> ## Documentation Index
> Fetch the complete documentation index at: https://docs.onsomble.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Product changelog

> Recent product updates across Onsomble. Scroll the timeline, click any entry for the full story.

A running log of notable product changes. Bigger releases link to a full writeup; smaller ones are summarised in place. This page is the curated, customer-facing version, not a commit log.

<Update label="13 Aug 26" description="References, rebuilt around what to do next" tags={["Results"]}>
  [References](/results/references) now opens with the questions that matter
  most: who's carrying your market's story, where you can win more of it, and
  what other sites are saying about your brand.

  **Where your citations come from** shows the split between your own sites,
  sites you could earn a place on, and your competitors' sites, plus how much
  of the whole picture rests on just five websites, all tracked against your
  last scan.

  **Top opportunities and Since last scan** point you at the independent sites
  already shaping AI's answers, ranked by how much they matter, and show which
  sites are new, rising, or falling since last time.

  **What sources say** pulls out specific statements other sites make about
  your brand, with the ones working against you first, so you know exactly what
  to address.

  **Brand coverage** replaces the old Brand Visibility card. It shows how often
  AI actually names your brand when it cites a site about your market, plus a
  list of silent sources that get cited a lot and never mention you.

  The source inventory also picked up new columns for **Influence**, **Claims
  backed**, **About you**, and **Competitors**, a **vs last scan** trend, and a
  **Community** role for forums and social platforms like Reddit.
</Update>

<Update label="10 Aug 26" description="Recommendations, reorganised around the evidence" tags={["Results"]}>
  Recommendations dropped its toolbar and filters: the list is short enough to
  scan without them, and priority now shows as a plain coloured word instead
  of an icon.

  Open a recommendation for the full picture: what to build, why it matters,
  which questions it targets, and where AI answers currently look for
  information, with category, effort, confidence, and impact all in one place.

  **A coverage chart for AI-engine gaps:** when one AI engine mentions your
  brand less often than the others for the same questions, the recommendation
  now shows exactly how each engine compares. Each targeted question also
  carries small badges showing which AI engines the recommendation is based on.
</Update>

<Update label="06 Aug 26" description="Site Settings, refreshed" tags={["Sites"]}>
  Site Settings has a cleaner layout, and the **Operating regions** picker now
  works properly. The market search used to be cut off inside the card, so you
  could not see or choose from the list. Selecting **Edit regions** now opens a
  searchable picker where you find and add markets, and your selected regions
  show as chips you can remove.

  The page is organised into clear sections (business profile, audience and
  markets, brand identifiers), each with its own save status so you can see at a
  glance whether a change saved. Editing a field and clicking away still saves it
  automatically, exactly as before.
</Update>

<Update label="06 Aug 26" description="Client Settings for agency accounts" tags={["For agencies"]}>
  Agency accounts can now edit a client's details and remove a client from a new
  **Client Settings** page, reachable from the bottom of the sidebar inside any
  client's workspace. Update the client's name, website, industry, and
  description, with each field saving as you finish it.

  **Deleting a client is now possible, and deliberate:** the Danger zone removes
  the client along with all of its Sites and their full scan history. The
  confirmation lists the Sites that will be deleted and asks you to type the
  client's name first, so an entire client and their history cannot be removed by
  accident.
</Update>

<Update label="31 Jul 26" description="Sentiment rebuilt on quoted evidence" tags={["Results"]}>
  Sentiment scores are now built from the actual statements AI models make
  about each brand. Onsomble extracts every claim a response makes — with the
  exact quote behind it — classifies the tone of each one, and averages them.
  Open any score and you can read the quotes that produced it.

  Scores now sit on a clearer **0 to 100 scale where 50 is neutral**: above 50
  reads favourable, below 50 unfavourable. Scores also spread out more —
  criticism that earlier scoring tended to flatten into a mildly positive
  average now shows up, so expect sentiment to move more honestly from scan to
  scan. Scans completed before this release used the previous −100 to +100
  scale, so comparisons across the changeover date aren't like-for-like.

  **Honest gaps instead of misleading zeros:** brands with no sentiment
  evidence in a scan now show a dash rather than a score, and Overall Health
  averages the metrics that exist instead of counting missing sentiment as
  zero. **Narrative got richer too:** the Narrative view now distinguishes
  trust signals (reviews, market-share proof, endorsements) and mixed
  statements ("powerful but pricey") instead of forcing everything into
  positive or negative.
</Update>

<Update label="23 Jul 26" description="Invite your team into one shared account" tags={["Account"]}>
  Settings > Team is live: invite colleagues by email and work together in one
  account — same sites, same scans, same history. Invitees accept from an
  emailed link, prove it's their address with the usual sign-in methods, and
  land straight in the account with no separate confirmation step.

  The page lists members alongside outstanding invitations, with resend and
  revoke for anything still pending. Seats count members plus outstanding
  invitations against your plan's allowance; expired invitations stop
  counting until resent.

  **Removal without loss:** removing someone ends their access immediately,
  but their scans and history stay in your account, attributed to them — and
  their login survives, so they can be re-invited or start an account of
  their own. **Billing stays with the Owner:** everyone invited joins as an
  Admin who can do everything except manage the plan.
</Update>

<Update label="23 Jul 26" description="Prompt generation now sizes itself from your strategy" tags={["Scans"]}>
  Generating prompts now always produces a complete question set sized by your
  confirmed strategy — the products, customer groups, and journey stages you
  selected — instead of a count derived from your plan's scan allowance.
  Setups that previously stopped with "We couldn't finish this step" during
  onboarding now generate every time, and plan limits apply where they belong:
  to how often and how widely your scans run.

  **Also improved:** after a plan change, **Generate prompts** in the scan
  expansion setup opens the same guided strategy wizard used during onboarding,
  with the choice to add the newly generated prompts to your existing library
  or replace it.
</Update>

<Update label="21 Jul 26" description="Ask questions about your Site with the Onsomble Assistant" tags={["Assistant", "Insights"]}>
  The **Onsomble Assistant** is now available across supported Site pages. Ask
  about current performance, competitors, prompts, models, regions,
  references, narrative, recommendations, and scan history, then continue with
  follow-up questions in the same conversation.

  The Assistant can also search the official Onsomble documentation for product
  help and link the pages it used. Activity disclosures show whether an answer
  checked Site analytics, read an underlying AI response, or searched the docs.
  Conversations are private to you within each Site and can be renamed or
  archived from conversation history.

  [Learn how to use the Assistant](/assistant/overview)
</Update>

<Update label="12 Jul 26" description="Connect Onsomble to your own tools" tags={["Platform"]}>
  Accounts with API access can now connect their own integrations to Onsomble
  through the **Onsomble API**. Create account- or client-scoped API keys in
  **Settings → Account**, then read Clients, Sites, Scans, and discoverability
  reports or trigger a Scan for an existing Site.

  [Get started with the API](/developers/quickstart)
</Update>

<Update label="09 Jul 26" description="Plan entitlements and smoother setup" tags={["Billing", "Platform"]}>
  Plans now work through **entitlements** you meet in context: creating sites and
  clients, expanding a scan, or picking a faster cadence offers an upgrade at
  the moment you hit a limit, instead of failing. Scan size is governed
  per-plan with a visible run budget while you configure prompts and models,
  and free allowances now differ between business and agency accounts to match
  how each actually uses the product.

  Site setup also became more resilient: a setup run that fails partway can
  be retried without re-entering anything.
</Update>

<Update label="05 Jul 26" description="The agency experience goes client-first" tags={["Agencies"]}>
  Agency accounts are now organised around **clients**, not a flat list of
  sites. Onboarding asks who you're setting up Onsomble for and branches
  accordingly. Agencies describe their own website, review an AI-researched
  agency profile, and land in a **client portfolio**. Each client gets a
  workspace holding their sites and activity, with an **Add Client** flow that
  researches the client's business from their website URL and pre-fills their
  profile for review.
</Update>

<Update label="26 Jun 26" description="Region Exploration, composite filters, and the prompt strategy wizard" tags={["Insights", "Scans"]}>
  Three additions to the discoverability toolkit.

  **Region Exploration**: when your prompts span more than one region, a new
  Insights view puts visibility on the map: drill from the globe into countries
  and cities, rank regions by visibility, citation quality, competitor gap, or
  composite score, and replay how it all changed across your scan history.

  **Composite dashboard filters**: every insights view now shares one filter
  bar that slices by model, prompt category, product, customer group, journey
  stage, tags, region, and time period, applied together.

  **Prompt strategy wizard**: generate a coherent prompt set from your business
  profile by choosing the products, personas, journey stages, and markets that
  matter, instead of writing prompts one at a time.
</Update>

<Update label="08 Jun 26" description="Read the answers, not just the numbers" tags={["Insights", "Platform"]}>
  A round of improvements to how you read and act on individual AI responses.

  **Prompt responses, readable**: the response view gained clear brand
  visibility cues, better navigation between responses, and honest empty states,
  so reading what a model actually said is as easy as scanning the metrics.

  **More regions**: scans can now cover a wider set of markets, laying the
  groundwork for region-level exploration.

  **Smarter upgrade moments**: after your first scan, the upgrade offer is
  personalised to what the scan actually found, instead of a generic pitch. The
  sign-in experience was redesigned along the way.
</Update>

<Update label="24 Apr 26" description="A new home for your results" tags={["Insights", "Platform"]}>
  The dashboard landing page was rebuilt around what changed since you last
  looked.

  **Activity feed with trend deltas**: scan completions and competitor movement
  across your account, with the numbers that moved.

  **Snapshot sparklines and insight statements**: headline metrics now carry
  their recent trend and a plain-language read of where you stand.

  **Scan runs, hands on**: a redesigned scan schedule card with a run-now
  control, and a scan-run detail panel that lets you regenerate analytics for a
  completed run.

  Under the hood, every surface moved onto the rebuilt visibility engine: the
  same metrics, computed one way, everywhere.
</Update>

<Update label="18 Apr 26" description="Documentation site launch" tags={["Platform"]}>
  The new Onsomble documentation site is live at
  [docs.onsomble.ai](https://docs.onsomble.ai), replacing the previous
  Nextra-based site. Full coverage of both the AI discoverability and
  Workflows pillars, dedicated sections for agencies and account
  administration, and educational concept pages on AI discoverability.

  [Read the full release →](/changelog/2026-04-18-docs-launch)
</Update>

<Update label="28 Mar 26" description="The insights suite" tags={["Insights"]}>
  The release that turned scan results into a full analysis surface, and the
  foundation the rest of this changelog builds on.

  **Brand Scorecard**: your core metrics tracked over time against every
  competitor, by brand and by metric.

  **Narrative Intelligence**: the attributes AI assistants associate with each
  brand: strengths, frictions, trust signals, and gaps.

  **References**: the sources behind the answers, with citation breakdowns per
  model and domain.

  **Per-prompt results and recommendations**: how each question performed, and
  prioritised actions generated from what the scans found.

  **Scan schedule and run history**: recurring scans with a full record of every
  run, plus **Discovered Competitors**: brands AI models keep mentioning that you
  aren't tracking yet.
</Update>

## How this changelog is organised

Every user-visible release gets an entry here. Each entry covers what changed and who it affects: business accounts, agency accounts, or a specific plan. Bigger releases link to a dedicated page with the detail and reasoning; smaller ones are fully described in the timeline.

Changes that require attention from existing users (migrations, deprecations, behaviour changes) are called out explicitly. Purely additive changes (new features, improved insights, better performance) are noted but don't need anything from you.

## Subscribing to updates

This page provides an RSS feed at [`docs.onsomble.ai/changelog/rss.xml`](https://docs.onsomble.ai/changelog/rss.xml). Drop the URL into your feed reader of choice, or wire it to Slack via a service like Zapier.