Field notes from the AI stack.
Deep-dive write-ups of representative engagements across banking, healthcare, telecom, and consumer — the hypothesis, the architecture, the rollout discipline, and the lessons we'd hand you before you build.
Consumer lending, decided in hours
How an underwriting agent assembles the credit file, applies policy, and drafts decisions — cutting time-to-decision by 60% while keeping a human on every approval.
Early-warning credit risk
A predictive ML program that blends bureau and behavioral signals to flag at-risk accounts before they roll — and routes the right treatment to each one.
The 6 a.m. advisor brief
Portfolio drift, suitability flags, market moves, and meeting prep — assembled by an agent overnight so advisors start the day advising, not gathering.
Tax-aware rebalancing
Drift detection across thousands of household portfolios with tax-lot-aware trade proposals an advisor reviews and approves — turning a quarterly scramble into a continuous process.
Personal quarterly commentary for forty thousand households
Grounded per-household reporting that explains what happened to this client's portfolio and why — drafted in minutes, checked against a compliance rulebook, signed by a human.
Private banking onboarding
Document AI reads the pack, verifies what it can, and assembles the file — while every screening judgment and every acceptance decision stays with a human.
Seeing assets leave before they leave
Outflows are almost never a surprise in hindsight. A model that reads the early signals — and hands advisors a specific, timely reason to make the call.
Reviewing every recommendation instead of two percent of them
Advice surveillance that reads the recommendation, the client profile, and the rationale together — flagging what a compliance officer should look at, and never issuing a finding on its own.
Prior authorization in minutes
Extraction pipelines that read patient records against payer requirements, assemble submission-ready packets, and flag gaps for human sign-off.
Trade surveillance
Legacy lexicon rules generated thousands of alerts a day and almost no findings. An agentic triage layer that correlates trading and communications, drafts a rationale, and leaves every disposition to a human.
An analyst copilot that reads everything and decides nothing
Filings, transcripts, and broker research made searchable with citations — so analysts spend their hours forming a view rather than locating the paragraph.
Post-trade breaks
Automated matching across custodian, broker, and internal records — with an agent that assembles the evidence and drafts the counterparty query for every genuine break.
The save desk that starts with context
Predictive models flag at-risk accounts; an agent assembles the full account story and drafts in-policy retention offers — reps approve, customers stay.
Support that cites its sources
A retrieval-grounded assistant that answers from the brand's own help center, policies, and live order data — deflecting routine volume and handing nuance to humans with context attached.
Forecasting demand where it's decided
From spreadsheet consensus to a governed ML forecast that planners can interrogate — fewer stockouts, less dead inventory, and a planning meeting about exceptions instead of averages.
Finding the money left in closed claims
A model plus a document agent surface missed third-party recovery in closed auto claims and draft the demand package — turning quiet leakage into recovered dollars.
The NOC's new first responder
A triage agent correlates alarms, enriches them with topology and history, and drafts resolution steps — so NOC engineers open every ticket already understanding it.
Giving clinicians their evenings back
A grounded documentation assistant drafts submission-ready notes from the encounter record, with a clinician reviewing and signing every one.
Shipping faster
How a B2B software company gave 300 engineers a copilot that actually knows their codebase, their runbooks, and their incident history — cutting ramp time and cross-team escalations.
Fifty thousand alerts a day
An agentic triage layer that enriches, correlates, and drafts a verdict for every alert — so analysts spend the night on the dozen that matter, not the twelve thousand that don't.
Invoice to cash
Document AI reads every invoice, matches it to the purchase order and receipt, and posts it — while exceptions route to humans with the evidence already assembled.
Catching defects at line speed with computer vision
A vision system that flags defects in under 200 milliseconds on the line — and, just as importantly, learns which flags the inspectors actually agree with.
Predicting transformer failures before the lights go out
Sensor telemetry, maintenance history, and weather combined into a risk score per asset — turning emergency truck rolls into planned work with months of warning.
Benefits casework
An agency cut its backlog by automating document intake and evidence assembly — while keeping every eligibility decision with a caseworker and auditing the system for disparate impact.
Four hundred thousand contracts, finally answerable
Clause extraction and obligation tracking across a decade of agreements — so legal can answer 'what are we actually exposed to?' in a day instead of a quarter.
Matching engineers to roles without letting the model decide
A talent platform that ranks and explains fit across a large candidate pool — with the model surfacing evidence, recruiters making the call, and every release tested for adverse impact.
Four thousand deliveries a day, planned in eleven minutes
Demand forecasting feeding a routing optimizer — plus a planner interface that lets dispatchers override any route, and a system that learns from every override.
Stories describe representative engagement patterns; metrics are design targets, and details are generalized to protect client confidentiality.
And what we make of where this is going.
Longer-form analysis on how AI is reshaping markets, economics, and work — written by the people doing the builds.
How AI is redrawing market structure
Model capability is commoditizing faster than almost anyone predicted. That doesn't mean the value disappears — it means it moves. A map of where it is going, and what that implies if you are buying rather than building.
Read the perspective →Agentic AI grows up: from impressive demo to system that holds
Most agent pilots stall in the gap between a demo that works once and a system that works ten thousand times. The difference is almost never the model.
Read the perspective →The quiet shift to small, specialized models
The largest model is rarely the right model. Why serious work is moving down the size curve — and what enterprises gain besides a smaller bill.
Read the perspective →Bring us a hypothesis. Leave with a system.
Tell us what's eating your team's time. We'll give you an honest read on whether AI is the right tool — and if it is, a scoped v1 with a timeline and cost.