Digvijay Singh
AI & Growth StrategyMadrid --:--
Work with me ES
← HomeFor businesses

Find what pays. Build it. Hand it over.

For scale-ups, growing businesses and teams bringing AI into real work. I do the strategy and the build myself, so nothing gets lost between a slide deck and a developer. You end up with something that works in your business, and a team that can run it.

  1. 01

    AI opportunity sprint

    Teams that suspect AI could help but do not yet know where it pays.

    • Interviews with the people who do the work, and a map of where time and money actually go.
    • A shortlist of use cases, each scored on value, feasibility, data readiness and risk.
    • A business case for the strongest one: what it costs to run, what it saves, and the metric that will say it worked.
    • A working prototype on a sample of your own data, with a small labelled test set that shows where it fails.
    • A clear go or no-go, with a pilot plan if it is a go.
    Timeline
    2–3 weeks
    You end with
    A yes or no on one AI use case, backed by numbers and a prototype.
    See: Location Intelligence →
  2. 02

    90-day AI pilot, forward-deployed

    Teams with a use case worth testing that need it working on real data, inside the real workflow.

    • I work inside your team: in your tools, on your data, with the people who will use it.
    • A data pipeline with checks that stop a bad build before anyone sees it.
    • An evaluation set labelled with your experts and reported with intervals, so we know how good it is before anyone relies on it.
    • Deployment your data rules allow: a hosted API, or models running on your own servers.
    • Handover: runbooks, training and a named owner on your side.
    Timeline
    90 days, reviewed at 30 and 60
    You end with
    A system running in a real workflow, measured against the metric agreed on day one, and owned by your team.
    See: Licita →

How an engagement runs

  1. 01

    Talk

    A short call about the problem and the decision it feeds, not about tools.

  2. 02

    Scope

    One page: what we build, how we measure it, what it costs and when it ends.

  3. 03

    Build

    Short cycles on real data. You see working output every week, not status slides.

  4. 04

    Hand over

    Documentation, training and a named owner, so it keeps working after I leave.

Questions

Do you work alone?

Mostly, yes. You get one person for the strategy and the build, which keeps things fast and means nothing is lost in a handover. If a project needs a specialist I do not cover, I say so in the scope and we agree who brings them in.

Which tools and models do you use?

Whatever fits your data rules and your budget. I build with Claude, Mistral and open-weight models, in Python, SQL and plain web code. If your data cannot leave your network, the model runs on your servers. I choose the smallest model that passes the evals, not the most famous one.

Who owns the code and the data?

You do. The code, prompts, test sets and documentation are yours at handover. Your data is used only for your project and stays where your rules say it must.

How do we start?

Send me a few lines on LinkedIn about the problem. We have a short call, and if there is a fit, I send a one-page scope with the timeline and the cost. Every package is scoped per project.

Start with the problem.

A few lines on what is not working are enough. If I am not the right person for it, I will say so.

Message me on LinkedIn