Lisbon has become one of Europe’s busiest places to hire AI talent, and the top AI agencies in Lisbon now range from 150-person consultancies to small senior teams that build one agent at a time.
For a buyer, the hard part is not finding a firm. It is telling apart the agencies that ship AI into production from the ones that added “AI” to a web design menu. If you want the wider national view, including Porto and Coimbra firms, see our list of AI development companies in Portugal.
Every agency below has a headquarters or delivery team in Lisbon or its immediate metro area. We read each firm’s own site and checked for named AI services, case studies with concrete outcomes, and partner or award evidence. We ranked on AI specialization first, then on published results and team depth. Claims we could not confirm on a firm’s own pages are not included.
|
Company |
Lisbon Base |
Best For |
Core Services |
|---|---|---|---|
|
Twistag |
Av. de Berna, Lisbon (HQ) |
Agents that take over back-office work |
Agent engineering, AI readiness, agent governance |
|
DareData |
Lisbon (HQ, remote-first) |
Large enterprises mixing ML and GenAI |
AI discovery, AI MVPs, GenOS platform |
|
Augusta Labs |
Lisbon |
Enterprise and PE-backed AI transformation |
Agentic workflow redesign, embedded AI pods |
|
BySix |
Algés, Lisbon metro |
Customer support and content agents |
AI agents, LLMOps, AI consulting |
|
Imaginary Cloud |
Lisbon office, plus Coimbra and London |
Software products that need ML features |
AI POCs, ML development, MLOps |
|
Altar.io |
Lisbon (HQ) |
Startups building an AI-first product |
LLM workflows, RAG, AI agents, MVPs |
|
Velv |
Lisbon and Porto |
Adding AI engineers to an existing team |
AI nearshoring, AI engineering, AI consulting |
|
Xpand IT |
Lisbon (HQ) |
Microsoft-based data and AI programs |
Data, AI solutions, analytics |
|
gamgi |
Lisbon, with London and New York presence |
SMEs wanting one team from audit to build |
AI audits, custom assistants, automations |
|
Digiton |
Delivers from Lisbon |
Founders hardening an AI app for production |
Agent systems, production reviews, AI governance |
Twistag is based on Av. de Berna in central Lisbon, the city where it was founded in 2016. It began as a product studio, moved into recommendation engines and machine learning, and added LLM and agent work in 2023. Today it calls itself an applied AI company that designs, builds and runs agents and data systems for operational work. The team is small and senior: 25+ engineers, designers and AI specialists with an average of more than eight years in their fields.
What earns the top spot is the published evidence. Twistag lists an accounts payable agent for manufacturer Aralab that processes more than 2,000 invoices a month, an AI communications assistant for a UK water utility that checks every draft against the utility’s guidelines, and a PepTalk platform that folded five tools into one agentic system. Buyers can start with an AI Readiness Sprint, a fixed-scope build, or a monthly “Loop” subscription for smaller AI work. The site shows Anthropic and AWS partner badges and a 4.9 Clutch rating.
DareData lists its headquarters in Lisbon and runs remote first through a network of more than 100 AI and ML engineers. Its process starts with AI discovery and assessment, then moves into AI MVPs scoped at four to eight weeks, custom data science and engineering, and its own GenOS platform for running GenAI solutions in production.
For Lisbon telecom NOS, DareData automated invoice processing so that about 65% of invoices go through end to end without an operator. For Coficab it built a simulation platform that explores tens of thousands of cable designs in seconds. The firm carries Google Cloud, Microsoft and NVIDIA Preferred Partner badges and appeared on the Deloitte Fast 50 Portugal list in 2025. It is the Lisbon agency to call when the buyer is a large enterprise with a mix of predictive and generative use cases.
Augusta Labs was founded in Lisbon in January 2024 and already reports more than 50 AI engineers. Its model is to redesign high-value workflows as agentic systems, deploy them into production with evaluations, and leave the client owning the system as its own IP. Engagements run through small hybrid pods of two to four people, beginning with a six to eight week AI value foundation and an evaluation gate.
Its best-known engagement was announced by Sage in January 2026. Augusta Labs teams are embedded in Sage’s product organization to speed up its AI Center of Excellence, working on Sage Payroll, Sage Active and Sage 300 to deliver agentic workflows, data pipelines and production AI features. The firm says it serves Fortune 500 companies, private equity firms and government agencies. It fits buyers who want a deep, multi-month transformation rather than a single tool.
BySix is based in Algés, on Lisbon’s western edge, and has close to a decade of engineering work behind it. It reports more than 70 AI specialists, engineers and data scientists. The service set is focused: custom generative AI agents from simple chatbots to multi-step agents, AI operations with LLMOps, autoscaling and monitoring across AWS, Azure and Google Cloud, and AI consulting for use case prioritization.
Its case studies are short and specific. A customer support agent made support 52% faster. A newspaper copywriting agent drafts breaking news in under five minutes. An online store publishing agent made e-commerce listings 96% faster across thousands of SKUs. BySix says it has shipped more than 100 projects for over 30 companies, with logos including Audi, Bentley and BNP Paribas.
Imaginary Cloud has a Lisbon office alongside Coimbra and London, about 100 people, and more than 300 products delivered since 2010. Its Applied AI and Machine Learning service covers AI proofs of concept, model development for classification, regression, deep learning and NLP, production engineering with MLOps, and the data engineering behind it all.
The studio’s other angle is how it builds. Its developers work as certified “agentic engineers” using Claude, Gemini, Cursor and GitHub Copilot, with coder, reviewer and DevOps agents preparing pull requests that a technical lead approves. On the ML side, its Environmental Intellect project is credited with saving that firm’s clients more than $200,000 in contractor costs. Buyers who need AI features inside a larger web or SaaS product will find a full product team here.
Altar.io is headquartered in Lisbon and is led by former startup founders. It works like a technical co-founder for early companies, starting with a 15-day product scoping process and moving into an MVP build, custom software and dedicated teams. The firm says more than 60% of its client engagements now include AI in some form.
Its AI service covers LLM workflows, retrieval-augmented generation, AI agents and multi-agent systems, computer vision with multimodal models, classical NLP and MLOps. One published project is Pair, a virtual assistant that answers employee questions in natural language and checks each user’s access rights before returning data. Altar.io shows a Clutch “Top AI Maturation Company 2024” badge.
Velv has built nearshore software teams from Portugal since 2017, with offices in Lisbon and Porto plus Amsterdam and Rio de Janeiro. It describes itself as an OpenAI Select Partner. Its AI offer has three parts: AI nearshoring, where every engineer, lead and PM is trained to spot AI opportunities; AI consulting from LLM integration to workflow design; and AI engineering for agents, RAG, automation and evaluation.
Velv offers two engagement shapes: team extension, which adds AI-enabled senior engineers to your squads, and a managed team that Velv owns and leads. Published work includes a rebuild of the Google Partner Portal in four months with agency Bynd and a 30% delivery speed increase for a Scandinavian biometric company. It fits companies that want Lisbon engineers inside their own process rather than a separate agency project.
Xpand IT was founded in 2003 and is headquartered at Praça Príncipe Perfeito in Lisbon, with offices in Braga, Germany, the UK and Sweden. Data journey and AI are two of its five specialist areas, and it lists Microsoft Partner of the Year awards for 2016 and every year from 2020 to 2024.
Its AI work covers predictive models, automation, advanced analytics and intelligent processing. For card processor UNICRE it used StoryWeaver and generative AI to improve how teams define user stories. Xpand IT also builds its own software products, including Xray. This is the safest Lisbon pick for an enterprise that wants AI delivered inside a Microsoft data estate by a large, established partner.
gamgi is based in Lisbon with a presence in London and New York. Its process runs in four phases: audit, strategy, build and evolve. The audit produces a prioritized AI opportunity map, and the same team then builds chatbots, apps and automations connected to OpenAI, Anthropic, Gemini or Mistral models and to tools such as HubSpot, Slack, WhatsApp and Stripe.
Its case studies are local and concrete. A website assistant for a veterinary clinic answers 60% to 70% of routine questions around the clock. LexAlert monitors the Lisbon prosecutor’s legislative database every three hours for a law firm. A school communication platform replaced manual WhatsApp messaging for a Portuguese education provider. It is a good fit for small and mid-sized Lisbon businesses with a defined first use case.
Digiton delivers from Lisbon, where founder and AI architect Brandon Da Costa is based, and is registered in Tallinn, Estonia. It began in 2018 as a digital marketing studio, moved into automation in 2023 and has built AI systems full time since 2024. Services include custom AI platforms built in phases, agent systems with a kill switch and a named owner, and production engineering for apps first built with Claude, Cursor or Lovable.
Recent work includes an operations platform for Lisbon EV charging network ChaZeMo, running on live charger data since July 2026, and a closed RAG tutor inside Moodle for a university in El Salvador that answers only from each course’s material. The first paid step is a five-day production review, which makes Digiton a low-risk choice for founders whose prototype needs to survive real users.
The Lisbon market is crowded, and many agencies sell similar words. These questions help you compare them on substance. For firms that focus only on agents for large organizations, our list of AI agent development firms for enterprise solutions is a good cross-check.
Ask for two or three systems that real users rely on today, with a volume number attached. Invoices per month, tickets handled, hours saved. Agencies with real work answer this quickly. Vague answers about “pilots” are a warning sign.
Some agencies build on their own platform and license it. Others hand over everything. Both can work, but you should know before you sign. Ask specifically about prompts, evaluation sets and agent configurations, since those carry much of the value in an AI system.
An agent that sends emails or posts invoices needs limits. Ask how the agency scopes permissions, where humans approve actions, how it logs decisions and how it shuts an agent down. Good answers mention evaluation sets, monitoring and named owners.
A Lisbon address does not always mean a Lisbon team. Ask where the people on your project live and work, and whether you can meet them. If in-person workshops matter to you, confirm it in the proposal.
An AI agency usually designs and builds the system: agents, assistants, data pipelines. A consultancy may stop at strategy and a roadmap. Most firms on this list do both, starting with a short assessment and then building. If you only need advice at this stage, see our list of AI consulting firms in Europe.
Often, yes. Lisbon shares the full European working day and the morning of the US East Coast, and several agencies here already serve North American clients. For a list built around US time zone overlap, see our guide to nearshore AI development partners for US companies.
Most Lisbon agencies scope a first engagement at between one and eight weeks. A readiness sprint or audit can take one to three weeks. An MVP or first production agent usually takes four to eight weeks. Larger transformation programs run six to twelve months.
The top AI agencies in Lisbon share one trait: they can point to systems that run every day. Twistag, DareData and Augusta Labs lead on agent and enterprise AI work. BySix, Imaginary Cloud and Altar.io offer strong build teams for agents and AI products. Velv and Xpand IT bring scale, and gamgi and Digiton suit smaller companies with a focused first use case.
Shortlist two or three, ask each for a production reference, and start with a small fixed scope. That approach will tell you more in a month than any sales deck. If you are open to firms outside Portugal, our guide to AI development companies in Europe widens the field.