About CoPlanAI

Collaborative planning with AI, for cities.

CoPlanAI is a platform for collaborative planning. Residents work on a real place together with the institution that must decide, in the room and online. A resident’s words become an image of the place in seconds. The platform evaluates it against the institution’s own policy, and people weigh it before anything is decided.

We work with cities, public agencies and international organisations, and run educational programmes for schools, universities and professionals. Built and run in Helsinki by SPIN Unit Lab Oy since 2018.

95 engagements73 cities38 countriessince 2021

A collage of workshop photographs. Group photos: participants in Kyiv holding a Ukrainian flag, and groups in Nusantara and Manila. Around them, people at work: hands on a tablet, residents at touchscreens in Bologna, residents at a laptop in East Cleveland, printed scenarios on a wall in Pristina, and boards and plans on workshop tables.
Workshops in Kyiv, Nusantara, Manila, Bologna, East Cleveland, Pristina, Turku, Kaarina, Dubai and Vienna. Vienna photograph: Urban Innovation Vienna.Workshops in Kyiv, Nusantara, Manila, Bologna, East Cleveland, Pristina and Turku.

Who we are

Researchers and designers, working from Helsinki.

Two co-founders, and the company that builds and runs the platform.

Damiano Cerrone

Co-founder · product design and innovation lead

Doctoral dissertation on collaborative planning with generative AI, 2026. Co-founder of SPIN Unit.

damiano.cerrone@coplanai.com

SPIN Unit Lab Oy

The company

An urban-data and participatory-design studio.

CoPlanAI is a service of SPIN Unit Lab Oy, based in Helsinki since 2018. From 2021 until September 2025, SPIN Unit worked with Toretei on the field application of UrbanistAI, the platform Toretei develops, and some engagements in our record were delivered with it. Ideas and feedback from that fieldwork helped shape the platform’s usability and functionality. The collaboration ended in September 2025.

Visit SPIN Unit (opens in a new tab)

Who we work with

For the people who convene the room.

Ways to work together

  1. 01

    Cities and planners

    Bring residents in while the plan can still change, and leave with evidence a council can act on.

  2. 02

    Innovation offices

    Our usual entry point: the team with the mandate to experiment before the line organisation commits.

  3. 03

    International development

    With UNDP and GIZ, from Nusantara and Skopje to Panama City and Manila.

  4. 04

    Schools, universities and museums

    Educational programmes, research partnerships, and a kiosk mode that runs without a facilitator.

Clients, partners and funders

64 organisations: cities, ministries and agencies, international bodies, universities and companies.

    Ethics and AI governance

    The platform informs the choice. People make it.

    AI in a planning institution changes how the institution decides. So we set out who does what, what we commit to, and how we handle the model’s bias.

    1. 01 · The platform

      Generates and evaluates scenarios

      From people’s own words, on the real place. Each scenario is evaluated against the institution’s own policy, design guides and law, and for its likely environmental and social impact.

    2. 02 · People

      Deliberate

      They compare, argue and vote. They see what each option would do, and can change their minds.

    3. 03 · The institution

      Decides

      It holds the mandate and sets the guardrails every option has to respect. What people agreed arrives documented, as evidence it can act on.

    EU AI Act · limited risk

    A decision-support tool: it informs decisions and takes none.

    Human in the loop, by design

    Four commitments. We ask every partner organisation to write its own.

    1. Augment judgement; never automate it away.

      Every generative step is wrapped in a human step.

    2. Preserve the right to know the consequences, and to change your choice.

      Participants control what the model produces, and see a scenario’s likely impact and trade-offs before anyone chooses.

    3. Make the model’s biases legible before asking anyone to trust its images.

      We show where the model leans, and the room can correct it.

    4. Represent those who cannot be in the room.

      The platform can stand in for future residents and ecosystems, to stress-test a scenario. It never replaces the living public.

    The participant app: a transformed city street with planters, benches and planting, and beside it a Transformation analysis panel listing the impact and the trade-offs, gains and losses.

    The participant app · IterateA scenario with its evaluation beside it: what it would improve, what it would narrow, who gains and who loses.

    Bias, curated in the open

    Choosing a bias is choosing whose city gets imagined more easily.

    Every tuning over-represents some features and under-represents others. We treat bias curation as design work, done in the open, in four practices.

    • Declare

      Every tuning ships with a note on what it favours and what it flattens.

    • Localise

      Tuned with local imagery and policy, with the client.

    • Counter-balance

      Deliberate weight for children, elders and biodiversity, set as open policy choices.

    • Supervise

      Professionals review against regulation and quality before any decision.

    Energy and footprint

    Compute only when we imagine.

    Inference runs only while a workshop is live. The model generates in seconds, then stands down.

    How the platform uses energy

    • Seconds per image, not training runs. We use existing foundation models at workshop scale and train no large models in-house.
    • European cloud. Hosted in AWS Europe regions. By AWS’s own reporting, Amazon matches its global electricity use with renewable energy.
    • Data minimisation. We store only what the service needs, so there is less to keep, move and power.

    Data and compliance

    Hosted in Europe. The outputs belong to the client.

    The platform runs on European cloud infrastructure, to the standards public institutions must meet.

    The data record

    Europe
    Hosting
    AWS, Europe regions, unless agreed otherwise with the client
    Data processor
    AWS, under GDPR
    Frameworks
    GDPR, through a Data Processing Agreement · EU–US Data Privacy Framework
    Personal data
    Only what the service needs: data minimisation, purpose limitation, security by design
    Facial analysis
    None. Facial features in uploaded images are neither collected nor analysed
    Encryption
    In transit and at rest, with access controls
    Your outputs
    Generated content and aggregated preference data belong to the client, to use, modify and distribute freely
    EU AI Act
    Limited risk. A decision-support tool, with transparency, human oversight and bias mitigation
    Request the full data and server sheet

    Service level

    Standard SLA
    Availability
    99.5% per month, during business hours
    Support
    By email to governance@coplanai.com, 09:00–17:00 EET, Monday to Friday
    Critical response
    2 business hours, or 1 hour during a live workshop on standby
    Workshops
    Readiness checked a day ahead; standby support for critical sessions, day or night
    Backups
    Daily · RPO 24 h · RTO 8 business hours

    The headline terms of our standard Service Level Agreement.

    Contact

    Come and say hei.

    One address reaches the whole team. Support, data and compliance have their own.

    Company details

    Company
    SPIN Unit Lab Oy
    Founded
    Helsinki, 2018
    Business ID
    2916199-1
    VAT
    FI29161991
    PIC (EU)
    884696552
    E-invoicing
    003729161991 · Apix Messaging Oy (003723327487)