Pulse Briefing

Traditional media monitoring showed what people said about your brand. Pulse shows how AI systems understand, mention and recommend it.

Contact PRecious Exit demo
Access checking
Index Synced (Idle)
Live model request running Pulling fresh AI model data.

Briefing data

This open preview uses recorded sample results, so you can see the report style without connecting private data or running a new scan.

Showing recorded sample results.

Active brand BYD Automotive Demo
Monitored competitors

Tesla, Deepal, BMW, GAC Aion

Default market

ASEAN 6 · English, Chinese, Thai, Vietnamese, Indonesian, Malay, Filipino

Dashboard chart view Apply one view across all dashboard charts
Visibility +2.1%
64.0% AI visibility index

How often and how clearly the brand appears in relevant AI answers. Peer movement stays separate until peers are independently scanned.

Source/Citation Quality +1.5%
38.0% citation support

Whether answers can be backed by usable owned, earned or independent sources. Missing source evidence is unavailable, not zero.

Message Clarity +0.8%
82.0% tone and clarity signal

Whether AI descriptions are clear and usable, with tone tracked across positive, neutral and negative signals.

82% Pos 14% Neu 4% Neg

Change badges compare the current demo snapshot with the previous 30-day baseline.

Mode Demo Last checked Local seeded snapshot Models used Demo model set Search channel Separate Google Search-grounded view Score basis Comparable core AI engines only Scan issues None Source basis Shown URL check pending Method Two reads: readiness + sampled AI answer Scope and limit Check scope pending Prompt lineage Prompts should carry rationale, source type, date range and audience intent; unavailable observed demand is not guessed Read 1 Source readiness: owned, validation and independent evidence Read 2 Sampled AI answer: platform, prompt/context, market, date and sample size Boundary Citation/source evidence is not recommendation or endorsement

Peer/Competitor Context

Point-in-time context for the selected brand and approved peer set. This is internal evidence, not a public ranking.

80% 75% 70% 65% 60%

AI Model Summary

AI mentions, source support and recommendation status by configured model family

Authority and recommendation drill-down Open for corroboration, perception triangle and buyer-journey detail.
Authority intelligence

Authority Explains Recommendation

Visibility is a discovery signal; authority explains whether AI should recommend the brand.

Authority pending
Corroboration -- Not scored yet
Citation quality -- Source weighting pending
Prompt distribution -- Prompt groups pending
Own It --
Back It --
Spread It --
Recommendation intelligence

Will AI Recommend The Brand?

Checks whether the brand is only visible or survives shortlist and recommendation prompts.

Readiness pending
Mentioned -- Appears in AI output
Positive -- Described credibly
Shortlisted -- Clear shortlist inclusion
Recommended -- Explicit recommendation
Evidence, research and methodology Open for narrative ownership, research QA and the Pulse operating model.
Operating model

Discovery, Authority, Recommendation, Action

Pulse treats visibility as discovery, authority as the core interpretation, recommendation as the commercial outcome and action as the user workflow.

Authority analysis

Evidence, Gap And Ownership

Maps what supports the desired narrative, where the gap sits and who appears to own the category story.

Evidence Map

Narrative Gap

Narrative Ownership

Competitive Corroboration

Research depth

Evidence Pipeline

Supporting QA view for source gaps, prompt tests and the next research job.

Research queue pending
Observed source discovery

Sources AI search found

Not run

These are the URLs returned by the saved source search. They are shown as observations; Pulse does not require a review receipt to display them.

Observed AI answers · one question per engine

What AI engines returned

Checking saved web-grounding receipt…

Read each engine's saved answer excerpt and returned URLs as the AI observation. A source link may be answer-linked or only in the provider's result list. This separate check does not change the Core AI benchmark.

    Next research step source-validation

    Validate the strongest claims before turning them into client-facing recommendations.

    1 Measure 2 Benchmark 3 Interpret 4 Act

    What This Means

    Plain-language interpretation for PR, communications and marketing teams.

    Your brand is visible in some AI responses, but the strength of source support and category association determines whether that visibility can turn into recommendations.

    Visibility Risks

    Issues that can reduce AI discovery, recommendation quality or market confidence.

      Recommended Actions

      Practical PR, content and SXO work that can close visibility and authority gaps.

      Reporting

      Create a client-ready output

      Use the fast buttons for leadership outputs; open the builder only when you need a fuller report or appendix.

      Fast outputs

      Leadership-ready exports

      Build the standard executive brief or one-slide active-brand snapshot without leaving Pulse Briefing.

      Advanced

      Report Builder

      Pick the format only when the quick outputs are not enough.

      Open builder
      Contact PRecious

      Request a Pulse demo through PRecious

      For now, demo requests and commercial enquiries go through the PRecious contact page so the existing HubSpot workflow captures the lead.

      Open PRecious contact page

      No direct Pulse signup yet. Requests go through the PRecious contact page so the existing HubSpot workflow captures the lead.

      Research and reports

      Turn a business question into a clear, sourced report

      Start with the decision you need to make. Pulse builds a preview from saved findings, shows what supports it and keeps missing evidence visible. It does not run new searches here.

      Plan a new check or investigation

      Preview the scope first. This does not run engines or change existing reports.

      Visibility counts come from saved answers. Source support is reviewed separately.

      Preview the scope before saving. No engines will be run.

      Checking saved-plan availability…

      Saved evidence to include

      Choose the evidence set for this report. Pulse keeps missing evidence visible and does not run new searches here.

      Checking whether reports can be saved. Keep this tab open until a save is confirmed.

      Prospect research

      Prepare for a stronger prospect conversation

      Start with a prospect name or domain. Pulse checks what is visible, what is supported and what may deserve a conversation—then keeps external use behind review.

      Save a prospect report as an internal draft, then request review in Library. Opening it does not save or approve this snapshot.

      Save preserves the original report and sources. Review and external-use approval are separate.

      New prospect

      Start here

      Create the first BD view, then choose whether to improve the evidence, prepare a share-safe card or export reviewed drafts.

      Quick scan Fast first view for a call, repitch or warm account check. Full scan Wider prompt and source coverage when the account matters. Review path Staff card first; prospect card only after source and claim review.
      Setup brand, market, language and lens Scan public proof and sampled AI answer examples Review proof gaps and next BD move Export staff cards, graphics and PPTX drafts
      How to read the result Public proof and AI answer examples are shown separately.

      Use the first result to decide the next action: add evidence, ask a better buyer question, prepare a reviewed card or run a wider scan.

      Public proof what buyers can verify AI answer example what appeared in this run Next action improve, review or export
      BD readout workspace

      Turn the Snapshot into a conversation starter

      Internal BD View stays staff-only. Prospect Share View is a reviewed, plain-language extract after source, claim and owner review.

      Agency workspace

      Business Development Snapshots

      Proposal and prospecting outputs for PRecious teams. Keep these separate from normal client dashboards and client report-builder flows.

      Available now PRecious team prospecting Output One-slide AI discovery, authority and recommendation snapshot Future lane API/MCP for approved sales automation, agency and white-label partners
      BD pack access

      PRecious-only revenue enablement first

      Prospect packs support initial emails, follow-ups, proposal teasers and repitches. Use them for PRecious team preparation until approved partner access and shared approval history are ready.

      Ad hoc Brand, geography, industry, topics and spokespeople Guided assistant support Future approved sales tools can request the same Pulse guidance Partner access Agency and white-label use requires a separate BD user category and tenant boundary
      Pitch evidence packs

      Internal pitch handoffs

      Use these to prepare PRecious follow-ups: the conversation angle, supporting sources, points to verify and the next action before anything is shared outside the team.

      Create Pulse Briefing → Report Builder → Prospect Evidence Notes Save Save proposal notes from the report window Use Prep for emails, calls, proposals and source checks Check first Source links, wording and claim strength before external use

      Saved locally for the current browser. Use the notes for proposal preparation; shared archive and approval history come later.

      Action workflow

      Turn findings into communication actions

      Choose a finding, then turn it into earned media, thought leadership, content or search work. Every suggestion keeps its supporting evidence and review limits visible.

      Start here: Review the top three ideas, open one action brief, then assign its evidence or approval gap.

      1 Find gap 2 Choose audience 3 Build proof 4 Brief action

      Recommended actions

      PR, content and search actions based on the selected workspace, competitors and current findings

      Uses the active workspace profile and canonical competitor set from Settings.

      Evidence basis for action ideas

      Each idea shows the source basis, confidence and output format so findings can become an action plan, content brief, prompt pack item, source-gap task or campaign brief seed.

      Premium action workflow

      Turn priority gaps into fuller campaign briefs

      This is the fuller version of PR Action Ideas. Use it when a suggestion needs audience context, evidence needs, message architecture, execution steps and a downloadable working campaign brief.

      1 Select gap 2 Add context 3 Build brief 4 Download brief

      Campaign Brief Builder

      Expanded campaign drafts for selected visibility, authority and recommendation gaps

      Ready to use now

      Campaign Brief Builder creates fuller on-screen briefs and downloadable documents. Direct sending to Slack, email, HubSpot, Notion or a CMS is optional and can be connected later.

      Uses the active workspace profile and canonical competitor set from Settings.

      Working brief basis

      Campaign briefs are working outputs. Review scan evidence, source URLs, client proof points and limitations before treating a brief as publication-ready or client-approved.

      Premium measurement workflow

      Connect AI visibility to traffic, leads and outcomes

      Use this screen to plan attribution once GA4, HubSpot and source tracking are configured. It links AI citations and referral paths to sessions, contacts and conversion evidence.

      1 Track citation 2 Tag traffic 3 Match leads 4 Report impact
      AI Referral Traffic (GA4) Preview
      Pending sample sessions from citations
      GA4 setup required
      CRM Lead Conversions (HubSpot) Preview
      Pending sample form conversions
      HubSpot setup required
      Attribution Rate Preview
      Pending citation click-to-lead conversion
      Tracking plan preview

      GA4 Referrals by Day

      Preview trend for the GA4 tracking plan; not live reporting until GA4 and UTM rules are connected

      HubSpot Conversions by Day

      Preview trend for the CRM attribution plan; not live reporting until HubSpot matching is connected

      Individual setup required

      Impact Attribution is a Premium module. GA4, HubSpot, source attribution, UTM rules, consent settings, and reporting definitions need to be configured separately for each client before these numbers should be treated as live reporting.

      Evidence workflow

      Test one real question across answer engines

      See whether an answer mentions the brand, recommends it and cites useful sources. Each result stays separate so you can compare like with like.

      Start here: Pick a buyer question, choose one answer engine and run that question.

      1 Pick query 2 Check mention 3 Review sources 4 Define gap

      Choose how widely to compare

      Core uses the standard comparable engines. Expanded adds more engines. The selected question still runs through one chosen engine at a time, while web-search results remain separate.

      Showing local demo simulation.

      Question library

      Choose a common buyer or stakeholder question, or enter your own. Nothing runs until you select Check this question.

      Uses the active workspace profile and canonical competitor set from Settings.

      Browser Prompt Validation

      Separate user-experience lane for prompts tested in actual AI browser sessions.

      Not run

      Use this for observed ChatGPT, Gemini, Claude, Perplexity or agent/browser answers. API connector results stay separate.

      Demo uses four local answer styles; live modes use the AI models configured in Admin.

      Selected Answer Rank #1/10
      AI Brand Mentions 2
      AI Recommendation Recommended
      ChatGPT (GPT-5.2) Configured model style
      Sources & References

      Default: last 6 months for relevance plus source depth.

      Brand and competitor comparison

      Compare a brand in its market

      Check one brand, compare its peers, inspect the sources and turn supported gaps into actions.

      Start here: Enter a brand for a focused check, or run a new comparison using the saved workspace setup.

      Workspace guardrails can limit scan depth, languages and available connectors.

      Languages Lite measures English only; selected local languages are context until validation.
      Answer engines Choose any answer engine available to this workspace. Search-grounded discovery remains separate.
      Workflow

      Run, validate, act

      Discovery, visibility, peer validation and recommended actions happen in one flow.

      1 Discover market 2 Check visibility 3 Compare peers 4 Suggest actions

      No AI authority check yet

      Enter any brand or domain to generate a quick working view of AI mentions, search presence, category association and competitor context.

      Positioning workflow

      See what the brand is associated with

      Use this screen to understand which topics, proof points, problems and buying situations AI systems connect to the brand, then spot positioning gaps before they become recommendation gaps.

      1 Map topics 2 Check sentiment 3 Verify sources 4 Refine positioning
      Evidence source

      Category Association Mode

      Choose the AI view, brand focus and peer set for the association map. Search-grounded discovery is supporting context, not a weighted association score.

      Showing local demo associations.

      Active profile Selected brand

      The canonical competitor set saved in Settings appears here.

      Category & Topic Associations

      Topics, phrases, proof points and attributes linked to the brand in AI-generated answers

      Association Sentiment Signal Strength

      Category Association Map

      Hover over nodes to see association strength, context and source links

      Advisory workflow

      Turn findings into client-ready guidance

      Use PR Advisory Chat to interpret the visibility checks, connect findings to PRecious services and draft practical recommendations for PR strategy, earned media, content, SXO and reputation work.

      1 Ask question 2 Ground in evidence 3 Link service 4 Draft advice
      Advisory source

      PR Advisory Mode

      Choose whether the advisor uses the local demo script, a comparable Core AI view, or a broader Deep AI view. Search-grounded findings are labelled as discovery context.

      Local rule-based copilot.

      PR Advisory Chat
      Website-grounded local advisor · no live AI model connected
      Workspace settings

      Workspace setup

      Set up this workspace: brand, audiences, markets, languages, competitors, sources and report preferences. Your role determines which fields you can change. Platform-wide controls stay in Admin.

      Workspace: checking the current selection…

      How Pulse works together

      Set the context once. Use it throughout Pulse.

      Your saved setup follows the selected workspace. Every finding keeps its question, engine, source and date so you can see what supports it and what is still missing.

      1 Settings configures brand, market, languages, competitors and sources 2 Briefing, Market Lens, AI Query Check and Category Association measure 3 PR Action Ideas and PR Advisory interpret evidence into bounded actions 4 Reports and exports combine the current workspace findings and status 5 Admin governs access, plans, engines, usage, storage and diagnostics

      Workspace profile

      The brand and audience context used when you explore answers, compare findings and build reports

      What Settings controls

      Normal workspace setup fields. Operator-only platform controls are not mixed into this form.

      Workspace access and plan

      Read-only summary of the current role, package and live-model access for this workspace

      Usage this month Update to see this workspace's live usage status. Saved results stay available if live work is paused.

      What your platform administrator manages

      Shared engines, platform health, access policies and plans across workspaces. These do not change when you edit this profile.

      Platform administration

      Manage Pulse across workspaces

      Review platform health, answer-engine availability, access and service limits. Use Settings when you need to change one workspace's brand, markets, languages, competitors or report preferences.

      Start here: Check the topline status, then open only the operational section that needs attention.

      Platform-wide controls; workspace-specific actions must identify their selected target.

      How Pulse works together

      Platform controls and workspace findings stay separate

      Settings supplies each workspace's context. Pulse creates scoped findings and actions from it, while Admin manages access, availability and limits without changing saved evidence.

      1 Settings configures brand, market, languages, competitors and sources 2 Briefing, Market Lens, AI Query Check and Category Association measure 3 PR Action Ideas and PR Advisory interpret evidence into bounded actions 4 Reports and exports combine the current workspace findings and status 5 Admin governs access, plans, engines, usage, storage and diagnostics
      Topline

      Platform status and actions

      Start here. Open an operational section only when you need to inspect or change its detailed controls.

      Loading Admin status Checking the current signed session, workspace, engines and review queue.

      Workspace governance

      Current data boundary, durable-store readiness and what belongs in Settings versus Admin

      Client pilot data must move from browser/runtime state into workspace-scoped durable storage before production use.

      Workspace diagnostics

      Admin readback for entitlement, durable storage, saved Core/Deep results, Category Association and Market Lens

      Active workspace pending This checks what Pulse has durably saved for the selected workspace. Missing and unavailable states stay visible rather than becoming zero.

      Refresh diagnostics before rerunning scans when a workspace appears stuck.

      Usage by workspace

      Usage status for every durable workspace. Expand a row for internal counters, request receipts and reset history.

      Monthly usage allowance Users see one usage-remaining meter. Saved results stay visible when live usage is paused. Admin can expand each workspace for internal counters, request receipts and reset history.

      Check usage overview to read durable counters, requests and reset receipts from the hosted workspace store.

      Provider resilience

      Privacy-safe latency, adaptive deadline, circuit and recovery state for matched comparisons

      Resilience status loading Only safe timing and failure classes are shown. Provider prompts, responses, credentials and client evidence are never included.
      LoadingReading durable provider state without making provider calls.
      Recent durable resilience events Circuit transitions and provider-free deployment gates only; newest first.

      Missing latency evidence remains unavailable, never zero. External notifications are not active.

      Engine availability and operating rules

      Core, Search and Deep engine readiness, held credentials, queue policy and budget guardrails

      Runtime governance pending Search is a separate evidence channel. Missing engines stay held or unavailable, not zero.
      Meta AI terminal receipt No secret-safe Muse Spark receipt has been captured since the current server started. Refresh engines after a bounded Meta run. Prompts and credentials are never displayed.

      Run runtime governance before live bake-off or model routing changes.

      Workspace plan and access

      Assign Lite, Premium or Custom workspace access, scan credits and invoice state

      Plan, engines and Search policy Lite uses the fixed Core four plus Basic Search; in-depth independent source search and discovery are not included. Premium adds approved Deep engines, Expanded Search and independent source discovery. Custom requires an eligible Search scope.
      Usage allowance policy Allowances protect shared test budgets and stop surprise provider spend. Users see one usage-remaining meter; Admin sees the internal counters and budget controls.
      Usage reset and request path Reset monthly allowance only from Admin, with a reason. Users can request more usage, reduce scan scope, or wait for the monthly refresh.
      Usage this month Update Admin to read durable usage counters, reset receipts and pending usage requests for the selected workspace.

      Select the active workspace above. Saving writes this entitlement to the durable workspace record.

      Feedback-for-discount lifecycle Not started
      Consent and outbound governance Not recorded

      Pilot counters only: scan credits and exports are local/demo usage signals for invoice planning, not live payment data. JSON is for audit handoff; CSV is for invoice spreadsheet handoff.

      Current Access

      Read-only people view built from the saved workspace access records Pulse uses for session and workspace gates.

      Admin-only access view Current Access is the admin readout. Saved workspace access records are the underlying runtime membership rows used by the app. Fixture/test rows are hidden from the main people view.

      Current Access is a read-only overview. Use Refresh access after saving the owner/admin key; use View access list for the pop-up view.

      Workspace Activity Log

      Admin-only timeline of access, reports, exports, requests and scan/usage events for the selected workspace.

      Admin-only activity view Shows redacted workspace activity from audit events, usage events, memberships, saved reports and snapshot requests. Raw prompts, provider payloads, secrets and private client data stay hidden.

      Use the owner/admin key, then refresh activity. This log is read-only and does not enable public signup, CRM writes, checkout or external sends.

      Request Access Queue

      Review public access requests and record GO, HOLD, REWORK or REJECT. Decisions create receipts only; they do not create accounts or workspaces.

      Owner/admin key
      Queue view
      Exports are owner/admin review files only. HubSpot API writes, workspace creation and public signup stay closed.

      Use this queue for reviews until email alerts are configured. Decisions are owner-review receipts only; public signup, automatic workspace creation, checkout, CRM and external sends stay closed.

      Source review approvals

      Name the people who can approve source-learning suggestions before anything is stored or applied.

      Draft queue only This is an owner-decision card. It does not apply source suggestions, write to storage, create client output or change scores.

      Apply queue remains disabled until storage, review owners, server write owner, retention and external wording owner are approved.

      Controlled Demo Guardrails

      Internal staff demo readiness, client selection rules, regulated-sector holds and feedback discount requirements from the RC1 baseline approval.

      First 30 days: trained Pulse staff only Pulse Champions are internal staff cleared to demo Pulse after a mock walkthrough. This is not a customer certification or public product badge.

      Internal demo readiness

      Person Status Mock demo Certified by Allowed scope

      Client selection eligibility

      Feedback discount quality gate

      Regulated-sector holds

      Healthcare, finance, banking, insurance, digital health, medical and pharma accounts need SLT selection before live reports or client demos.

      This panel is an internal operating guardrail. It does not open public signup, public workspace creation, checkout, CRM writes, external sends or production OAuth expansion.

      Access Role Boundaries

      Who can operate platform, staff, client and partner access layers before client login opens.

      Role split before login Master Admin and PRecious Internal are active internal roles. Client Admin, Client User and Partner Admin are future roles until Auth, membership and workspace-boundary gates pass.

      This is an internal access model. It does not open client login, partner access, public signup or public workspace exposure.

      Sign-in and access health

      Credential-owner path for one controlled invite or magic-link execution.

      Runtime-only execution gate This view shows whether the approved internal-action path is ready. It does not display secrets and does not send invites unless the staging execution gate is explicitly enabled.

      Public signup, client login and public durable writes remain closed while this gate is being cleared.

      Release readiness

      Compact go/no-go view before the controlled-pilot release candidate.

      Do not cut rc.1 until Auth execution, membership readback, private API matrix, GA4 journey proof, owner review, hosted smoke and LarsOS go/no-go pass.

      Create Workspace

      Create the client workspace shell, client record, owner membership, invite status and billing posture together. Public signup and client login stay closed.

      Invite-only access model This prepares an internal workspace plus invited owner membership. Send/resend only marks the invite as prepared; Supabase Auth email or magic-link delivery remains held until the invite-only Auth gate is approved.
      Add approved user Use this for a reviewed person who should get managed Pulse access. Choose the workspace, enter their work email, name, role and status, then save. Pulse routes the login method after the user enters their email. Choose password setup if you want Pulse to send the setup invite now. Same email plus same workspace updates the existing row; same email plus another workspace adds another controlled workspace. Public signup, checkout and CRM records stay closed.

      Public signup remains closed. Brand, competitor, market and language setup happens in Settings after the workspace is created.

      Production Foundation Readiness

      Auth & Enforcement Readiness plus Render environment, Supabase storage/Auth, invite acceptance, durable membership readback and client-login guardrails

      Use staging enforcement first. Production client login stays closed until Render, Supabase, membership storage, invite acceptance, readback and audit events are all ready.

      Pulse-Live Channel Gate

      External promotion readiness for PRecious, GA+AI, RevGrowth, SEA Connect, KEK and future partners

      Downstream teams can build held assets now. Active external Pulse-backed promotion waits for Pulse-live clearance.

      Snapshot Request Queue

      Captured BD interest, CRM handoff and launch-readiness storage status

      Runtime request queue is enough for private demos. Supabase is required before public multi-person request handling.

      Proof-Pack Archive

      Review internal proof-pack handoffs only when preparing BD follow-up or owner approval

      Show proof-pack archive and review gates Collapsed by default because this is evidence review, not day-to-day platform setup.

      Proof packs remain internal review records until evidence labels, caveats, source review and durable storage are cleared.

      Workspace Evidence & Output Readiness

      Bundle A/B/C bridge: saved evidence ownership, source mix, commercial usage context and report-output regeneration readiness

      Use this before treating any Authority Snapshot, report, proof pack or campaign brief as a controlled-pilot output. Runtime fallback is not production durability.

      Launch Readiness Acceleration

      Detailed RC evidence board for launch owners. Keep it collapsed unless checking launch gates.

      Show RC evidence board and launch-readiness detail For launch owners only. Public signup, client login, pricing and public claims remain held.

      RC evidence board is read-only. It shows what still blocks controlled pilot readiness before any public/client activation.


      Hosted evidence capture is internal-only. It collects launch-readiness proof while guarded RC submission remains token-blocked.


      RC evidence closure is internal-only. It defines the exact proof packet needed before guarded RC submission can be attempted.


      This panel is internal-only. Public promotion, public claims, pricing, signup, sends, partner access and public report links remain blocked.

      Legacy profile recovery tools

      Administrator-only recovery and cross-profile operations, including resets and deletion. For everyday workspace edits, use Settings. These tools retain their existing permissions.

      Market defaults

      Used as the default market and language context for live scans and reports. Choose a preset, then edit label or languages only when needed.

      Changes are stored in this browser until a backend is connected.

      Public demo guardrails limit editable profiles, monitored competitors and broad scan depth. Trusted local sessions can save and run broader checks.

      Production Integration Path

      What needs to change before this becomes multi-user and live-data ready

      Public access

      Host the app on Vercel, Netlify, Cloudflare Pages, or an internal PRecious domain. Add login before exposing client data.

      GA4

      Use a backend service account with the Google Analytics Data API. The browser should call your backend, not Google directly.

      HubSpot

      Create a private app or OAuth app, store tokens server-side, and expose only aggregated lead/conversion data to this UI.

      Editable CRM data

      Move this browser-stored dataset into a database such as Supabase/Postgres, Airtable, Google Sheets, or HubSpot custom objects.

      Master AI Model Setup

      Review main routes first; expand fallback, held and candidate routes only when needed. Configuration, dated runtime proof and workspace policy remain separate. Workspace-specific access is managed in Workspace Entitlements above.

      Configuration does not prove execution, and a passed smoke does not override a held workspace policy. Lite uses the approved Core four only when runtime-ready. Premium and Custom may use only engines configured, successfully tested and explicitly allowed for that workspace.

      Model & API Operations

      Connector health, active/preferred/fallback model readback and full-audit readiness. Official provider retirement or latest-model checks still require the Engine Baseline review.

      Model checks run from Admin when stale or when triggered manually.

      Provider status Shows live status, active model, preferred route, alternate route, last success and recent issue class.
      Check cadence Wake check, then 6h active / 24h idle. Weekly review before default model changes.
      Model answers vs Search channel Gemini is model-only. Search-grounded discovery is shown as a separate channel and can fail separately.
      Usage guardrails Admin-only: watch at 30% remaining, warning at 15%, critical at 7% or projected 48h remaining. Account states may remain unavailable_not_zero where no safe balance API exists.
      Alert history Admin queue records usage, slow-response and repeated-failure signals for review before heavy scans.
      Client-facing wording Slow, limited or fallback connectors use neutral partial-scan wording; billing and credit detail stays admin-only.

      The app flags newer recommended models but does not silently switch major model families without review.

      Master Route Health

      Authenticated Master Admin only. Run exactly one fixed-sentinel route at a time through the durable private runner. Initial 11-route pass; no retries or raw responses.

      Loading durable route receipts. No provider is called by this readback.

      Initial ceilings: 11 calls, US$0.40, 10 credits. Four later family retests stay held pending safe repairs; absolute ceilings are 15 calls, US$0.60 and 15 credits.

      Repair-family retests: 0/4. No retries. Controls remain disabled until durable initial-matrix and Master-session readback pass.

      Final repair verification: 0/8. One call per unresolved route, no retries, US$0.32/8-credit hard cap.

      Final targeted retest: 0/3. No retries, US$0.12/3-credit hard cap.

      Four-route outcome repairs are loading. Provider execution remains held pending separate authorization.

      Four-route repair v4 is loading. Provider execution is held.

      Source Providers & Evidence Review

      Check search/source-provider readiness and attach source candidates for review when live discovery misses known proof

      Optional source candidates for review Use this for known source URLs that live scans might miss, then review claim fit before use.

      Approved fallback sources are stored with the active profile. They are useful for demos and evidence QA, but they do not replace live third-party source discovery or durable workspace evidence storage.

      App Version & System Monitors

      Release tracking, app responsiveness and cost-control monitors

      Current version Version loading...
      Last updated Checking...
      Latest change Loading latest change...

      Version syntax: Major.Minor.Patch. Major = bigger release, Minor = medium feature change, Patch = small fix or copy/UI adjustment.