What's new

AI-backed prompt, timeline, study, population, and survey flows now accept optional model selection controls

Image generation workflows now support approved model choices in addition to platform defaults

Automation destinations now support event subscriptions, test sends, and replayable delivery logs for outbound workflows

Survey-fill integrations can dispatch structured Typeform and Qualtrics-style payloads from persona responses

Wikipedia-assisted persona creation now extracts factual profile data with source provenance before draft generation

Country-based Wikipedia discovery can enqueue multiple eligible profiles to seed larger draft cohorts

When a requested model cannot be applied, requests now continue on auto behavior instead of failing hard

Fallback outcomes now include response warning metadata so clients can audit downgrade behavior safely

SDK and API contracts stay aligned with generated request types for model-aware integrations

Persona draft review now provides clearer approve, revise, and discard decision handling before finalization

Queued AI workflows now preserve selected model settings consistently across callback execution paths

Population task and study paths now enforce soft-delete guards so archived records do not leak into runtime operations

Temporary API keys with configurable expiration windows for short-lived integrations and handoffs

Permission and role-aware API keys so each key can be scoped to a specific purpose (for example prompt-only or read-only)

R SDK for researchers and data scientists — install via devtools with httr2-based HTTP client, built-in retries, and full API coverage

Auto-generated types across all SDKs — TypeScript types and Python Pydantic models are now generated from the OpenAPI spec to stay in sync with the API

TypeScript SDK v2.0 with openapi-fetch under the hood, streaming support, and expanded endpoint coverage for populations and studies

Dashboard key creation now includes access profiles, optional role assignment, and expiration controls in one flow

API authorization now enforces key permissions across route groups to reduce accidental over-privileged access

Python SDK promoted from beta to stable with retry logic, default tenant ID support, and expanded API surface

All SDKs now include built-in retries with exponential backoff and Retry-After header support on 429/5xx responses

SDK test harnesses integrated into the build pipeline for continuous verification across TypeScript, Python, and R

Widget key verification now recognizes prompt permissions in addition to legacy widget scope metadata

Context source composer for studies now supports text, URL references, and document IDs on both dashboard and generator surfaces

Study status guidance is clearer (draft, ready, archived), including safer run gating and better visual cues

Run history now includes at-a-glance summary cards for completion, failures, and average duration

Mutation workflows now provide action-level error feedback with one-click retry for create, update, run start, and export prepare

Export actions now include clearer format guidance and preparation-state feedback for faster downstream use

Unknown study statuses now default to a safe non-runnable state in the UI to prevent accidental run starts

Optional multi-party skill sessions for debate, negotiation, and cooperative exercises with structured turn flow

Participant subset selection across individual persona runs, population runs, and study runs

Turn transcript views for population and study runs across dashboard and generator surfaces

Run creation now validates participant subsets up front to prevent invalid run starts

Execution mode now aligns automatically with selected participants and skill configuration

More resilient run-state handling for pause, resume, and cancellation during long-running sessions

Studies in dashboard and generator — save one population + one skill + one context as a reusable unit

Study run history with status tracking so teams can monitor repeat runs over time

Collated study output exports in JSON, JSONL, and CSV with one-click download for downstream processing

Study runs now snapshot context and web-access settings per run for clearer historical interpretation

Prepared exports are cached for fast repeat downloads and archived for retrieval after processing

Safer run execution by blocking archived studies and validating callback/run linkage before processing

Expanded route and contract smoke coverage for studies across web, generator, and API surfaces

Skills naming and interface updates across dashboard and generator for clearer workflow semantics

Skill versions and lifecycle states (draft, published, archived) to manage evolution safely

Runtime controls for population runs with pause, resume, and cancel actions

Execution traces and evaluation scores per run for faster debugging and quality analysis

Typed runtime input and structured output validation with surfaced schema issues per result

Population aggregates now include average quality and schema-validation health signals

Generator workspace sidebar navigation — persona panels are now individually routable pages with a persistent sidebar for faster context-switching

Demographic model editor — visually compose and adjust population distributions directly in the dashboard

New-population creation flow in the web dashboard with guided setup and template selection

Use cases page — browse common research patterns and example workflows

Public status page for real-time platform health visibility

Admin access management panel for generator workspace

Python SDK expanded from 10 to 18 endpoints with SSE streaming support

TypeScript SDK gains listPromptSessions and exportPersona methods

Playground moved under the dashboard for consistent authenticated navigation

Dashboard layout polish — larger headings, better spacing, and clearer visual hierarchy across all views

Billing entitlement checks and subscription webhook handling made more resilient

Route handlers consolidated with shared helpers across all API boundaries

Sign-in and sign-up flows now support multi-step authentication callbacks

Python SDK correctly preserves HTTP status codes on non-JSON error responses

Deployment startup failures from module format mismatches resolved across all serverless functions

Population panels — define demographic distributions and generate representative persona cohorts at scale

Task execution engine — run freeform prompts, structured surveys, form fills, voting, and writing tasks across personas

Task templates with web access — personas can search the web and read pages during task execution for grounded responses

Batch run management — queue, monitor, and aggregate results across full population panels

Census-based population modeling — seed populations from real demographic distributions

Stronger request hardening with SSRF protections on all outbound web access

Dashboard views for population management, task configuration, and run monitoring

Persona creation from demographic seeds with automatic timeline memory generation

Full persona lifecycle management — create, update, export, and delete via API and dashboard

Timeline memory system with structured entries (facts, feelings, interpretations, emotions) and revision tracking

Timeline consistency checking to detect contradictions, chronology gaps, and sequence issues

Study sessions in three modes: synchronous, streaming (SSE), and async with webhook callbacks

Document upload and knowledge base ingestion for context-grounded persona responses

Consistency and safety controls — contradiction detection, blocked terms, PII filtering, and guardrails

TypeScript SDK — npm install @person.run/sdk

Python SDK (beta) — pip install person-sdk

Batch panel runs for high-volume survey and poll execution

Team management with role-based access control (Owner, Admin, Editor, Viewer)

Audit log with filtering by action, actor, and time range

API key management with scoped access

Subscription billing and usage tracking with per-tenant quotas

Persona headshot generation

Interactive study playground for testing persona responses

REST API with OpenAPI 3.1 specification

Documentation site with guides for quickstart, streaming, panel setup, guardrails, and versioning