add_reference_video
ChatGPTAttach a reference video to a UGC playbook. url is a TikTok or Instagram post link, a direct video link, or a URL from suggest_reference_videos. The first reference video becomes the playbook cover and thumbnail in the creator app; an external link is downloaded and rehosted by the video worker within a few minutes. preview_add_reference_video takes the same arguments and returns the change without applying it.
configure_campaign
ChatGPTSet up or change an organic campaign in one call: the content format (and with it who can post), where, the budget, and the Noise Unlimited options. During onboarding it applies to the Starter Campaign (get_onboarding_status names it). creator_pool follows playbook_format unless passed (slideshow: 'all' = All Creators, any creator from any account; UGC: 'gold_plus' = Gold+ Creators, Gold-level and above posting from an account that fits the brand's categories). Budget: daily (at least $50 a day; Noise sets the rate and aims to spend around it, a target not a hard cap) or monthly (Noise Unlimited: the brand's own rate per 1,000 views and an optional monthly limit). The preview lists missing answers, every defaulted value, what needs Noise Unlimited, and warnings; saving is refused until nothing is missing. Noise Unlimited settings save only once the brand has subscribed in the portal (the preview gives the link). A campaign keeps its budget type once set. preview_configure_campaign takes the same arguments and returns the change without applying it.
create_campaign
ChatGPTCreate a new campaign. kind 'organic': creators post about the brand on their own social accounts; the same settings as configure_campaign plus a hashtag (the hashtag creators add so posts count). kind 'ads': creators send the brand videos to run as ads on Meta or TikTok instead of posting; Noise takes a 5% fee on the brand's ad spend, plus optional submission and selection bonuses. An ads campaign is UGC with Gold+ Creators and has no platforms, dedicated account or budget. It is a Noise Unlimited campaign ($99/month, credited to the brand's balance straight away to spend on creators): it can be created now, and activating it needs the subscription, which the brand completes in the portal. Name is optional: without one, the campaign gets the standard name ('Starter Campaign' for organic, 'Ad Submissions' for ads, numbered if taken). Playbooks for an ads campaign are UGC, created on its campaign_id. The preview lists the defaults. preview_create_campaign takes the same arguments and returns the change without applying it.
create_slideshow_playbook
ChatGPTCreate an image-slideshow playbook in a chosen format. Creators assemble these from images the brand supplies, with no filming: the fastest way to get a brand's first playbook live. The format id comes from list_playbook_formats. It is created inactive and empty: slides are added with update_playbook, applied with publish_playbook, and the playbook goes live with set_playbook_status. preview_create_slideshow_playbook takes the same arguments and returns the change without applying it.
create_ugc_playbook
ChatGPTCreate a UGC playbook in a chosen format. The creator films these from a script of scenes, so they take more setup than a slideshow and need a reference video (the example the creator studies) before they can be published; suggest_reference_videos lists candidates and add_reference_video attaches one. The format id comes from list_playbook_formats. It is created inactive and empty: scenes are added with update_playbook, then the reference video, then publish_playbook applies them and set_playbook_status makes it live. preview_create_ugc_playbook takes the same arguments and returns the change without applying it.
explain_billing
ChatGPTHow Noise billing works and, for a linked brand, its own billing state. Posting campaigns bill for performance: the brand pays for views on creator posts, nothing upfront. Ads-only campaigns take a 5% fee on ad spend plus optional creator bonuses. Noise Unlimited is optional ($99/month, credited to the brand to spend on creators). Returns the charge cadence (accrued views are checked every 5 minutes and the card is charged when they reach the brand's threshold: $10, $25 or $50 at onboarding, higher tiers after $500 in charges), what a failed charge does, the brand's threshold, card state and last charge, and who to contact: the brand's Noise rep, or support for callers with no organization. Works with or without a Noise organization.
explain_flywheel
ChatGPTThe Noise flywheel: brief creators with a playbook, let creators find the campaign, find the winning posts, amplify them on Noise or as Meta and TikTok ads, repeat. Returns the tools for each step, what makes a post a winner, and a prompt the brand can schedule in this client to run the loop several times a day. Works with or without a Noise organization.
explain_noise_effectiveness
ChatGPTHow Noise shows up in a brand's numbers, and the brand's own signals. Explains the effect (an organic install uptick; App Store search and Apple Search Ads lift, because viewers search for the app after seeing creator posts; installs rising first in the countries the posts ran in), then reports over a window: installs on quiet versus Noise-active days, installs above the no-Noise "sea level" attributed to Noise and the resulting cost per attributed install, whether install movement follows view movement a few days later, which countries and App Store install sources rose during Noise activity, and the store listing funnel. Ends with the limits of the data. Defaults to the trailing 180 days ending 8 days ago (store and estimate data arrive late). Brands without a connected app store get the explanation and how to connect one.
find_ideal_creators
ChatGPTFind real high-performing creator posts in a brand's niche from campaigns on Noise, showing what creators in that category make and which formats earn views. Takes the brand description and a short niche label; competitor names add posts made for those competitors (flagged competitor_example: true) when they run on Noise. Returns creator handle, platform, post URL, views and a thumbnail, one post per creator, plus the Noise categories matched. The brands behind the posts are not named. Works with or without a Noise organization.
find_us_markets
ChatGPTLook up US metro areas by name for creator targeting, which is Noise Unlimited. Takes the places the brand named (cities, metros or nicknames such as 'Austin', 'the Bay Area', 'NYC'); each comes back with its matching metros and their ids, best match first. The ids go to configure_campaign or create_campaign as targeting.us_market_ids. A place with no match comes back empty. Only the United States is split into metros; other countries are targeted whole with countries.
get_audit_log
ChatGPTRecent change history for the brand, newest first: who changed what and when (campaign activations and pauses, budget, target and rate changes, playbook edits). A performance shift in get_daily_report can be matched to a configuration change here.
get_campaign_details
ChatGPTCurrent configuration of the brand's campaigns (not performance history, which is in get_daily_report): hashtag, type, dynamic or static, active, daily target, budget (amount, spent, utilization), current CPM rates and the campaign's playbooks. campaign_id selects one campaign; without it, all campaigns.
get_daily_report
ChatGPTDaily performance metrics for the brand on Noise: spend (USD), views, posts, unique creators, CPM and estimated_installs (first-party App Store / Play data where the brand has connected a store, otherwise third-party SensorTower estimates; day grain only, for brands with an installs source). Grouped by UTC day; optionally broken out by campaign and/or playbook, or filtered to one. Dates are inclusive YYYY-MM-DD, up to 366 days; 90 days or less keeps the output compact. Returns period totals plus a per-row breakdown. Campaign and playbook ids come from list_campaigns and list_playbooks. Installs, countries and App Store search lift are in explain_noise_effectiveness, not in this report.
get_onboarding_status
ChatGPTWhere the brand stands on the way to its first live campaign, and the step that comes next (next). Stages: no_organization, needs_campaign_setup (the campaign is not set up yet; next lists the questions for each branch, in the portal wizard's order), needs_playbook, needs_slides, needs_activation (blockers lists what the brand clears in the portal first: a card, the Noise Unlimited subscription, an active playbook), awaiting_first_posts, live. It reports on the campaign being set up, named in campaign: the ads-only campaign if the brand chose ads, otherwise the Starter Campaign. Works with no organization (returns no_organization). help names the brand's Noise rep, or support.
get_playbook_details
ChatGPTA playbook's full content. Returns draft, the editable document (name, hashtags, descriptions, and slides with their instructions, on-screen text options and images), plus version, which update_playbook requires so an edit built on a stale read is rejected instead of silently overwriting a concurrent one. editable: false means this playbook's format cannot be edited here, with the reason in not_editable_reason. The legacy top-level fields (content_template_slides and others) are the published playbook, kept for older clients. Generated image assets are omitted.
ideate_playbook
ChatGPTGuidance for designing creator content for a brand: the content and hook principles Noise content follows, what makes a good Noise playbook, the questions that shape one, the structure of the two playbook kinds (slideshow: 5 to 7 image slides in a listicle format with caption options and how to source images; video/UGC: a Hook and a Demo scene, each with instructions for the creator), and the sequence of tool calls that builds one once an organization and campaign exist. The applied results carry a portal_url for the user. Works with or without a Noise organization.
list_ad_destinations
ChatGPTWhere a paid ad can be built: per connected provider (Meta, TikTok), the campaigns and their active or paused ad sets (Meta) or ad groups (TikTok), from Noise's mirror of the brand's ad account. Each ad set carries delivering: true (spending now) or false (paused by itself or by its campaign); delivering ones list first. A paused ad set is a valid destination: an ad built there starts delivering when the brand switches the ad set on. Per provider, syncPending: true means the account was connected moments ago and the mirror lands within about 5 minutes; otherwise the mirror refreshes every 4 hours, so a brand-new ad set appears after the next sync. An empty list with syncPending false means the account has no active or paused ad set.
list_ads
ChatGPTThe brand's paid ads on Noise: every post pushed to Meta or TikTok (a 'placement'), newest first. Per row: ad_status (queued or processing = the creative is still being delivered; available = it is in the ad library with no ad yet; active = a matched ad is delivering; paused = matched ads, none delivering; not_created = it is in the ad library but building its ad failed, with the reason in last_error, usually a permission in the brand's own Meta setup; error = the push failed, see last_error; last_error is null once an ad matched the placement), provider, post URL, destination ad set, matched provider ad ids, ads (one entry per matched ad: adId, adsetId, adsetName, adName (the ad's name in Ads Manager), status, effectiveStatus; a creative can run in several ad sets and the row's totals sum across them), fees paid to Noise (USD, the license fee), provider ad spend tracked (USD), impressions, clicks, CTR, primary action and CPA. Also lists which ad providers are connected. Paid ads are a separate channel from the organic Noise effect, which explain_noise_effectiveness covers.
list_campaigns
ChatGPTList the brand's campaigns (id, name, active). The id is get_daily_report's campaign_id filter.
list_playbook_formats
ChatGPTThe playbook formats a brand can create, with what each is for, and the format id that create_slideshow_playbook and create_ugc_playbook take. Two families: SLIDESHOW playbooks are assembled from images the brand supplies and need no filming, the fastest route to something live; UGC playbooks are filmed by the creator from a script.
list_playbooks
ChatGPTList the brand's playbooks (id, name, active, campaign_id: the campaign the playbook belongs to, null when unassigned), including id 0 "Custom Posts". The id is get_daily_report's playbook_id filter.
list_posts
ChatGPTPer-post drilldown for the brand on Noise: posts created in the date range, each with creator handle, title (the post's caption text), posted_at (ISO 8601 UTC), platform, range_views (views gained inside the range), lifetime stats (views, likes, comments, bookmarks, shares), engagement_rate (lifetime engagements ÷ views, 0..1), playbook and campaign attribution, spend (USD brand cost, lifetime), range_spend (the same basis, accrued inside the range) and the post URL. Sorted and paginated on the server; campaign_id, playbook_id and a modest page_size keep the output compact. Filter ids come from list_campaigns and list_playbooks.
noise_facts
ChatGPTNoise's confirmed prices, fees, limits, billing rules and policies for brands, from the Noise facts registry, with the registry version. Facts not settled yet are listed apart and marked: they have no known value, and the brand's Noise rep answers them. An optional topic narrows the list (for example billing or Unlimited). Works with or without a Noise organization.
noise_glossary
ChatGPTNoise terminology: brand, creator (tiers), campaign (basic, niche, Dedicated; static vs dynamic), playbook, slideshow, UGC, slide, caption, description, hashtag, post, views, CPM, rate, budget, daily target, dynamic pricing, billing threshold, billing, ad, Noise effect, brand rep, connection code, audit log. Takes the words the user used, in any wording (template, influencer, carousel, daily budget), and returns the Noise term, a one-line definition, the detail and the mapping. Without terms, returns the full glossary. Works with or without a Noise organization.
portal_navigate
ChatGPTA page of the Noise brand portal (FU-1441). Takes the destination; campaign_edit and playbook_edit need the id. Inside the portal the page opens beside the chat; from any other client the result carries the URL to give the user. Read-only: nothing changes.
preview_add_reference_video
ChatGPTPreview of add_reference_video: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_configure_campaign
ChatGPTPreview of configure_campaign: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_create_campaign
ChatGPTPreview of create_campaign: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_create_slideshow_playbook
ChatGPTPreview of create_slideshow_playbook: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_create_ugc_playbook
ChatGPTPreview of create_ugc_playbook: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_publish_playbook
ChatGPTPreview of publish_playbook: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_remove_reference_video
ChatGPTPreview of remove_reference_video: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_set_campaign_status
ChatGPTPreview of set_campaign_status: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_set_daily_target
ChatGPTPreview of set_daily_target: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_set_playbook_status
ChatGPTPreview of set_playbook_status: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_update_campaign_budget
ChatGPTPreview of update_campaign_budget: takes the same arguments and returns what it would change, with its consequences, without changing anything.
preview_update_campaign_rates
ChatGPTPreview of update_campaign_rates: takes the same arguments and returns what it would change, with its consequences, without changing anything.
publish_playbook
ChatGPTApply a playbook's saved draft to the playbook, in one transaction. Publishing is not activating: creators see a playbook only while it is active (and its campaign is active). On an ACTIVE playbook, publishing updates what creators see right away; on an INACTIVE one (every new playbook), creators see nothing until set_playbook_status activates it. Publishing does not activate, so an edit cannot silently un-pause something the brand paused on purpose. preview_publish_playbook takes the same arguments and returns the change without applying it.
remove_reference_video
ChatGPTTake a reference video off a UGC playbook. reference_id comes from get_playbook_details or suggest_reference_videos (the current list). Removing the first one makes the next the cover; removing the last one blocks publishing again. preview_remove_reference_video takes the same arguments and returns the change without applying it.
search_playbook_images
ChatGPTImages already available to the brand for slideshow slides: free, already moderated and tagged, and including the brand's own uploads. Semantic search, so a description ("woman drinking iced coffee outdoors") works better than guessed keywords. Without query, returns what the brand has that matches its categories. Each result's url and id go into update_playbook's slide images as {url, library_id}.
search_posts
ChatGPTThe brand's approved posts from the last 7 days (days widens the window), most lifetime views first, as candidates to run as paid ads. Leaves out posts already pushed to an ad account, posts that Noise content review holds or rejected, and posts by creators under review. Views are reported as recorded: the tool does not judge whether they are genuine, and low engagement is not treated as a red flag, because engagement can be missing on a post for reasons unrelated to view quality. Returns 5 posts unless limit is set. Each hit carries the post id, views, likes, comments, bookmarks, shares, spend, caption (title), creator handle, playbook, campaign and post URL.
set_campaign_status
ChatGPTActivate or pause a campaign. Activating needs a payment method on file (the portal Billing page) and, for a Noise Unlimited campaign, the subscription; get_onboarding_status and the preview list what is missing. Activating auto-enrolls existing brand members and, for dynamic campaigns, rolls the budget period; pausing also pauses its playbooks. A new campaign goes live through set_playbook_status, which turns its campaign on too. The preview lists the consequences. preview_set_campaign_status takes the same arguments and returns the change without applying it.
set_daily_target
ChatGPTSet the daily spend target (USD/day) for a dynamic campaign. The pricing engine paces to it; the per-post cap recomputes and the budget period rolls forward. Guardrailed by a per-change percentage cap. preview_set_daily_target takes the same arguments and returns the change without applying it.
set_playbook_status
ChatGPTActivate or pause a playbook. Activating shows it to creators. It needs a payment method on file (the portal Billing page) and, for a Noise Unlimited campaign, the subscription; the preview lists what is missing. Pausing the last active playbook also pauses the campaign. preview_set_playbook_status takes the same arguments and returns the change without applying it.
suggest_reference_videos
ChatGPTReference videos a UGC playbook could reuse, when it has none or needs another. Returns the playbook's current reference videos, then candidates from the Noise template library in the same format and from the brand's other UGC playbooks, each with the playbook it comes from. A TikTok or Instagram post link (for example a winning post from search_posts) or any direct video link also works; the video worker rehosts it. add_reference_video attaches one.
update_campaign_budget
ChatGPTSet a static campaign's budget amount (USD). Dynamic campaigns have no budget row and are paced by their daily target (set_daily_target); get_campaign_details reports is_dynamic. A campaign's first budget is set with configure_campaign. Guardrailed: a change cannot exceed the per-change percentage cap or the absolute ceiling for AI changes. The preview shows current versus proposed. preview_update_campaign_budget takes the same arguments and returns the change without applying it.
update_campaign_rates
ChatGPTReplace a static campaign's CPM rates (USD per 1,000 views). Dynamic campaigns are paced by their daily target instead (set_daily_target). CPMs are hard-capped per campaign type. Only 'views' rates change: creator splits, per-post caps and daily targets carry forward, and per-post bonus rates are left alone. preview_update_campaign_rates takes the same arguments and returns the change without applying it.
update_playbook
ChatGPTEdit a playbook by saving a DRAFT: nothing reaches creators until it is published and the playbook is active, so it saves without confirm. Requires the version from get_playbook_details, so an edit built on a stale read is rejected rather than silently overwriting a concurrent one. Every field is optional and an omitted field is left unchanged, except slides: when sent, it REPLACES the whole list, so it has to hold every slide to keep, in order, each with its key from the read. The brand sees the draft in the portal and can review it before publishing. publish_playbook applies it.
why_noise
ChatGPTHow Noise works and whether it fits a brand: creators post about the brand on their own TikTok and Instagram and the brand pays per 1,000 views; why many creators beat a few (fast creative testing, native distribution, organic install lift, winners pushed to Meta and TikTok ads at no upfront creative cost); questions that make the case specific to the brand; and which tool comes next. Works with or without a Noise organization.