Capabilities Sectors Platform How we work Portfolio Case studies Responsible AI Company FAQ Book a call
2026 Now onboarding Q4 delivery slots

We build AI that survives contact with real operations.

WasabiLabs is an applied AI studio. We design, ship and operate production systems — autonomous agents, forecasting engines, computer vision and document intelligence — for companies whose margins depend on getting the details right.

90 daysIdea → production median
11 sectorsShipped and supported
99.95%Managed uptime SLO
wasabilabs · deployment.yaml
# production agent — claims triage, EU region
pipeline: claims-triage-v7
runtime:
  orchestrator: wasabi-runtime
  region:       eu-west-1
  residency:    eu-only
models:
  - role: extraction   latency_p95: 410ms
  - role: reasoning    latency_p95: 1.2s
  - role: verification latency_p95: 380ms
guardrails:
  groundedness:  0.96   # min. score
  pii_redaction: strict
  human_review:  below_confidence(0.88)
observability:
  traces: full   evals: nightly + shadow
deployed · healthy 142k decisions / month

Trusted by operators in regulated and high-throughput industries

NORDBANK Meridian Health CARGOLINK Vessa Retail Kōbe Industrial ATLAS ENERGY Lumière Group Civitas Public Orbit Telecom PRIMA AGRO NORDBANK Meridian Health CARGOLINK Vessa Retail Kōbe Industrial ATLAS ENERGY Lumière Group Civitas Public Orbit Telecom PRIMA AGRO
Capabilities

Six practices. One delivery team.

Most AI projects die between the demo and the first real user. We organise around the disciplines that close that gap — modelling, engineering, evaluation and operations — so the thing you approve in a workshop is the thing that runs on Monday.

Agentic systems

Multi-step agents that read, decide and act inside your stack — with tool access, memory, escalation paths and a human in the loop where it matters.

  • Tool & API orchestration
  • Deterministic guardrails and fallbacks
  • Supervisor / worker topologies

Forecasting & optimisation

Demand, price, risk, capacity and maintenance models built on your own history — paired with solvers that turn a prediction into a decision someone can act on.

  • Hierarchical & probabilistic forecasting
  • Constraint and route optimisation
  • Scenario simulation and what-ifs

Computer vision

Inspection, counting, safety and quality models that run at the edge, on the line, in the yard or on a drone — offline-tolerant and calibrated to your defect taxonomy.

  • Defect detection & anomaly scoring
  • Edge inference on constrained hardware
  • Active learning from operator feedback

Document intelligence

Contracts, claims, invoices, dossiers and scans turned into structured, auditable data — with citations back to the source page and confidence on every field.

  • Layout-aware extraction at scale
  • Field-level confidence & provenance
  • Straight-through processing rules

Data & retrieval platform

The unglamorous half of AI: pipelines, lineage, chunking strategy, embeddings, hybrid search and permissions that mirror the ones you already enforce.

  • Hybrid semantic + keyword retrieval
  • Row-level access inheritance
  • Freshness & drift monitoring

AI assurance & MLOps

Evaluation harnesses, red-teaming, drift alarms, cost ceilings and audit trails. If it cannot be measured on a dashboard, we do not call it done.

  • Golden sets & regression gates
  • Shadow deploys and canaries
  • EU AI Act documentation packs
140+Models in production across client estates
€48MAnnualised value measured by client finance teams
11Regulated sectors with shipped deployments
4.8/5Average post-engagement client score
Sectors & use cases

Domain first. Model second.

A fraud model and a defect model share almost no vocabulary. We keep standing teams per sector so the conversation starts at your workflow, your regulator and your failure modes — not at a generic architecture diagram.

Financial services

Models that have to explain themselves. We build under the assumption that every decision will eventually be read out loud to a regulator, an auditor or an unhappy customer.

−62%Manual review queue on retail onboarding
+18ppFraud recall at constant false-positive rate
9 daysAverage model documentation turnaround

Transaction fraud & AML

Graph-aware scoring with narrative explanations attached to each alert, so analysts spend their time deciding rather than reconstructing.

Risk

KYC & onboarding automation

Identity documents, proof of address and corporate structures parsed, cross-checked and routed — with straight-through processing above a tuned confidence floor.

Operations

Credit decisioning support

Alternative-data features and reason codes that slot into existing scorecards without breaking model governance.

Lending

Advisor copilots

Portfolio context, product rules and suitability constraints in one assistant that drafts client-ready notes with citations.

Front office

Regulatory change monitoring

Continuous ingestion of circulars and consultations, mapped to the internal controls each one touches.

Compliance

Complaint triage

Intent, severity and redress classification across channels, with automatic detection of systemic issues before they become a thematic review.

Conduct

Healthcare & life sciences

Clinical-grade caution with startup delivery speed. Everything is built assuming PHI never leaves the perimeter and every output is traceable to a source.

−41%Clinician documentation time per shift
3.2×Faster cohort discovery for trial feasibility
0PHI egress events across managed deployments

Clinical documentation

Ambient note drafting that respects local templates, coding rules and the clinician's own editing habits.

Care delivery

Prior authorisation

Policy criteria matched against the chart, with the supporting evidence assembled into a submittable packet.

Revenue cycle

Medical imaging triage

Worklist prioritisation and second-read flags for radiology and pathology, deployed as decision support with clear uncertainty bands.

Diagnostics

Pharmacovigilance

Adverse event detection from case narratives, literature and unstructured intake, coded to MedDRA with human sign-off.

Safety

Trial feasibility & recruitment

Protocol criteria turned into executable cohort queries across EHR and registry data, with site-level yield estimates.

R&D

Patient navigation

Multilingual assistants for scheduling, preparation and follow-up that escalate cleanly to a human at defined clinical boundaries.

Experience

Retail & e-commerce

Margin lives in a thousand small decisions per day. We automate the ones that are mechanical and sharpen the ones that are judgement calls.

+11%Gross margin on dynamically priced lines
−34%Stockouts on high-velocity SKUs
72%Tier-1 support contacts resolved without a human

Demand forecasting

SKU × store × week forecasts with promotion, weather and cannibalisation effects, reconciled across the planning hierarchy.

Planning

Dynamic pricing & markdown

Elasticity-aware price recommendations bounded by brand, competitive and legal guardrails.

Commercial

Catalogue enrichment

Attribute extraction, deduplication and copy generation across millions of listings in every market language you sell in.

Content

Conversational commerce

Assistants that know stock, delivery promises and return policy — and hand over with full context when they should.

Service

Returns & fraud abuse

Behavioural scoring that separates genuine dissatisfaction from organised abuse, without punishing good customers.

Risk

Store operations vision

Shelf availability, planogram compliance and queue analytics from existing camera infrastructure.

Stores

Manufacturing & industry

Shop-floor reality beats laboratory accuracy. We deploy to the hardware you already have, tolerate bad lighting and patchy networks, and measure success in scrap and downtime.

−27%Unplanned downtime on instrumented lines
−19%Scrap rate after vision rollout
6 wksFrom pilot cell to multi-line rollout

Visual quality inspection

Defect detection tuned to your taxonomy, retrained continuously from operator corrections on the line.

Quality

Predictive maintenance

Vibration, thermal and telemetry fusion that gives maintenance a ranked worklist, not another alarm to ignore.

Reliability

Process parameter optimisation

Closed-loop recommendations for setpoints that hold yield while cutting energy and cycle time.

Process

Maintenance knowledge agent

Decades of manuals, work orders and tribal fixes made searchable in the language technicians actually use.

Knowledge

Supply risk monitoring

Early signals on supplier distress, lead-time drift and single-source exposure across the bill of materials.

Supply chain

Workplace safety vision

PPE compliance and exclusion-zone monitoring designed with works councils, privacy-preserving by construction.

EHS

Logistics & mobility

Networks where a five-minute decision delay compounds across a continent. Our systems run at the edge of the network and degrade gracefully when the link drops.

−14%Empty kilometres on optimised fleets
+23ppETA accuracy within the promised window
<200msDispatch decision latency at peak

Route & load optimisation

Fleet, driver-hours and customer-window constraints solved continuously, not once per night.

Planning

ETA prediction

Live traffic, dwell-time and handover models that replace static transit tables with honest probabilities.

Visibility

Freight document automation

Bills of lading, customs forms and proof of delivery extracted and reconciled against the transport order.

Back office

Yard & dock vision

Trailer identification, occupancy and damage capture without new gate hardware.

Terminal

Exception management agent

Delays, damages and misroutes detected, explained and actioned with the carrier before the customer notices.

Operations

Network design simulation

Scenario modelling for hub placement, mode shift and capacity commitments under demand uncertainty.

Strategy

Energy & utilities

Physics-constrained models for assets that cannot be A/B tested. We work alongside your engineers, and we never ship a recommendation that violates the grid code.

−8.4%Day-ahead forecast error on renewables
+2.1%Asset availability on wind portfolios
31%Faster outage cause identification

Generation & load forecasting

Weather-ensemble driven forecasts at asset and portfolio level with calibrated uncertainty for trading desks.

Markets

Grid anomaly detection

Sensor-level deviation detection that distinguishes instrumentation faults from genuine network events.

Networks

Asset inspection vision

Drone and line-crew imagery triaged for corrosion, vegetation encroachment and component wear.

Field

Flexibility & storage dispatch

Battery and demand-response scheduling against price, constraint and degradation objectives.

Optimisation

Customer energy assistants

Tariff, consumption and efficiency guidance grounded in the customer's own meter data.

Retail

Permitting & compliance drafting

Environmental and regulatory submissions assembled from project data with full traceability.

Development

Legal & professional services

Where a hallucinated citation is a career event. Every answer is grounded, quoted and linked, or the system says it does not know.

−55%First-pass contract review hours
100%Answers carrying a verifiable source span
4.6×Throughput on due-diligence data rooms

Contract review & playbooks

Clause extraction scored against your fallback positions, with redlines drafted in your own house style.

Transactional

Due diligence acceleration

Data-room triage that surfaces change-of-control, indemnity and unusual terms before the deadline, not after.

M&A

Matter knowledge search

Precedent and memo retrieval that respects ethical walls and matter-level permissions.

Knowledge

Regulatory horizon scanning

Obligation extraction and impact mapping across jurisdictions your clients actually operate in.

Advisory

Audit evidence assembly

Sampling, tie-out and workpaper drafting with provenance to the underlying ledger entry.

Assurance

Pitch & proposal drafting

Firm credentials, prior matters and pricing precedent composed into a first draft in minutes.

Growth

Public sector & education

Accountability is the product. We build for procurement scrutiny, freedom-of-information requests and the day a journalist asks how the system decided.

−48%Backlog on citizen case processing
27 langsSupported in citizen-facing assistants
FullDecision audit trail retained by default

Citizen service assistants

Multilingual, accessible guidance on entitlements and procedures, grounded strictly in published policy.

Service

Casework triage

Application routing and completeness checks that shorten queues without removing human judgement from outcomes.

Administration

Procurement analysis

Tender comparison, spend concentration and anomaly detection across framework agreements.

Finance

Records & FOI response

Retrieval, redaction and response drafting with statutory deadlines tracked end to end.

Transparency

Learning support

Curriculum-aligned tutoring and formative feedback with strict age-appropriate safeguards.

Education

Infrastructure asset vision

Road, rail and utility condition surveys from vehicle-mounted and satellite imagery.

Works
Platform

The Wasabi Runtime

Every engagement stands on the same foundation: a modular runtime we maintain, deploy inside your cloud or ours, and hand over with the source. No black boxes, no lock-in, no surprise per-seat bill in year two.

Layer 01

Ingestion & data fabric

Connectors, change-data-capture and normalisation across the systems of record you already run — with lineage preserved end to end.

SAPSalesforceS3 / ADLSKafkaFHIRSWIFT
Layer 02

Knowledge & retrieval

Hybrid vector and lexical search, structured joins, and permission inheritance so the assistant can never surface what the user could not open themselves.

pgvectorOpenSearchGraph indexRBAC / ABAC
Layer 03

Model layer

Frontier, open-weight and in-house models routed per task on quality, latency and cost — swappable without touching business logic.

ClaudeOpen weightsFine-tuned SLMsClassical ML
Layer 04

Orchestration & agents

Typed tools, deterministic state machines around probabilistic steps, retries, compensating actions and explicit human checkpoints.

Tool registryWorkflow engineHuman-in-the-loop
Layer 05

Assurance & observability

Offline evals, online shadow scoring, groundedness checks, cost ceilings, drift alarms and a complete trace for every decision made.

Eval harnessOpenTelemetryRed teamingCost guardrails
Layer 06

Deployment & residency

Your VPC, our managed estate, on-premise or the factory floor. Infrastructure as code, reproducible builds, and data that stays where your counsel says it must.

KubernetesTerraformAir-gappedEdge / ARM
How we work

Four phases, one fixed rule: production or it does not count.

We do not run open-ended discovery. Every engagement starts with a decision we are trying to change and a number we expect to move, and ends with that number on a dashboard your CFO can open.

01

Frame

Two weeks. We map the workflow, quantify the decision, and agree what "good" looks like before any model is trained.

  • Value hypothesis & baseline
  • Data and access audit
  • Risk and regulatory scan
02

Prove

Four to six weeks. A working slice on your real data, evaluated against a golden set your experts built with us.

  • Offline evaluation harness
  • Cost and latency envelope
  • Go / no-go with evidence
03

Ship

Six to twelve weeks. Hardening, integration, security review, change management and a staged rollout behind flags.

  • Shadow then canary release
  • Runbooks and on-call handover
  • User training and adoption
04

Operate

Ongoing. We run it, watch it, retrain it and keep proving the value — or we hand the keys to your team and stay on call.

  • SLOs and incident response
  • Quarterly value review
  • Full source and IP transfer
Portfolio

Products born out of client work.

When we solve the same problem for the fourth client, we turn it into a product. Each one ships with its own deployment, its own pricing and its own landing page — and each one is still maintained by the team that built the original.

Document intelligence

Atlas

Turns contracts, claims and dossiers into structured, cited data. Field-level confidence, human review routing and straight-through processing out of the box.

ExtractionCitationsSTP rules

Forecasting

Kuro Forecast

Hierarchical demand and capacity forecasting with probabilistic outputs, promotion effects and automatic reconciliation across planning levels.

DemandProbabilisticScenarios

Agent platform

Mise

The orchestration layer we use internally: typed tools, deterministic workflows around model calls, replayable traces and built-in human checkpoints.

OrchestrationTracesHITL

Industrial vision

Shoku Vision

Edge-deployed inspection for production lines and yards. Ships on commodity hardware, learns from operator corrections, works when the network does not.

EdgeDefectsActive learning

AI assurance

Nori Evals

Evaluation and monitoring for AI systems in production: golden sets, regression gates, groundedness scoring, drift alarms and EU AI Act evidence packs.

EvalsMonitoringCompliance

Knowledge & search

Tanso

Enterprise retrieval that inherits your existing permissions. Hybrid search across documents, tickets and databases, with answers that always cite the source.

Hybrid searchPermissionsGrounded
In development View product
Selected work

Numbers our clients put their name to.

Every figure below was measured by the client's own finance or operations team, against a baseline we agreed before the work started.

Insurance · Europe

Claims triage for a top-five motor insurer

Eleven thousand daily FNOL submissions across five languages, previously read by hand. We built an extraction and severity pipeline with automatic fraud flags and a confidence-gated fast lane.

−63%Time to first decision
74%Straight-through rate

Manufacturing · DACH

Vision inspection across 22 production lines

A pilot cell proved out in six weeks, then rolled to every plant on existing industrial cameras and on-premise GPUs. Operators label corrections inline; the model retrains nightly.

−19%Scrap rate
€6.1MAnnualised saving

Retail · Iberia & LATAM

Forecast and replenishment for 1,400 stores

Replaced a spreadsheet-driven planning process with hierarchical forecasts, promotion modelling and automatic order proposals that planners approve rather than assemble.

−34%Stockouts
−12%Working capital in stock

Healthcare · Nordics

Ambient documentation for a hospital group

Deployed entirely inside the hospital's own data centre. Clinician-authored templates, local language models, and a hard rule that nothing is filed without a human signature.

−41%Documentation time
+31%Clinician satisfaction

Logistics · Global

Exception agent for a freight forwarder

An agent that watches milestone events, detects at-risk shipments, drafts the carrier chase and books the recovery option — escalating to a human when cost exceeds a threshold.

+23ppOn-time delivery
3.1×Shipments per coordinator

Banking · UK

Model governance platform for a challenger bank

A single evaluation and documentation pipeline covering 40+ models, cutting the regulatory submission cycle from quarters to weeks with evidence generated automatically.

9 daysDocs turnaround
40+Models under governance
Client voices

What it is like to work with us

“
We had two failed AI pilots behind us. WasabiLabs spent the first week telling us which of our ideas would not work. That is why the third one shipped.
EK Elena KovačCOO, Meridian Health Group
“
They handed over the repository, the runbooks and the on-call rota. Six months later our own team is shipping changes without them. That was always the deal.
MR Marcus ReinhardtCTO, Kōbe Industrial
“
Our regulator asked for the decision trail on a Friday afternoon. We exported it in twenty minutes. That single moment paid for the engagement.
AF Amara FofanaHead of Risk, Nordbank
Responsible AI

Governance is not a slide at the end.

We work in sectors where a wrong answer has consequences. Safety, privacy and auditability are engineering requirements in our backlog from sprint one — priced in, not bolted on.

Data residency by design

Your data stays in the jurisdiction you nominate. Air-gapped and on-premise deployments are a standard option, not an enterprise upsell.

Traceable decisions

Every output stores its inputs, retrieved evidence, model version and prompt. Replay any decision from any date, for as long as your retention policy requires.

Humans where it counts

We define the confidence thresholds and failure modes that must reach a person, and we design the review interface so that person can act in seconds.

Adversarial testing

Prompt injection, data exfiltration, jailbreak and bias probes run on every release — with the results written into the release notes your auditors read.

ISO/IEC 27001Information security
ISO/IEC 42001AI management system
SOC 2 Type IIAnnual attestation
GDPRDPA & DPIA support
EU AI ActRisk classification & technical file
HIPAABAA available
Company

A small team that has shipped this before.

WasabiLabs was founded in 2019 by engineers and researchers who were tired of AI projects that ended in a slide deck. Forty-six people across Barcelona, Amsterdam and Tallinn. No account managers between you and the people writing the code.

SN

Sara Nakamura

Co-founder & CEO

Fifteen years in applied ML, previously led decision science at a European payments network.

DV

Daniel Vargas

Co-founder & CTO

Distributed systems and MLOps. Built the inference platform behind a 40M-user marketplace.

IL

Ingrid Larsen

Head of AI Research

PhD in probabilistic forecasting. Publishes on calibration and uncertainty in production models.

OA

Omar Aziz

Head of Delivery

Runs the engagement model. Previously delivered core banking migrations across three continents.

PRINCIPLE 01

Ship the boring parts

Monitoring, permissions, retries and rollbacks are what separate a demo from a system. We build those first and treat them as features, not overhead.

PRINCIPLE 02

Your data, your models, your IP

We never train shared models on your data. Contracts assign the work product to you, source included, from the first commit.

PRINCIPLE 03

Say no early

If we think a use case will not repay its cost, we say so in the first workshop. Roughly a third of our proposals recommend a smaller scope than the one we were asked to quote.

Insights

Field notes from production

Written by the engineers on the projects, not by a content agency.

Why your RAG system fails on the twentieth question

Retrieval quality degrades quietly as a corpus grows. A practical guide to chunking strategy, hybrid scoring and the evaluation set that catches the regression before your users do.

Read the note

An EU AI Act technical file you can actually maintain

What high-risk classification means in engineering terms, and how to generate most of the required documentation from artefacts your pipeline already produces.

Read the note

The unit economics of an agent that runs 140k times a month

Token cost is rarely the problem. Routing, caching, retries and the cost of a wrong answer are — with the arithmetic we use to size an engagement.

Read the note
FAQ

The questions we get in every first call

Two weeks for a framed problem with a measurable baseline, four to six more for a working slice on your own data. We do not run six-month discovery phases; if a use case needs one, it is not ready to be an AI project yet.

Usually not. Most of our engagements begin with data that lives in an ERP, a shared drive and three spreadsheets. We build the minimum pipeline the use case needs and leave it behind as reusable infrastructure rather than asking you to finish a two-year platform programme first.

Where you tell us. We deploy into your cloud account, our managed EU estate, your own data centre or a fully air-gapped environment. We never train shared models on client data, and every deployment ships with a data flow diagram your DPO can sign.

Whichever fits the task. A frontier model for open-ended reasoning, a fine-tuned small model for high-volume classification, gradient boosting for tabular problems. Model choice is a routing decision inside the runtime, so swapping one costs a config change rather than a rewrite.

We design for it. Confidence thresholds route uncertain cases to a person, guardrails block actions outside a defined envelope, and every decision is replayable. We agree the acceptable error profile with you before launch and monitor against it continuously.

Fixed price per phase for Frame and Prove, so you can stop cleanly after either. Ship is time and materials against an agreed scope, and Operate is a monthly retainer tied to SLOs. For some products we also offer outcome-linked pricing once a baseline is established.

You do. Source code, models, training data and documentation are assigned to you. Our platform components are licensed perpetually and irrevocably for the deployed use case, with the source escrowed. There is no scenario in which turning off a WasabiLabs subscription turns off your system.

That is our preferred mode. We embed two to five people alongside your engineers, pair on the work, and plan the handover from day one. About 60% of our engagements end with the client's own team operating the system independently.

Get started

Tell us the decision you want to change.

Send us one workflow that costs you money or time. Within five working days you will get a written assessment: whether AI is the right tool, what it would take, what it would cost, and what could go wrong.

  • No obligation and no charge for the assessment
  • Signed NDA before any data is discussed
  • An engineer on the first call, not a salesperson
  • A clear answer if we think you should not do it
Thanks — your request is noted. We reply within two working days.

By submitting you agree to our privacy policy. We never share your details with third parties.