How upestate Built One Shared System for Three Ways of Working

Headquarters

Company Size

Project Duration

Meet upestate, a three-person, architecture-led real-estate developer and asset manager in Hamburg. Between them, the team handles acquisitions, financing, construction, investor reporting, and administration. Each person had developed a different way to organise that work.

That became a constraint when upestate began exploring AI. The team had ambitious ideas for automated investment memos and property reporting, but the context those workflows needed still lived across email, spreadsheets, OneNote, SharePoint, and individual memory.

During an eight-week Notion Transformation, we helped upestate build a shared data foundation and a set of practical workflows around the work the team already did. The result was a workspace they could use, understand, and extend. By the end of the sprint, the first new workflows were already coming from the team itself.

Back To OverviewReal Estate
  • Working with Matthias and John was a genuine pleasure. What we valued most was the open, uncomplicated communication and the honest interest in our goals. You never had the feeling of being just another client; it felt like working with a genuine peer, not a vendor.

  • Three People, One Room, Three Versions of the Truth

    Once a month, upestate’s three-person team prepared investor reports for each property. The previous full round had taken all three people four days.

    The problem wasn’t effort or expertise. The information required to do the work lived across email attachments, spreadsheets, SharePoint folders, OneNote, and three individual ways of keeping track.

    A calculation could be correct when someone created it. Then another copy appeared with a date and someone’s initials in the filename. Which one was current?

    Tasks followed a similar pattern. One person kept a personal list in OneNote. Another worked from the inbox and the context carried in their head. Reporting moved between spreadsheets, documents, and email attachments.

    The team sat in the same office and spoke throughout the day. That still didn’t give them one reliable view of tasks, decisions, files, and current numbers.

    And upestate wanted to go much further.

    The vision was an operating system for the business: information arriving at the front of the acquisition process, a finished investment memo emerging at the other end, with AI helping in between.

    AI needs context. At upestate, too much of that context still lived with individuals.

    We call this the Human Context Window: the finite amount of active context one person can carry. As long as the business relied on that window, an AI assistant could answer one question after a long explanation. The next conversation would start from scratch.

    So the first decision of the transformation was simple: create a shared foundation before automating the impressive stuff.


    Meet upestate

    upestate is an architecture-led real-estate development and asset-management company based in Hamburg. Founded by two friends who met while studying architecture, the company combines architectural and commercial thinking to identify, develop, reposition, and manage properties across northern Germany.

    Its projects range from residential portfolios to mixed-use developments and the revitalisation of existing buildings. Across them, the same idea runs through the work: recognise the architectural and economic potential of a place, then develop it responsibly and sustainably around the people who will use it.

    That is a wide operating surface for three people. It also explains why each person had built a working style around their own responsibilities. The team wasn’t starting from chaos. They had processes, spreadsheets, templates, and years of property knowledge.

    The next system had to preserve that expertise while making it usable across the company.

    Rolf Walzel, managing director at upestate, described the adoption test early in the project:

    It would be genuinely useful for us, even though there are only three of us, if we had one place. And if everyone has to discipline themselves to use that one place, then it has to work well. You don’t use something that doesn’t work well for you.


    A House You Build While Living In It

    Notion and AI operations are bespoke. The workflows depend on how people make decisions, where information enters the company, which exceptions matter, and what “good” looks like in that specific business.

    A consultancy can’t learn that behind a closed door and return eight weeks later with “the solution.” The system has to be fitted while people use it.

    A tailored suit works the same way. The measurements give you a starting point. Then come the fittings. You move, the tailor adjusts, and the shape gets better. Skip that loop and an expensive suit fits even worse than that off-the-rack option in “probably my size”.

    Our projects follow the same principle. We start with a Blueprint, build in short cycles, and take the client through the workspace every week. At upestate, those sessions became construction-site tours. The team could see the rooms taking shape, try the first workflows, and tell us when a wall needed to move.

    John Hodgkinson, our project’s lead consultant, used this (fitting) metaphor to describe our approach during the Blueprint Call:

    We’re not building a turnkey house. We’re building a foundation, the supporting structure, and three rooms that are ready to move into. The rest remains available for future fit-out.

    The foundation connected the information the team needed to share: properties, tasks, meetings, reports, contracts, loans, and deal status.

    The supporting structure established the rules between them. A loan belongs to a property. A meeting produces tasks. A document stays connected to the object it describes. The acquisition history follows a property into the portfolio.

    Existing tools stayed where they were strong. Excel remained the right place for complex financial calculations. SharePoint continued to hold files. Notion became the shared map: the place that showed what existed, how it connected, and which information was current.

    Then we focused on the rooms people would use every day.


    Three Rooms With Daily Pull

    Meetings That Leave Work Behind

    The first room connected meetings, notes, and tasks.

    A recorded meeting could produce proposed actions, wait for human confirmation, and link the approved tasks back to their source. The team rebuilt its Monday operating meeting inside the workspace using the agenda it already knew.

    That changed what a meeting left behind. Decisions and tasks no longer depended on someone writing separate notes and copying actions into a personal list. The existing ritual now created reusable company context.

    And a personal dashboard answered the question that matters on Monday morning: what needs my attention?

    A Portfolio Spine for Reporting

    The second room created a shared structure for the property portfolio.

    Portfolios connect to properties. Properties connect to buildings and units. Contracts, loans, reporting periods, and key documents attach at the level where they matter. The model follows how upestate acquires and manages real assets instead of forcing the business into a generic CRM.

    This is a complex flow – because real estate development is a complex business.

    A Notion System needs to represent your business reality. As simple as possible, but complex enough so that you’re not constantly feeling “ah yes, but ACTUALLY it’s like this…” – the sentence which describes the daily user experience with most out-of-the box tools.

    The team also tested AI-assisted extraction from real documents. Selected information from contracts and loans could be proposed for the relevant records, with every imported value marked for human verification.

    The full monthly reporting cycle was still running through parts of the established process at the end of the sprint. But it now had a data spine to draw from. The report no longer had to become another isolated file with its own version of the truth.

    Acquisition Context That Compounds

    The third room connected acquisition work to the information already gathered elsewhere.

    Instead of opening documents and searching them one by one, the team could begin asking questions across the context it had brought together.

    Rolf tested a question about annual rent excluding service charges. The answer came from an email and its attachment from the previous September.

    Instead of me going in and searching myself, it worked it out.

    Fiete Bengelsdorf tried a purchase contract: “I say, here’s a purchase contract, please look, there’s some lease issue in it—and it finds it. That’s wild.

    The files remained in SharePoint. The difference was that the folder structure no longer had to serve as the team’s primary map of the business. Notion held the relationships and context that made those files easier to find and use.

    The end-to-end investment-memo workflow was still ahead. The ingredients now had somewhere to accumulate.


    The Team Started Building Without Us

    Five days after the workshop on writing AI skills, Fiete returned with one nobody had asked him to build.

    It reconciled broker offers arriving in his inbox against the acquisition pipeline. Rolf had begun experimenting with a task-check skill and a personal weekly overview. These weren’t items from our delivery plan. They were workflows the team had identified and started extending themselves.

    That’s exactly what we want to see – because it means the system gains traction.

    At the beginning of the project, upestate wanted an operating system for the company. By the end of the first phase, the team had started learning how to build inside it.

    Particularly when it comes to onboarding AI to your company, it’s helpful to internalise a few specific perspectives. Here’s an idea that really resonated with Fiete:

    “You tell it what to do, then you ask: did you understand, what else do you need? That idea helps enormously.”

    In the final conversation, he described starting with a question, discovering which information Notion was missing, and then adding the relevant document or master data. Before ending the session, he checked that the new context had been stored in the right place.

    One conversation made the next conversation easier.

    This is why we begin with Skills Before Agents. A person runs the process, sees where it breaks, improves the instructions, and decides which judgement calls must remain human. More autonomy comes once the workflow has survived real work.


    Ready for the next building phase

    Workspaces – particularly in the AI age – always grow at different speeds in different areas.

    That’s normal. Building out AI-enabled ops takes time and attention. But every bit starts to compound and before you know it, your whole way of working has changed in a fundamental way.

    The first phase at upestate did exactly this.

    • one shared structure for the company’s operational context;
    • working flows for meetings, tasks, property information, and document-assisted research;
    • a team that understood the architecture well enough to question it and extend it;
    • clear next candidates for reporting, acquisition analysis, and document integration.

    And nothing is more exciting than when things click and the team sees the potential for growth come true:

    “What makes me happiest is that we’ve been trying this for two years now, two and a half. And this is the first time we’re heading in a direction where it gets genuinely exciting.”

    Here’s to upestate – we can’t wait to see how their new building takes shape and evolves over the coming weeks & months.

    Need a Notion Consultant for Real Estate? Let's talk!

    Browse More Case Studies

    Deutsch