> ## Documentation Index
> Fetch the complete documentation index at: https://docs.surfacd.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Changelog
> New features, improvements, and fixes in Surfacd, newest first.
## Members can now manage every part of a topic
Members could already create, rename and assign topics, but deleting one needed an Admin.
Members can now do all of it from **Manage Topics** or the Prompts page. Deleting a topic never
deletes prompts: any prompts still in it move to Uncategorized.
On the Prompts page, the selection toolbar now shows only the actions your role can use. If you
can assign tags but not create them, the tag picker lists your project's existing tags without
offering to create new ones. See [who can manage topics and tags](/docs/tags-organisation#who-can-manage-topics-and-tags)
for the full breakdown by role.
## Deactivate prompts and keep their history
Replacing a prompt used to mean deleting it, and deleting a prompt removed its history from
your reports. Select prompts on the Prompts tab and click **Deactivate** to stop collecting
them instead. Their topic, tags and past results stay in place, so you can see how a new
prompt performs against the one it replaced.
Deactivated prompts move to a new **Inactive** tab. From there you can reactivate a prompt to
start collecting again, or delete it to remove it and its history from your reports. Active prompts are now deactivated before
they can be deleted, so history is never removed by accident.
If you deactivate every prompt in a project, collection pauses but your reports stay
available. Collection resumes when you activate or add a prompt.
Through the API and MCP server, set a prompt's `status` to `inactive` to deactivate it. The same
rule applies there too: only inactive prompts can be deleted.
## Reputation dates now ignore deleted prompts
When you delete a prompt, its responses drop out of your reports. Reputation's first-seen
and last-seen dates now follow the same rule, on the Narratives and Themes lists, on each
detail page and through the API. The **Latest** sort on Narratives uses the same dates.
Related narratives on a narrative's detail page are also ranked using only responses from
prompts you still track.
## Introducing Reputation
Visibility tells you whether you appear in an AI answer. Reputation tells you what that
answer means. You can win the mention and still lose the customer if ChatGPT, Gemini or
Perplexity describe you poorly, and until now there was no way to see that at scale.
Reputation analyses the responses Surfacd already collects and surfaces the recurring
stories AI platforms tell about your brand. Narratives such as "praised for fast support"
or "criticised for pricing" are grouped into themes, scored for sentiment and tracked over
time, so you can see which stories are growing, which are fading, and which platforms are
driving them. Competitor views show how the same themes play out for the brands you are
measured against.
Reputation is included on every plan. Open a project, click **Reputation** in the sidebar
and enable it. Analysis starts with your next collection and there is nothing extra to
configure. It is in beta and free to use while it is.
## Faster Source Analysis
Source Analysis loads faster on projects with large source histories, including when opening a website's cited pages or exporting results.
## Download Rankings and Answers as a PDF or Markdown report
Until now the only way to take a report out of the app was a spreadsheet of the underlying rows. Rankings and Answers now also download as a formatted report: the charts, tables and every active filter laid out as a document you can share.
Choose **Export as PDF** for a branded, print-ready version, or **Export as Markdown** to paste the same report into an AI assistant and ask it to analyse or summarise your results. Each report includes a link that reopens the live page with the same filters applied.
Both live in the **Export** menu on the page, alongside the existing spreadsheet download.
## Manage your topics in one place
Topics group your prompts by theme, but until now you could only shape them while
editing prompts. The new Manage Topics page, under Manage Tags in the project sidebar,
shows every topic with its description, prompt count and creation date.
From here you can create a topic ahead of assigning prompts, rename one, or update its
description. Click a topic's prompt count to jump straight to its prompts.
Deleting is safe by design: you are asked to confirm first, and prompts are never
deleted with their topic. Any prompts still assigned move to Uncategorized, and the
confirmation tells you how many prompts that affects.
## See the searches AI ran before it answered
Open any answer from Mentions, the Dashboard or Reputation and a new Search Queries
section sits beside the sources. It shows every web search the platform issued while
building that answer, in the order it ran them, including operators such as `site:`
and quoted phrases.
Search-backed platforms often run several queries for one prompt. Seeing them tells you
which questions the AI actually asked, which competitor sites it went looking at, and
which regulations or terms it considered essential. Pages that rank for those searches
are what the AI reads before it answers.
Platforms that answer from their own knowledge run no searches, so the section stays
hidden for them. The queries are also returned by the response detail API endpoint and
the MCP response tool.
## In-app explanations for every metric
Every report now explains itself. Hover a column heading like Visibility Score, Avg.
Position or Citations, or the info icon beside a card, and a small panel tells you what
the number means and how to read it — with a concrete example, such as "If AI gave 100
answers and 40 mentioned the brand, visibility is 40%".
Bigger topics come with a short walkthrough video embedded right in the panel: creating
a project, adding prompts, organising with topics, comparing visibility over time,
tracking competitors, browsing answers, and the Analyse matrix. Where a docs page covers
the topic in depth, the panel links straight to it.
You will find these across the Dashboard, Rankings, Answers, Sources, Analyse,
Reputation, Prompts, project settings and team settings.
Reach out to [support@surfacd.com](mailto:support@surfacd.com) if you have any questions or feedback.
## Monitor your brand in Claude
Claude is available as an early-access platform. It is not included in plans by
default, so contact support to enable it for your team. Surfacd queries Claude with web
search switched on, so responses reflect both the model's knowledge and the pages it
cites.
Claude's data is significantly more expensive to collect than most platforms, so each
prompt check costs 30 credits rather than 1. The platform selector now shows a credit
badge next to any platform that costs more than the standard rate, and the credit estimate
for a project always reflects the real total.
Claude is monitored through the model provider's API rather than the consumer app, and
carries an **API** badge in the selector to make that clear.
## Filter API and MCP reports by tag or topic name
Every reporting and reputation endpoint in the public API, and every report tool in the
MCP server, now shares the same four filters: platforms, countries, tags and topics.
Tags and topics accept either a UUID or a name. Names are matched case-insensitively
within the project. Topics also accept `no_topic` to select prompts that have not been
assigned a topic.
Unknown values are rejected with a clear validation message rather than silently
ignored, so a typo in a tag name will not quietly return unfiltered data.
## Grok now monitored through its API
Grok is now monitored through the model provider's API, with an **API** badge in the
platform selector.
Because the way Grok is collected has changed, it has been deselected on every project
that had it enabled. Any credits reserved for Grok on scheduled projects have been
returned to your team. To keep monitoring Grok, open your project's platform settings and
select it again.
## Colour-code your tags
Pick a colour when you create a tag, whether that is from **Manage Tags** or by typing a
new tag name while adding prompts. Ten colours are available: grey, blue, cyan, purple,
pink, red, orange, yellow, green and teal.
Colours appear wherever tags do: on prompt rows, in tag dropdowns, in the bulk-tagging
tools and in report filter controls. Existing tags and tags created by CSV import show as
grey until you choose a colour from **Manage Tags**.
The public API returns a `color` field on every tag and accepts one when creating or
updating a tag.
## Save and reopen your favourite report views
Set up the filters you use most, then open the chevron next to the page name in the
breadcrumb and choose **Save current view**. The dialog shows exactly which filters will
be saved before you confirm.
Saved views are listed in the same menu. Views for the current page appear first, and
you can search across every page's views, rename them, or delete them.
Dates can be saved three ways: leave them out so the view uses whatever range you have
active, pin the exact dates, or choose a rolling range such as **Last 28 days** or
**Previous month** that resolves to fresh dates each time you apply it.
## Report filters and views now live in the link
Filters, selected tabs, sort orders, search terms and comparison sets on the reporting
pages are now stored in the page address. Copy the link and whoever opens it sees the same
view you do.
Only the settings that differ from the page default are added, so links stay short.
The browser back and forward buttons step through your filter changes, and reloading a
page no longer resets it.
This applies across Dashboard, Rankings, Mentions, Analyse, Sources, and the Reputation
pages.
## Sentiment score alongside visibility in Rankings and the Dashboard
Visibility tells you how often a brand appears. Sentiment tells you how it is described
when it does. Both now sit side by side.
In Brand Rankings, each brand has a **Sentiment Score** column with a colour band and a
trend against the previous period. The Rankings CSV export includes the same values.
On the Dashboard, the Industry Ranking card shows each brand's sentiment score, and the
visibility chart has a **Visibility / Sentiment** toggle so you can chart either metric
for every brand in the ranking. A period with no mentions has no sentiment to report, so
the line bridges the gap rather than dropping to zero.
The public API and MCP tools return the same `sentiment_score` and
`previous_sentiment_score` fields on brand rankings and visibility-over-time responses.
## Follow what's new in Surfacd
Improvements used to land quietly. Now every change that affects what you can see or
do in Surfacd gets written up here as it goes live.
Each entry explains what changed and why it matters, so you can tell at a glance
whether something is worth a second look.
Prefer updates to come to you? Subscribe to the RSS feed from the icon at the top of
the changelog page and pipe it into Slack, email, or your reader of choice.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.