Nexos is the AI and software division of Abrams & Luchanski, a San Diego firm founded in 2024 with no New York office. It builds AI agents, private LLMs, custom applications and system integrations for New York businesses, whether a Manhattan finance or professional firm, a hospital or nonprofit back office, or an upstate manufacturer or dairy operation, and every New York engagement is remote. Workshops, reviews and training take place over video during New York business hours, and each design is planned with the New York rules that may apply in view: the city’s bias-audit law for hiring tools, the SHIELD Act’s security duties and, for banks and insurers, NYDFS Part 500, with your counsel confirming what applies to you.
By the Bureau of Labor Statistics’ August 2026 preliminary count, New York has just under ten million nonfarm jobs, and the biggest single block of them — 2.43 million — is education and health services. Financial activities account for 759,100 more, and professional and business services for another 1.40 million. The Census Bureau puts health care and social assistance receipts at $252.4 billion for 2022. That is an economy built on documents, approvals and regulated records: claims, filings, referrals, audits and client files, most of which pass through several systems before anyone acts on them.
It is also an economy of small operators. The Census Bureau counted 541,313 employer establishments across the state in 2023 and nearly two million nonemployer businesses — close to four owner-only outfits for every location with a payroll. A business with no employees has no IT department either, so every hour its owner spends reconciling systems by hand comes straight out of billable time.
Upstate reads differently. Empire State Development says more than 156 semiconductor and supply-chain companies employ over 34,000 New Yorkers, names Micron in Central New York, GlobalFoundries in the Capital Region and Wolfspeed in the Mohawk Valley, and promotes the Finger Lakes as an optics and photonics hub. Agriculture matters as well: the USDA’s 2025 overview lists 30,000 farm operations, with milk the top commodity at $3.61 billion. Fabs, their suppliers, machine shops and dairy cooperatives all live on lot records, certifications and settlement statements.
Nothing below describes past work. These are the projects that New York’s mix of regulated services and upstate production makes most worthwhile, and the reasons why.
Insurers, lenders, money transmitters and agencies licensed by the Department of Financial Services have to be able to show an asset inventory, multi-factor coverage and a working incident process whenever the regulator asks. Pulling that evidence automatically from the identity provider, the device manager and the ticketing system turns an annual scramble into a standing report the compliance officer can sign.
Clinics, home-care agencies and nonprofits spend staff hours reading faxes, referral packets and eligibility forms and retyping them into an EHR or case system. An agent that extracts the fields, checks them against the record and queues anything uncertain for a coordinator keeps a person on every decision while removing the retyping.
Firms whose value sits in years of engagement letters, workpapers and memos cannot put those files into a public chatbot. A private language model lets staff search and draft from that archive without handing it to a public service: installed on your own hardware, which we scope as a build, or hosted for you from $890/month.
Component makers serving the chip fabs and the photonics cluster are asked for lot histories, material certificates and on-time figures by customers with exacting quality systems. With the ERP, the quality database and shop-floor records joined in one data platform, a lot-history or certificate request becomes a query rather than a manual hunt through three systems.
Milk moves from farm to plant on hauler tickets, component tests and pay-price calculations that often live in spreadsheets. A small internal application that ties those together gives the farm, the hauler and the cooperative the same numbers each pay period.
For a business putting AI or automation to work in New York, six sets of rules come up most: a New York City law on hiring tools, the statewide SHIELD Act on data security and breach notice, the state’s cybersecurity regulation for financial companies, a disclosure law for prices set by algorithm, DFS guidance on AI in insurance underwriting, and — for model developers only — the RAISE Act. A separate bill to protect health data held outside HIPAA passed both houses in January 2025 but was vetoed on December 19, 2025, so it is not law.
Employers and employment agencies may not use an automated employment decision tool on job candidates or employees in New York City unless the tool has had a bias audit within one year of its use, information about that audit is publicly available, and the required notices have gone to the people assessed. The Department of Consumer and Worker Protection enforces it under its published rules.
Any person or business that owns or licenses computerized private information of a New York resident must keep reasonable administrative, technical and physical safeguards, such as a named security coordinator, risk assessments, staff training and vetted service providers. Small businesses under the statute’s employee, revenue or asset tests may scale those safeguards to their size.
“Private information” includes biometric data, medical and health-insurance information, and a username or email address combined with a password or security answer, so intake records and stored credentials are both in scope. The current text requires notice to affected residents within thirty days after a breach is discovered, with copies to the Attorney General, the Department of State, the State Police and, for DFS-covered entities, the Department of Financial Services.
Banks, insurers and other companies licensed under the Banking, Insurance or Financial Services Law must run a cybersecurity program, report qualifying incidents to DFS within 72 hours, use multi-factor authentication and maintain an asset inventory. Firms under 20 staff, $7.5 million in revenue or $15 million in assets get a limited exemption from parts of it, but must still use multi-factor authentication for remote access to their systems, for remote access to third-party applications holding nonpublic information, and for privileged accounts. Since November 1, 2025, every other covered entity has had to require it for anyone accessing its information systems.
A business that sets a price with an algorithm using a consumer’s personal data must show the words “THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA” next to it. Insurers, financial institutions covered by Gramm-Leach-Bliley and certain existing-subscriber discounts are excepted, and a court may impose up to $1,000 per violation.
Addressed to every insurer authorized in New York, it says AI systems and external consumer data should not be used for underwriting or pricing where that would permit unfair discrimination. Insurers are expected to govern that use at board and management level, test for discrimination before an AI system goes into production and regularly afterwards, and tell applicants the reasons behind an adverse decision.
Binds only developers of frontier models, meaning foundation models trained with more than 10^26 operations; a company using a commercial model inside its own tools is not one.
General information, not legal advice. Laws, regulations and their effective dates change; confirm how they apply to your business with your own counsel. Last reviewed 2026-09-26.
A New York engagement starts with the free audit: a video call in your afternoon, then a written audit of which workflow to automate first, noting where Part 500 or Local Law 144 may come into play for your counsel to confirm. Next comes a fixed-scope proposal: the build, the exclusions and the price, in writing. Once the build begins, the staff who handle the work share their screens and walk us along the real path of a file, a claim or an order, including the spreadsheets nobody mentions in meetings. Through the build we hold a video check-in every week, keep progress on a board you share, and put early working versions in your hands, so a regulated firm’s compliance lead or outside counsel can look at it before anything connects to production.
All of New York keeps Eastern Time, three hours ahead of our Pacific clock. That puts a 9 a.m. start for us at noon for you, so reviews are set for your afternoon, between noon and 5 p.m. Eastern. Training is recorded so teams on shifts (a hospital unit, a plant crew) can watch later instead of joining live.
Remote: discovery, reviews and training over video, with shared documents and working software in between. Eastern Time statewide — three hours ahead of Pacific.
The free technology audit is a call and a written audit, wherever you are, Southern California included. Outside Southern California, all of it runs over video.
Audit — free: a call and a written technology audit.
Build — from $4,900 per fixed-scope project.
Run — from $890/month for a hosted private model.
The same prices everywhere. See pricing.
Aidan Luchans founded Abrams & Luchanski in San Diego in 2024, and designed and built the Nexos Platform, a multi-tenant AI operations platform: AI agents, multi-step workflows with human approval, connections to roughly 90 business tools, and private knowledge search.
Aidan leads both the firm’s advisory work and Nexos’s engineering, so the person who scopes a project also builds it.
A firm in Watertown or Jamestown works with us on exactly the same terms as one in Midtown, because a New York engagement would run over video either way.
Our prices do not change with your ZIP code. The audit is free, and builds start at $4,900 for a fixed-scope project, and a hosted private model, which keeps client files off public chatbots, runs from $890/month; installing one on your own hardware is scoped as a build. A single-workflow agent, such as an intake or evidence-collection agent, sits at the low end. Each extra system we connect and each regulated review step adds scope, so a New York agent climbs above the floor when your counsel’s Part 500 review sets requirements the build must meet, or when it has to fit around an independent Local Law 144 bias audit, which a third-party auditor performs.
Yes, but if the tool counts as an automated employment decision tool under Local Law 144, it needs a bias audit within the year before use, a public summary of that audit and notice to the people it assesses. The law covers tools that “substantially assist or replace” discretionary hiring or promotion decisions, and DCWP’s rules read that phrase narrowly, as three cases: the tool’s output is the only factor, it is weighted more than any other criterion, or it is used to overrule a human decision. A screening tool outside those cases can be used without the audit and notice duties. We design screening assistants so every step is logged, and we flag any design near that line so your counsel can decide whether a bias audit is needed.
Yes. Part 500 does not forbid AI; it expects the tools you add to sit inside your cybersecurity program, with multi-factor access, an entry in your asset inventory, logging and vendor due diligence. DFS’s industry letter of October 16, 2024 says the same: it adds no new requirements, but asks covered entities to address AI-driven threats such as deepfake social engineering in their risk assessments, training and third-party reviews. A private model that keeps nonpublic information in your own environment is one way to meet that, and we document how the system works so your team and counsel can reflect it in your own risk assessment.
No — work in New York is remote. In-person sessions are limited to Southern California, and a New York engagement runs its discovery, weekly check-in calls, design reviews and training over video, scheduled for your afternoon.
Start with the task where staff copy information from one document or system into another — intake forms, invoices, compliance evidence or client onboarding. It is easy to measure, low-risk to put a reviewer on, and the free audit is how we find which one costs you the most hours.
A short description is plenty. We reply within one business day.
Nexos works alongside Abrams & Luchanski consultants so technical solutions are designed around the business — not the other way around.