, cross-referencing, and verification — tasks that demand rigor but not insight. We automate that ninety percent with deterministic pipelines that leave a complete audit trail. The remaining ten percent — the synthesis, the strategic read, the recommendation that accounts for what the data cannot say — is where human judgment operates, constrained and supported but never supplanted.

The architecture determines the outcome.

Fig. I  ·  Reconnaissance to Delivery  ·  The sequence of the work
A horizontal systems-flow diagram with five stages from left to right: Brief, Reconnaissance, Synthesis, Decomposition, and Delivery. A dashed recursion arc loops from Synthesis back to Reconnaissance above the main flow. Small iteration glyphs mark each decomposition sub-component. A dotted reframing arc loops from Delivery back to Brief below the main flow. the work, front to back Brief Terms of reference Reconnaissance parallel inquiries · terrain Synthesis signal & gaps recursion — narrower, again Decomposition auditable components Delivery traceable · actionable ↻ ↻ ↻ ↻ iteration — each component refined in place reframing — the brief revised by what was found

Ninety percent deterministic pipeline. Ten percent constrained synthesis. The proportion is not arbitrary.

§ How We Begin

Most advisory relationships begin with credentials. Ours begin with work.

By the time we meet, we have usually done something already — a draft, a prototype, a careful reading of the sector. Not a pitch; actual work, done to the standard the material requires. This is preparation. The conversation begins further along than it would have otherwise, and you take it from there.

§
Est. MMXXVI · Ashcroft & Thorne
Careers
Ashcroft & Thorne Est. MMXXVI San Francisco

The Practice

Ashcroft & Thorne is an applied intelligence consultancy. We provide institutional-quality research deliverables — tender proposals, market studies, feasibility assessments, regulatory landscapes, implementation strategies — through an AI-native research methodology that runs parallel investigative streams across regulatory, market, technical, and stakeholder dimensions. The result: deliverables that arrive in days, with deeper sourcing, broader coverage, and full analytical auditability.

Engagements are commissioned across the Americas, Africa, and Europe. Work is accepted selectively and executed directly. Deliverables are produced for the partner named on the engagement letter and are circulated nowhere else.

The practice publishes no client list. The intelligence briefs, read alongside this page, are the closest public account of what we think and how we work.

Inquiries are read by a partner and answered within two business days. There is no formal pitch process. We prefer a paragraph to a deck.

Careers

Careers

Open positions are listed below.

ML Optimization Engineer San Francisco · Remote

Quantization, LoRA fine-tuning, and inference optimization for transformer models deployed on resource-constrained hardware. Demonstrated experience shipping optimized models to production — not benchmarking, shipping. Comfort working at the boundary between model performance and physical device limitations. 3B-parameter models on 4GB of RAM is the operating range.

Security & Privacy Architect San Francisco · Remote

Architecture-level design for systems where private data remains under user control throughout the computation lifecycle. Encrypted containers, local inference, secure enclaves, key management. Strong technical opinions about the tradeoffs between trusted execution environments, homomorphic approaches, and simpler solutions that actually ship. The output of this role is working infrastructure, not documentation.

Visual Designer San Francisco · Remote

Editorial sensibility. Typographic command. Range across print and digital. The portfolio should demonstrate restraint and a clear point of view. If your instinct is to add, this is not the right role.

AI Training & Knowledge Engineer Remote

Construction of structured knowledge bases and domain-specific training datasets across a wide range of subjects — southeastern fly fishing, Middle Eastern telecommunications infrastructure, field geology. Part research, part taxonomy, part applied ML. Must be able to design evaluation frameworks that determine whether a fine-tuned model actually knows what it claims to know.

Applied Intelligence Analyst Remote

Sector research, cross-source synthesis, and written analysis for decision-makers with limited time and high standards. Publication-grade prose. Defensible reasoning. A writing sample on a topic of your choosing will be requested.

Engagement structure discussed individually. We value alignment over availability.

To inquire, send a note to careers@ashcroftandthorne.com with relevant work and a brief description of what drew you to this practice. No formal application. No résumé required.

careers@ashcroftandthorne.com
Of all the trees that grow so fair,
Old England to adorn,
Greater are none beneath the Sun,
Than Oak, and Ash, and Thorn.

— Rudyard Kipling, "A Tree Song"