What's new?

Blog editors can now upload, describe and place self-hosted images, videos and audio in Markdown, with resumable transfers, bilingual accessibility metadata and the same validated media on the public article.

The mobile journey now keeps public intent through sign-up, sign-in and recovery, then resumes the right product action from the dashboard with clearer navigation and touch-ready controls.

Harmate now applies one persistent appearance preference across the showcase, dashboard and respondent journeys while preserving readable controls, charts, documents and shared assets.

Questionnaire owners can now manage participant cohorts, monitor delivery and response states, send safe reminders and configure durable automatic follow-ups without leaving the attendee workspace.

Questionnaire review now replaces heuristic readiness with structural evidence and actionable corrections. Guided proposals persist through progressive conversations, show before-and-after content and always keep authors in control of confirmation.

Recruitment projects now connect questionnaire candidates to explicit campaigns, explain shortlist rankings with stable evidence, preserve revisioned needs and produce a shareable interview brief without mixing applicants across contexts.

Managers can now choose aggregate confidentiality before collection, review its consequences and propagate the policy through editor, campaign, simulation and restitution flows with fail-closed boundaries around sensitive outputs.

Operations projects can now define questionnaire-first population filters and declarative comparison rules, preview them safely and reuse the exact scope across the shared screen, PDF, CSV and optional assistant framing.

External consumers can now call versioned stateless text-insight endpoints with scoped organization keys, hashed idempotency, minimized audit records and explicit usage accounting without persisting source texts.

Response imports now provide governed provider connections, explicit credentials and entitlements, reviewable column routing, lifecycle reconciliation and visible operational states across supported spreadsheet, form and survey sources.

Real results, topic analysis, strategic reading and exports now use the same submitted corpus and freshness identity, disclose missing sections and avoid an aggregate score that would overstate what the evidence supports.

Organization administrators can now deliver localized invitations by email, preview their destination, resend or revoke them safely and follow a focused public acceptance flow without exposing invitation secrets in list payloads or navigation state.

Respondents can now attach required documents inline with validation, progress and retry states, while supervisors see private processing status from attendee details without turning stored files into public assets.

Real mirror questionnaires now open a dedicated matching cockpit with authoritative populations, pair labels, scores, coverage and source revision states instead of hiding the result behind the legacy tester route.

Topic analysis now builds canonical macro concepts and evidence-supported subthemes, preserves the source concept tree through computation and caching and keeps weak or invalid splits from becoming authoritative categories.

Response imports now classify columns from their values instead of bilingual header dictionaries, support local multi-sheet Excel reading and expose a review step before question, identity and technical metadata routes are persisted.

Group names now exclude identity and non-positive signals, adapt to French or English, and stay consistent across results, exports and future publications. Supervisors can replace an automatic label with an audited official name.

Strategic analysis now names and counts non-overlapping segments, reframes questions when their wording drifts, and provides a justified questionnaire verdict. Every exposed claim is linked to its source question, evidence and confidence level in both the dashboard and complete report.

Topic analysis now selects an appropriate response regime for each question, preserves repeated values and local sub-themes, and exposes coverage and diagnostics so the interactive tree stays readable without flattening question-specific meaning.

Questionnaire managers can prepare a recipient list from Contacts or email addresses, create and copy a read-only presentation link, send it to up to 50 people, and keep presentation access separate from respondent collection.

Strategic reading now keeps its source and run freshness visible, links findings to the questions that support them, and focuses the corresponding analysis view so a decision can be checked against the underlying evidence.

Complete reports now provide clearer progress, a readable cover and navigation, stable PDF delivery, and consistent simulation evidence and persona references from the results workspace to the exported document.

Simulation setup now adapts its selection controls to the available population, keeps coverage and constraints visible, and improves bilingual search labels so managers can make a more informed choice before launching a run.

Questionnaire managers can configure attention thresholds for drop-off, find operational alerts in the inbox, and manage notification and delivery preferences separately instead of losing important signals in one undifferentiated panel.

The questionnaire quality review now separates its axes and signals more clearly, with calibrated scoring that helps managers act on wording, structure, burden, and duplication issues before sending.

Pricing now makes simulation entitlements, credit usage, and the live Stripe checkout handoff more explicit, so plan selection and account billing state stay aligned with the value of a simulation run.

A new public use case explains how exploratory interviews can turn open conversations into structured, decision-ready learning while keeping the questionnaire journey and next actions visible.

Simulation workflows now connect preparation, exact grouping results, participant-level reading, progress recovery, and guided analysis so a manager can follow a run from generation to interpretation without losing its context.

Results now support persistent grouping controls, explicit must-link criteria, bounded manual adjustments, and a reviewable history so managers can understand how a grouping evolved before using it in analysis or simulation.

Questionnaire managers can preview one participant’s exact view, publish a versioned group assignment with messages, next steps, resources or dates, schedule its availability, withdraw it and retry minimal notifications. Participants find the official follow-up in My Responses, with strict audience and minor-protection safeguards.

Questionnaire managers can set an explicit respondent access policy, preview the exact access flow and confirm it before sending. Adult-only questionnaires verify the respondent's declaration before any identity or response flow begins.

The language analysis now separates five structural dimensions, the strength of the analysis base, and comparable human benchmarks with explicit provenance and limitations.

Strategic analysis now contextualizes sentiment according to the questionnaire, consolidates priorities and recommendations, and presents themes, questions, and score components in one clearer reading flow.

Complete PDF reports now prioritize interpreted findings, comparable language measures, strategic signals, and readable charts while omitting sections that do not provide useful evidence.

The results workspace now lets you resize the analysis panel, preserve the active source context, and move between reading and conversation with less visual friction.

Topic clustering now preserves question context, builds more meaningful hierarchies, and reduces misleading groupings caused by mixing responses from different prompts.

Product surfaces now explain credit usage more consistently across analyses and one-shot actions, making costs easier to anticipate before execution.

Simulation results now keep progress and visible runs available throughout execution, with more reliable navigation between persona selection and results.

Persona selection now localizes search, highlights coverage gaps, and provides clearer diagnostics for choosing a representative simulation panel.

The dashboard home now surfaces shared operational context alongside recent work so collaborators can identify the right dossier, questionnaire, and next action faster.

Quick Start, the questionnaire assistant, and job-description generation now reuse the preflight score to revise weak drafts once and block payloads that still fall below the readiness threshold.

The pricing page now separates collected responses, analysis credits, subscriptions, and one-shot packs with a clearer commercial structure.

A dossier owner can now send an external link, let a contact request access, and approve or reject the request when email matching is not automatic.

The simple editor now shows a readiness score, useful blockers, and quick actions to improve clarity, measurement quality, and analysis potential before launch.

The website now explains the full Decision Files workflow with evidence, hypotheses, test questions, and decision-oriented reading in a dedicated public surface.

Hypotheses no longer depend on a single primary questionnaire: a dossier can attach several questionnaires and convert a test question to the intended target.

The guided framing flow was simplified to chain decisions more clearly, surface the next useful action, and keep dossier preparation easier to read.

New accounts, active work recovery, and discovery prompts no longer render the same way: the dashboard home recomposes itself around your questionnaires, dossiers, and active usage.

After the first follow-up, the framing grid stays editable: you can correct the slots manually and continue the exchange before building the draft.

Operational dossiers can now structure hypotheses, attach evidence, follow test questions, and expose a clearer level of decision readiness.

Harmate now has a dedicated blog page to publish practical guidance, product notes, and strategic content around questionnaire design and group formation.

Harmate now supports JSON data import and surfaces more context about the imported content.

Volumetry now drills down to the group level to make it easier to read what brings subgroups together or sets them apart.

The What's New page centralizes the most visible product additions and fixes.