The problem this solves
After a Discovery Sprint, you know what to build. The Concept Sprint answers the next question: how, with what, and at what cost?
The gap between “great idea” and “approved project” is almost always a technical one. Finance needs a cost estimate. Engineering needs a clear specification. Legal needs to understand what data is involved. The Concept Sprint produces all three, in one session.
Without a concrete architecture, the project stalls. Procurement opens a vendor evaluation. Engineering starts from scratch. Timelines slip from weeks to quarters. The Concept Sprint shortcircuits that cycle by producing a ready-to-approve technical specification before anyone writes a line of code.
Who this is for
You've been handed an approved AI use case and need a real architecture, not a whiteboard sketch. You need a specification your team can actually build from.
You need to validate AWS service selection, estimate infrastructure costs, and produce a PoC spec that finance can approve before committing engineering resources.
You own the timeline. You need a scoped prototype specification with realistic build estimates so you can set stakeholder expectations and resource a team.
How it runs
A typical architecture output
The architecture diagram varies by use case, but this is a representative pattern for an AI document processing or extraction system, one of the most common outputs from Discovery Sprints:
This pattern handles the ingestion, OCR or extraction layer, AI analysis with Bedrock, and integration back into downstream systems. The same structure applies to video processing, transcription pipelines, and multi-agent workflows, with different services at each layer.
What the cost estimate covers
A Concept Sprint cost estimate breaks down into:
- Prototype costs: What it costs to build and run the demo for one month. Typically €200–800 for serverless, AI-heavy workloads.
- Production costs: What it costs at the target scale, with proper monitoring and redundancy. Typically 5–15× prototype costs.
- Data preparation costs: One-time engineering work to get data into the right format. This is often the largest line item and is frequently overlooked.
For projects with a credible production business case, the cost estimate feeds directly into an AWS PoC funding application, where up to €10,000 is available for qualifying projects.
Why architecture before build matters
Building without a clear architecture is how organizations end up rebuilding. The Concept Sprint catches decisions that are cheap to change on paper and expensive to reverse in code.
The PoC specification document produced in this session is the complete brief for the Prototype Sprint. Every technical question, service selection, data flow, cost model, security model, is answered before anyone writes code.
What “done” looks like
A full system diagram: data sources, ingestion, processing layers, storage, API surface, and security model, built to be understood by both engineers and non-technical stakeholders.
Prototype costs (typically €200–800/month) and production costs at target scale, broken out by service. Includes data preparation costs and AWS funding eligibility assessment.
The complete brief for the Prototype Sprint: scope, data inputs, expected outputs, AWS services, build sequence, and acceptance criteria. Shareable with technical and non-technical stakeholders.
Realistic estimates for prototype, hardening, integration, and production launch phases, with team size requirements at each stage. Directly inputs into internal resourcing and budget approval.
"Linda brought both the strategic overview and the engineering execution. We had an architecture concept, cost estimates, and a working demo. All within 3 weeks."
Three hours of structured technical decisions. The output is the document that PoC funding applications get approved on.
Frequently asked questions
The Concept Sprint produces the architecture engineering can build from and finance can fund.