What is and what's not Tech R&D

In the tech world, almost everyone claims to be “doing R&D.” But if you look closely, most companies are actually doing product development — implementing known methods, combining existing tools, and packaging them in new ways.

Real Research and Development (R&D) in technology goes deeper. It’s about solving problems that have no established solution, advancing knowledge, and taking measurable technical risks.

This post breaks down what true tech R&D looks like, explains the Technology Readiness Level (TRL 1–9) framework, explores what qualifies as AI R&D, and shows when tech projects actually meet the criteria for EU or national innovation funding.

1. What R&D Means in a Tech Context

In technology, R&D means using structured investigation to create new knowledge, methods, or systems — not just new apps or products. If your work involves uncertainty (you don’t yet know if it will work), experimentation, and systematic testing, you’re likely in R&D territory.

If you’re mainly combining existing APIs, building standard front-ends, or implementing known algorithms, you’re probably doing product engineering, not R&D.

To put it simply:

- Tech R&D = advancing what’s technically possible

- Product development = applying what’s already known

Both are valuable — but only the first one is eligible for R&D funding. The other could be eligible in other funnding programmes but you need to know what is and what's not R & D so you can find the right funding programme for you.

2. The OECD’s Frascati Manual: The Global R&D Benchmark

The OECD’s Frascati Manual — used by the European Commission and funding agencies — defines what counts as R&D. According to it, R&D must include:

1. Novelty – creating something new to the state of the art

2. Creativity – not just routine application of known tools

3. Uncertainty – outcomes are unpredictable

4. Systematic investigation – structured, hypothesis-driven process

5. Transferability or reproducibility – results can be shared or validated

For tech, that might mean:

- Inventing a new algorithm or training technique

- Creating a data compression or encryption method that breaks performance limits

- Experimenting with new architectures, protocols, or data models

If you’re doing that kind of work — pushing beyond existing solutions — you’re clearly in R&D.

3. The Three Phases of Tech R&D

Most digital R&D projects move through three overlapping phases:

1. Basic Research (TRL 1–3) – Exploring new computational ideas, models, or architectures.

2. Applied Research (TRL 3–6) – Using that research to solve practical problems.

3. Experimental Development (TRL 6–9) – Building prototypes, testing performance, and validating in real environments.

Funding bodies use this scale to assess risk and maturity — many grants focus on projects moving from TRL 3 to TRL 6, when technology is still experimental but has clear commercial potential.

4. The TRL Framework for Software and AI

Technology Readiness Levels (TRL 1–9) describe R&D maturity:

1. Basic principles observed

2. Concept formulated

3. Proof of concept

4. Lab validation

5. Validation in relevant environment

6. Demonstration in relevant environment

7. Prototype in operational environment

8. System complete and qualified

9. Proven system

Knowing where your project sits on this ladder helps you decide if it’s ready for funding, scaling, or commercialization.

5. What Actually Counts as R&D in Tech

True tech R&D is rooted in uncertainty. You’re trying to achieve a technical goal that’s not straightforward and not guaranteed to work.

Examples of what qualifies as R&D in technology:

- Developing new algorithms or optimizing existing ones beyond known results.

- Inventing novel data-handling, storage, or transfer techniques.

- Creating new architectures (e.g., distributed AI systems, federated learning).

- Experimenting with AI explainability, fairness, or interpretability methods.

- Testing innovative integrations of emerging technologies (e.g., edge AI + blockchain).

- Creating performance breakthroughs in computation, scalability, or automation.

Usually not R&D:

- Writing standard code or UI components

- Integrating pre-built APIs (like ChatGPT, Google Vision, or Stripe)

- Migrating legacy systems

- Building products around existing open-source models without modification

- Routine debugging or performance tuning

If your team could deliver the same outcome by following documentation — it’s probably implementation, not R&D.

6. What Counts as R&D in Artificial Intelligence

AI sits at the heart of most tech R&D today. But not every AI project is R&D — most are applications of R&D, not the R&D itself.

Not R&D:

- Using an existing model (like GPT or BERT) via API.

- Training off-the-shelf models on your dataset without architectural change.

- Applying standard classification or prediction pipelines.

Is R&D:

- Designing new model architectures.

- Developing novel training techniques.

- Reducing computational or data requirements through innovation.

- Exploring new forms of explainability or bias mitigation.

- Creating domain-specific AI systems with measurable breakthroughs.

If you’re uncertain whether your approach will work and you’re running structured experiments to find out, you’re doing R&D.

7. Is AI R&D Within Reach of a Normal Tech Team?

Yes — modern open-source frameworks and cloud compute have democratized R&D. Small teams can now run legitimate research sprints if they approach the work systematically.

How to structure AI R&D:

1. Start with a hypothesis.

2. Design controlled experiments.

3. Benchmark results.

4. Document findings.

5. Iterate.

R&D doesn’t require a lab coat. It requires curiosity, structure, and evidence.

8. Quick Diagnostic: Is It Tech R&D or Product Development?

Use this checklist to test your project’s nature:

- Are you solving an unsolved technical problem?

- Is the outcome uncertain?

- Are you designing experiments and measuring results?

- Are you generating new technical knowledge?

- Could your findings contribute to your field?

If yes to most, it’s R&D. If not, it’s development.

9. When Does Tech R&D Qualify for EU Funding?

To qualify for tech R&D funding, your project must:

1. Advance the state of the art.

2. Carry technical or scientific uncertainty.

3. Follow a systematic research process.

4. Produce new, transferable knowledge.

For instance:

- A new deep-learning approach for anomaly detection ✅

- An AI model that explains its reasoning in real time ✅

- A platform that just integrates APIs ❌

Funding assessors look for one thing above all: is this project creating knowledge or just applying it?

10. Turning Tech Innovation Into Fundable R&D

If you’re doing something innovative but not yet formalized as R&D, here’s how to elevate it:

- Frame your work as an experiment.

- Document uncertainty.

- Compare with the state of the art.

- Plan TRL progression.

- Collaborate strategically.

Good documentation and clear methodology are as important as good code.

11. Find Out if Your Tech R&D is Eligible for Funding

If your company is experimenting with algorithms, data models, or architectures, check whether your work qualifies as R&D under EU or national criteria. Projects that create new technical knowledge can access significant support through innovation programs.

And if your work focuses more on adopting or implementing existing tech — that’s still valuable, but it falls under digital transformation funding, not R&D grants. Knowing which applies could open major funding opportunities.