Meotis Finance · Neuchâtel

The investment lab installed under your roof.

Installed on infrastructure you control, Noésis holds the research and the regime reading, Atlas the portfolio oversight, custody controls, compliance, approvals and client reporting, in one governed workspace, built for the constraints of FINMA-regulated firms: every decision traced, roles kept apart, and your data, documents and logs held inside the installation.

What the division represents

Meotis’s financial laboratory.

Within Meotis, Finance is where cognitive heritage is applied to capital management. It captures specialist methods, turns them into explicit processes, then encodes them in tools that humans can supervise.

Vision

Augment judgment without ever moving responsibility.

Artificial intelligence should make every decision better informed, more adversarial, and easier to explain, without deciding in place of the professional.

Mission

Build the Swiss tools for governed finance.

We combine domain expertise, agentic research, and sovereign engineering to ship products that are useful, measurable, and auditable.

01

Observe

Start with the real work of managers, analysts, and researchers: their data, trade-offs, and constraints.

02

Formalize

Turn that knowledge into explicit rules, controls, veto rights, and records that can be reviewed.

03

Deliver

Build those methods into Atlas, test them in practice, and publish what we learn.

Sovereignty

Installed on your side, and it stays there.

Atlas is installed on the organisation's local infrastructure. The database, documents and operational history remain within the installation. Cloud connectors and external AI are disabled by default. Access to official Swiss public-data services is optional, explicit and logged.

Without external AI

The core of the product stands on its own.

Atlas's core portfolio, compliance, document, approval and audit workflows remain available without external AI. When a local inference engine is configured, AI features run inside the installation, under the same human-approval and audit controls.

What Atlas does not do

It prepares analysis, it does not execute.

Whether inference is local or externally authorised, Atlas only prepares analysis. It cannot execute at a custodian or send client-facing material without human approval.

Product · Atlas

The operating cockpit for wealth managers.

Atlas is a locally installed PMS built for independent wealth managers and family offices. It prepares the analysis, organises the debate, and documents the file. The manager keeps the decision.

Released July 3, 2026

Atlas

01

One cockpit

Portfolios, clients, alerts, compliance, and decisions awaiting approval live in one workspace.

02

An adversarial committee

30 specialist AI agents make the case, 15 bulls against 15 bears. The Risk and Compliance seats hold veto rights.

03

A human decision

Every proposal lands in an approval queue. Nothing leaves without being read, edited, or signed.

04

A verifiable memory

Assumptions, checks, and decisions are timestamped so the reason behind every action can be reconstructed.

See Atlas in detail
The Atlas cockpit showing portfolios, alerts, and proposals awaiting approval

Atlas · demonstration environment. Modules vary with the plan.

Product · Noésis

The control room behind your decisions.

Ahead of Atlas, Meotis runs its own investment lab: close to 200 indicators across 13 categories, reviewed every night, which set the house view on the cycle, on where risk is building and on which asset classes deserve more weight. Every scenario carries its own probability.

Noésis Macro Asset Allocation module: the four strategies with their model weightings and layer-by-layer attribution
Built to measureFor FINMA-regulated firms

Noésis

Regime reading, signals, and the team's research, brought together on one screen.

The shared memory

Each analyst's research, their notes, sources, and conference takeaways, is centralized in a shared memory. This corpus feeds the firm's cognitive core, which re-reads it every night: testing theses against market data, re-estimating the probability of each scenario, and logging every revision, strictly within the firm's investment criteria.

The system proposes, the manager decides

When recent research or an event warrants it, the system proposes new indicators on its own, built exclusively on verifiable data with named sources. The manager decides: nothing enters the cockpit without approval. Validated signals then tilt the model portfolios' allocations, and every weighting stays decomposable.

The first agentic financial product

This research feeds a first agentic financial product running under real market conditions, on a pilot portfolio: an agent investment committee that builds the case for and against every file, coded risk guardrails, and one rule without exception: every proposal passes a human approval gate, every action is logged in a chained, verifiable audit trail.

Our division's conviction fits in one sentence.

“Agentic AI must be grounded in a soundly designed process, never assuming the agent itself can be the source of domain expertise.”Meotis R&D Finance

AI agents are already beating the 60/40 portfolio on a risk-adjusted basis, according to J.P. Morgan Research.

A regulated asset management structure is in preparation to offer these services.

Advisory mandates

Accounting and tax expertise, at your company's service.

The Finance division doesn't just build software: it works under mandate for companies and their leadership. Cost control, budget planning, tax, valuation, project delivery: the same practitioners who define our tools hold the mandates.

AccountingBookkeeping and account optimization

Maintaining, reviewing, and structuring accounts: expense processing, cost capitalization, financial statement presentation.

TaxVAT and direct taxes

VAT filings, tax returns, and correspondence with the authorities: compliance first, optimization second, always within the Swiss legal framework.

SteeringBudget, cash flow, and cost tracking

Budgets, cash-flow plans, and variance analysis: knowing where the money goes, planning your flows, and anticipating what’s left in six months.

ValuationBusiness valuation

Valuing companies and equity stakes: recognized methods, explicit assumptions, ranges defensible in front of a third party.

Securities & dividendsSecurities transactions and distributions

Tax structuring of securities transactions and dividend distributions, arbitrated case by case.

SuccessionSuccession and transmission

Preparing the transfer of a business or estate with a controlled tax burden, planned across multiple fiscal years.

Seven years of audit and advisory work behind every mandate, with SMEs, groups, and executives. Advisory mandates are neither investment advice nor a licensed activity.

Papers & publications

Research documents what the product claims.

We publish the methods, limits, and results that shape the division’s products. No promise without a protocol or a record.

Analysis · published 27 August 2026

The AI build-out is now financed by credit

Hyperscaler capex has moved off free cash flow and onto debt, chip-backed lending and insurance-funded private credit. That changes whose balance sheet takes the loss.

Read the publication
Analysis · published August 6, 2026

The tungsten signal: the war premium left oil for a metal nobody tracks

Oil is falling on peace talks while a strategic metal almost nobody follows has risen tenfold since 2022. Why an oil premium fades and why this one sticks.

Read the publication
Data paper · Scientific Data (Nature) · 2026

A Dataset of Early Blockchain-Registered AI Agents on Ethereum

The first 10,000 AI agents registered on-chain under the ERC-8004 standard: identity, declared services, mint economics and reputation, as open, reproducible data. By Dr Yulin Liu, the division’s research partner. Published in a peer-reviewed journal; the open version remains on arXiv.

Read in Scientific Data
Analysis · published July 27, 2026

The model layer is commoditizing: sell the frontier, trade the token tax

Open-weight and Chinese models reach the frontier in months, not years. What that compresses, and the coming policy fight over a token tax.

Read the publication
Analysis · published July 25, 2026

Compute as an asset class: the physical-bottleneck trade comes to Switzerland

AI’s constraint has shifted to electricity, memory, and chip fabs; a site under study in Switzerland embodies the thesis.

Read the publication
Product & research · published July 3, 2026

Atlas: the software that steers your clients’ portfolios

A guided tour of the cockpit, the 30-agent committee, and the human approval queue.

Read the publication
NewsroomSee all Meotis publications
Meotis Finance

You manage capital. Let’s discuss your process.

We work with organizations that want to expand their analytical capacity without giving up control of their decisions or their data.

hello@meotis.ch

Informational and educational content only: it does not constitute personalized investment advice, an offer, or a solicitation.