Documented engagement model
AI Agencies for Fixed-Scope Projects in Europe (September 2026)
Agencies that document project-based engagements on their own sites or in published case studies, ranked by portfolio quality and credibility.
Each listed agency documents project-based engagements on its own site or in a published case study; the model is never inferred from a service list.
Top picks
- 1. Merantix Momentum — Berlin · 19 documented clients
- 2. Satalia — London · 6 documented clients
- 3. Datatonic — London · 8 documented clients
Documented clients include
Tesco · KRASS Optik · Deloitte · AstraZeneca · TÜV Rheinland · Bill and Melinda Gates Foundation
Ranked agencies for project-based engagements
Rankings updated · Newest review
Top 30 of 96 agencies with documented project-based work.
Berlin, Germany
Client evidence: 19 documented clients overall
Best for: German enterprises and federal or city administrations taking a machine-learning or LLM use case from feasibility to an operated system.
Satalia
#2London, United Kingdom
Client evidence: 6 documented clients overall
Best for: Enterprises with a scheduling, routing or workforce-allocation problem big enough to justify operations-research specialists.
London, United Kingdom
Client evidence: 8 documented clients overall
Best for: Enterprises standardizing ML delivery on Google Cloud or AWS, especially where a regulator has to be satisfied.
Unit8
#4Lausanne, Switzerland
Client evidence: 5 documented clients overall
Best for: Industrial and financial enterprises building on Palantir Foundry, or weighing whether they should.
Faculty
#5London, United Kingdom
Client evidence: 3 documented clients overall
Best for: Enterprises and public bodies that need AI deployed under regulatory or safety scrutiny.
Munich, Germany
Client evidence: 4 documented clients overall
Best for: Regulated German industrial and mid-market firms that want the model running inside their own Azure or AWS tenant.
Visium
#7Zurich, Switzerland
Client evidence: 5 documented clients overall
Best for: Pharma and life-sciences teams that need the method described in full when the client name cannot be.
Helsinki, Finland
Client evidence: 5 documented clients overall
Best for: Regulated Finnish and Nordic enterprises that need an AI system governed, certified and operated, not just delivered.
London, United Kingdom
Client evidence: 8 documented clients overall
Best for: Pharma, research and public-sector buyers with an NLP problem on unstructured documents that has to survive review.
dida
#10Berlin, Germany
Client evidence: 3 documented clients overall
Best for: Organizations with a hard perception or document problem who want the method documented, not just the result.
Madrid, Spain
Client evidence: 10 documented clients overall
Best for: Spanish utilities, public hospitals and large employers bringing a forecasting, clinical-data or Spanish-language modeling problem.
Ergo
#12Dublin, Ireland
Client evidence: 4 documented clients overall
Best for: Irish public institutions and regulated financial firms adding AI to a Microsoft estate they already run.
Ghent, Belgium
Client evidence: 5 documented clients overall
Best for: Enterprises adding AI to a product or support workflow they already run, with the delivery team embedded alongside their own.
Squirro
#14Zurich, Switzerland
Client evidence: 4 documented clients overall
Best for: Regulated enterprises buying a retrieval and knowledge platform with the delivery work attached, not a bespoke build.
Zühlke
#15Zurich, Switzerland
Client evidence: 8 documented clients overall
Best for: Regulated enterprises buying AI inside a system an engineering firm will also build, run and document.
Sparkbit
#16Warsaw, Poland
Client evidence: 5 documented clients overall
Best for: Early-stage product companies that need the ML core of their product built, tested and handed over with the IP.
Dublin, Ireland
Client evidence: 20 documented clients overall
Best for: Irish and European SMEs and founders buying a whole product build, with AI as one layer inside it rather than the whole engagement.
Kortical
#18London, United Kingdom
Client evidence: 8 documented clients overall
Best for: Enterprises with a defined prediction or document-automation problem that want a model in production in weeks.
Imobisoft
#19London, United Kingdom
Client evidence: 4 documented clients overall
Best for: Health and industrial companies taking a risk-scoring or compliance-reporting workflow from paper to a regulated digital platform.
Brainpool
#20London, United Kingdom
Client evidence: 3 documented clients overall
Best for: Operators with a measurable production bottleneck AI can remove—from timber design queues to e-commerce pipelines.
elunic
#21Munich, Germany
Client evidence: 6 documented clients overall
Best for: Manufacturers automating one inspection station or technical-document search on an existing line, not buying an open-ended AI program.
Inspari
#22Copenhagen, Denmark
Client evidence: 3 documented clients overall
Best for: Nordic enterprises already running a Microsoft or Snowflake data platform that want AI built on top of it, not a standalone pilot.
Dmlab
#23Budapest, Hungary
Client evidence: 2 documented clients overall
Best for: Hungarian operators in energy, logistics or media that need a forecasting system wired into how the business already trades.
Limebit
#24Berlin, Germany
Client evidence: 2 documented clients overall
Best for: Pharmaceutical companies and statutory health insurers that need analysis to survive a regulatory review and the code to come with it.
Datasparq
#25London, United Kingdom
Client evidence: 6 documented clients overall
Best for: Operational businesses with data already flowing that want a system built on it rather than a strategy deck.
Flobotics
#26Warsaw, Poland
Client evidence: 3 documented clients overall
Best for: Operations teams that want automation bought against a payback number, especially in healthcare revenue cycle.
craftworks
#27Vienna, Austria
Client evidence: 5 documented clients overall
Best for: Industrial operators who need a vision or maintenance system running on their own lines and, where required, on their own infrastructure.
10Clouds
#28Warsaw, Poland
Client evidence: 5 documented clients overall
Best for: Banks, insurers and credit funds automating a defined back-office or sales workflow on a platform already run in production.
Hiflylabs
#29Budapest, Hungary
Client evidence: 8 documented clients overall
Best for: Banks, energy and manufacturing groups pairing an LLM or forecasting build with the data-platform work underneath it.
Lautmaler
#30Berlin, Germany
Client evidence: 8 documented clients overall
Best for: Enterprises putting voice or chat in front of real customers, where a wrong answer reaches the public.
About this list
A project is the default shape of AI work and the easiest to buy badly: a defined scope, a delivery date, a handover, and then a system that nobody on the client side can retrain. The agencies on this page document project engagements with a beginning and an end on their own sites or in published case studies, and the useful question is not whether they run projects but what their projects leave behind. The chip is a documented engagement model, never an inference from a service list.
The best-documented projects here shipped systems that are still running years later. Satalia's last-mile engine for Tesco, built in 2015, still schedules 100,000 deliveries a day. Merantix Momentum's recommender for KRASS Optik has served 900,000+ customers daily on automated retraining since 2021, and its vehicle-damage detection for TÜV Rheinland shipped as a product the same year. Kortical's tax-automation build for Deloitte took a computation from five hours to six minutes over a six-month engagement, with a partner quoted by name. Datatonic's retrieval pipeline for AstraZeneca ingested more than 20,000 scientific documents on AWS Bedrock, and Fast Data Science's Clinical Trial Risk Tool for the Gates Foundation was deployed publicly and open-sourced under MIT.
Read a project proposal for the handover. Who runs the system after go-live, and did the agency document that in earlier work? Are the evaluation sets, pipelines and retraining procedures deliverables, or do they stay on the agency's side? And is the scope a system or a proof of concept? A pilot that never shipped is the modal AI project outcome in this market, and the agencies ranked highest here are the ones whose case studies describe operations, measured results and adoption rather than a demo. Agencies that also document retainer or team-extension work appear on those pages; for a bounded build with a clear end, this is the list. Scores rank portfolio quality and credibility; placement is never sold.
Intro by Gabor Kiss, curator · How we rank
Expert Insight
Why the engagement model matters
A fixed scope protects both sides only when the acceptance criteria are written: ask for them before the rate, because a project without them becomes a retainer nobody agreed to.
The handover decides whether the project was an asset or a dependency: pipelines, evaluation sets and retraining procedures belong in the deliverables list.
The pilot-to-production gap is the main risk of project work in AI: the case studies here that describe operations and measured results are the ones that show an agency has crossed it.
Frequently asked questions
A documented engagement with a defined scope, a delivery and a handover: a model built and deployed, a system shipped as a product, a proof of concept delivered for integration. The agency describes the model on its own site or a published case study documents it, with the source URL stored per client. Most agencies on this register run projects; the chip marks the ones whose published work documents that shape rather than asserting it.
The documented builds here range from a few weeks for a scoped agent or classifier to six to eighteen months for a platform: Kortical's Deloitte engagement ran six months, Datatonic's MLOps platform for Vodafone eighteen. A discovery phase longer than the build is a sign the proposal is sized for a different buyer. Ask for a dated plan with a named handover point.
From the hourly bands each agency publishes and a written scope with acceptance criteria. Published rates for the agencies on this page span the European range on our rate index, and a project total depends more on the data work around the model than on the model itself. Treat a quote without a scope as a retainer in disguise.
Code and infrastructure definitions, the evaluation sets and metrics the acceptance was measured against, retraining and monitoring procedures, and documentation a new team can operate from. The case studies here include handovers of exactly that kind: EORTC received In The Pocket's protocol tools in its own environment, and Merantix Momentum's proof of concept for the Bundeskanzleramt was handed over for integration.