Skip to content

New Model Advisory · Hamburg · since 2015

Better software already exists.

Startups build better tools than the standard enterprise suppliers do. The hard part is finding the right one, buying it without a two-year procurement cycle, and getting it to run. That is what NMA does, as an advisory boutique for venture clienting and agentic AI.

2015Founded as next media accelerator
78Startups backed from Europe, Israel and the US
7Exits from the portfolio so far
2Pan-European programmes · STADIEM · MEDIA FUTURES

What we do

Three practices, one model.

We started by investing in startups. We now help large organisations buy from them, and build the software that makes that work at scale.

Practice · since 2023

Venture Clienting

Work with startups as a customer, not an investor. No cap table, no board seat. A problem statement, a shortlist and a paid pilot with a decision date.

How the engagement runs →

Practice · since 2026

Agentic AI

Where agents actually pay off, what they must never be allowed to touch, and what to build first. Board-level advisory, then hands-on delivery.

Where agents pay off →

Platform · tectic.ai

Tectic

Our own platform. Agents that run on a written leash, a spending ceiling, an egress allowlist, a human-approval threshold, and file a receipt for every run.

Inside the platform →

Venture clienting

Buy from startups. Don't invest in them.

A venture client purchases. The instrument is a purchase order, not a term sheet, and the return is a working system this quarter rather than a markup in eight years. The difficult part is never the startup — it is writing a problem statement precise enough to shortlist against.

  1. 01 / Frame

    Get the problem right

    Two workshops with the business owner who actually feels the pain. Outcomes: one problem statement, one success metric and a budget envelope. If none of the three survives contact, we stop. And that is a good outcome as well.

    Week 1–2
  2. 02 / Scout

    Find the five that matter

    Our own network first, then the open market. Forty to sixty candidates in, five out, each one contacted and pre-qualified against your metric, not against a category on an analyst's quadrant.

    Week 2–5
  3. 03 / Pilot

    Run it for real

    One startup, one scoped and paid pilot. Real data, real users, a fixed end date, and the go / no-go criteria written down before it starts rather than argued about after.

    Week 5–12
  4. 04 / Scale

    Hand it over cleanly

    Procurement and IT inherit a finished package: security review, reference architecture, commercial terms and an internal owner who already ran the pilot.

    From week 12
Typical engagement — 8 to 12 weeks · one problem · one signed pilotTalk through a use case

Agentic AI

Most agent projects fail on governance, not on the model.

The demo always works. What stops it reaching production is that nobody can answer three questions: what is this thing allowed to spend, what is it allowed to reach, and what is it allowed to decide without a human. We start there, then build.

We advise on it and we ship it. Both partners work with clients directly, and the engineering happens on our own platform rather than on a slide.

Where agents earn their keep

  • Back officeDocument-heavy, repetitive, high volume: contracts, invoices, claims, supplier onboarding.
  • MonitoringMarket, competitor and regulatory watching that has to happen every single day, not when someone remembers.
  • InboundQualifying what arrives by mail, web form and public tender before a human spends an hour on it.
  • ResearchBriefing work that currently costs an analyst a week and is stale by the time it lands.

Formats

Board briefing (half day) · AI readiness assessment (three weeks) · Use-case sprint (two weeks) · Build with our engineers.

tectic.ai — our own platform

Agents you can hand to an auditor.

Tectic is the platform we build and operate ourselves. Every agent deploys against a written policy — a budget ceiling it cannot exceed, an allowlist of what it may reach outside your network, and a threshold above which a person has to approve. Every run produces a receipt: what it did, what it touched, what it cost.

agenttender-watch / eu-public
run2026-09-03 06:04 UTC
sources read143
matches kept6
egress3 allowlisted hosts
budget€0.41 / €5.00
human approvalnot required
statuswithin policy

Policy leash

Spending ceiling, egress allowlist and human-in-the-loop threshold, set before the agent is ever deployed.

Receipts

Every run metered and recorded per agent and per task, so cost and behaviour are reviewable after the fact.

A catalogue

Specific agents for specific jobs, assembled per client and not one general chatbot asked to do everything.

European hosting

Runs on infrastructure we operate ourselves in Europe, which is usually the first question we are asked.

The firm

Eleven years of the same question.

How do you get good technology from a small company into a large one? The answer kept changing, so the model did too.

2015

next media accelerator opens in Hamburg as early-stage program for media and marketing startups

2015-2022

NMA invested in 78 startups from Europe, Israel and the US and connected them to the German-speaking media industry. The leading pan-European program for innovation in media and marketing.

2020-2023

Part of STADIEM and MEDIA FUTURES, two pan-European innovation initiatives under the Horizon 2020 program.

2023

Focus shifts to venture clienting, and beyond media into other industries.

2026

Agentic AI joins the practice, as advisory and as our own platform, Tectic.

Today

7 exits from the portfolio, and more expected. Advisory boutique for venture clienting and agentic AI.

Nico Lumma, Managing Partner at NMA

Nico Lumma

Managing Partner

Has not been offline since 1995. Twenty years in the German startup scene, first in online marketing, then as one of the people who brought blogging and social media to Germany. Scouting, advising and investing since 2000; our expert on products and business models. Nico builds and runs Tectic and the client verticals.

Christoph Hüning, Managing Partner at NMA

Christoph Hüning

Managing Partner

Degree in business mathematics, and a track record as consultant and entrepreneur in digital media and communications since 1999. Christoph connects startups with corporates, agencies and investors, runs NMA operations, and takes clients through their first steps in the agentic AI world.

Get in touch

Tell us the problem, not the technology.

Send one paragraph about what is not working. Within a week you get an honest answer on whether venture clienting, an agent, or neither is the right move. And if it is neither, we will say so.

Office
NMA Venture Capital GmbH
Am Sandtorkai 27
20457 Hamburg, Germany
Platform
tectic.ai