One API call captures what was expressed — not inferred — and issues a portable .ddna record with consent, provenance, and retention rules built in. Designed for reuse. Designed for audit.
Today's memory layers capture feelings, moods, and emotional signals — but they do so without consent, provenance, or portability. That data is either discarded, retained illegally, or locked inside a single system. Under the EU AI Act, inferring emotional states in workplace and education contexts becomes prohibited by August 2026.
.ddna record — not inferredA .ddna record is a portable artifact for emotional context. Issue it once, and any system can consume it for its own purpose — without restructuring data or rebuilding memory.
Query by resonance
Retrieve memories by emotional significance, not just keywords or timestamps. Gravity-weighted recall surfaces what matters.
Consent in every record
Jurisdiction, consent basis, retention rules, and deletion rights travel with the data. Audit-ready from issuance.
Portable across systems
Records travel with users via AuraID. Switch therapists, change apps, keep emotional history. Like moving a PDF, not exporting a database.
Therapy & Health
Clinical context, governed
Session continuity for mental health platforms. Emotional themes carry forward between sessions and across providers — HIPAA-ready, non-inferential.
Companion AI
Emotional depth without explicit state
Give AI companions rich emotional context without engineering your own emotional memory logic. The .ddna record is the memory.
Legacy & Memoir
Preserve affective context
Life story preservation as structured emotional records. Durable, portable, owned by the subject.
Human-like Service
Surface affective context, not just what was said
Customer service and support AI that references affective context — empathetic interactions grounded in expressed significance.
.ddna record in one call_You already store conversations. The Extraction API captures the emotional context that was expressed and returns a structured artifact. Then you issue a portable .ddna record with consent, provenance, and retention rules built in. Non-inferential. No behavioral prediction.
import { DeepaData } from "@deepadata/sdk";
const client = new DeepaData({ apiKey: "dda_live_..." });
// Extract emotional content from text
const artifact = await client.extract({
content: "Today I finally told my therapist about the grief.
It felt like releasing something I'd held for years.",
});
// Issue a governed .ddna envelope
const envelope = await client.issue({
artifact,
pathway: "subject",
authority: "user:jane_doe",
});
console.log(envelope.certificate_id);
// "cert_a1b2c3d4..."EDM captures what most formats miss: not just what was said, but what it meant — as expressed context, not inferred state. It is non-inferential, governance-first, and portable. Published on Zenodo with a DOI and MIT licensed. Implement it yourself, or use our hosted infrastructure.
10 domains. 96 fields. Designed for verification and reuse.
EDM is to affective context what JSON is to structured data. DeepaData is a commercial implementation — the standard belongs to everyone.
If your system infers emotional state in workplace or education settings, you may not be able to retain that data. If your system captures what was expressed and issues it as a governed record — you can retain it, reuse it, and audit it. That's the DeepaData distinction.
Article 5(1)(f) prohibits emotion inference in workplace and education contexts. DeepaData issues records of expressed context — not inferred state — so they can be retained and audited as governed artifacts.
Jurisdiction, consent_basis, and subject_rights fields support Article 17 (right to erasure). Governance travels with the artifact.
Policy labels, masking rules, and k-anonymity support handling PHI for healthcare and mental health applications.
Retention and exportability controls support right-to-delete and portability. Rules are embedded in the record.
“We capture what's expressed, not what's inferred — so you can keep what you're allowed to keep.”
Keep your current data model. Call the API where it matters, issue a .ddna record, and carry that context forward. No migration. No rebuild.
Using Mem0, Zep, or custom memory? Extract expressed emotional context and store the governed record alongside what you already persist.
Your first issuance
Issue a .ddna record from a conversation your memory layer already stores.
Running multi-agent workflows? Issue a .ddna record at handoff so the next agent inherits context that is governed, timestamped, and auditable.
Your first issuance
Issue a .ddna record at the moment an agent hands off a session.
Therapy or coaching? Issue a .ddna record after each session so the next session — or the next provider — starts with structured context, not raw transcript.
Your first issuance
Issue a post-session .ddna record capturing themes for the next session.
Each record carries a cryptographic proof (eddsa-jcs-2022). Verify any artifact independently.
EDM v0.4 is published on Zenodo with DOI 10.5281/zenodo.17808652 — citable and permanent.
Schema and signing tools are MIT licensed. Inspect it, fork it, extend it.
226 tests passing across SDK, signing tools, and MCP server — built for production use.
.ddna records_Early access is free. No credit card. Or work with us as a design partner and shape the issuance model.
Get an API key and issue governed affective records in minutes. 100 extractions/month free. Extract, validate, issue, and verify — end to end.
+ REQUEST API KEYBuilding something where emotional context matters? We're working with a small group of teams to refine the issuance API and certification model. Direct access, priority support, and roadmap input.
+ APPLY AS DESIGN PARTNERIssue non-inferential records you can retain and audit. Get an API key — or join the waitlist for updates.
EU AI Act enforcement begins August 2026 for prohibited emotion inference in workplace and education contexts.