Two ways to understand your organization’s metabolism. Same diagnostic engine, same 2,200+ case repository. You choose how it’s delivered.
● Available Now — Alpha
Run it on your own business or department, using your own data
Using Claude Desktop, point it at a folder containing board packs, strategy memos, and/or due diligence reports. The plugin extracts decision episodes locally, scores them against the repository, anonymises and benchmarks against our GDPR compliant engine, and returns a full diagnostic. Nothing traceable or privacy sensitive leaves your machine. 1
Learn More● Available Now — Alpha
Board-ready, from existing artefacts
This is a bespoke process where we co-design the analysis in breadth, depth and scope. This is particularly relevant in pre-deal environments or before embarking on a large-scale transformation initiative. After mobilisation we will be able to tell you within weeks if your initiative has potential or is DOA. You will receive:
Archetype
Classification
Load
Absorption
Capacity
Forward
Degradation
Signals
Intervention
Window
& Timing
1 The self-serve plugin runs on Claude Desktop. Before using it on confidential documents, confirm with your CISO or data privacy officer that your organisation’s AI usage policies permit processing these materials through a large language model. DecisionDNA does not assume liability for data shared with the underlying model provider.
On the Roadmap
Quarterly metabolic pulse across portfolio companies. Detect degradation before it reaches the P&L.
Express interestPrivacy & Security
The diagnostic is designed around a single principle: your documents never leave your machine.
PDF, DOCX, PPTX — every source file remains on your workstation. The plugin reads them locally.
Document content is parsed and processed entirely client-side. No text is transmitted.
The plugin identifies and structures decision episodes from your documents locally.
Numerical feature vectors — the six dimension scores as numbers. No text, no context, no identifiable content.
Aggregate counts — how many decisions of each type and their status distributions. No narrative content.
Statistical measures of language patterns. Not the language itself.
No personal data transmitted. Anonymized payloads retained 90 days, then auto-purged. Immediate deletion available via diagnostic UUID (Article 17).
Scoring infrastructure on GCP europe-west4 (Netherlands). Data does not leave EU jurisdiction.
API gated through Cloudflare Access. Authenticated access, identity-blind payloads. No document reconstruction possible from transmitted data.