Top agentic AI use cases for Media and Publishing in H2 2026

Zero-click triage for topics losing clicks to AI Overviews

Chartbeat data in the Reuters Institute’s 2026 trends report shows Google search referrals to more than 2,500 news sites fell by about a third in the year to November 2025, and media leaders surveyed expect search referrals to fall by 43% over three years. The audience team owns the response; the agent finds which topics are losing clicks and recommends where format or distribution should change.

the workflow, in one room

1. ai teammatespeople
2. TriggerPull query-level clicks and impressions from Search Console each week and join them to section and topic tags in the analytics tool.
3. Flag topics where impressions hold but clicks fall, the pattern of an AI answer satisfying the query on the results page.
4. Check which of those queries show an AI Overview or AI Mode answer and which sources it cites.
5. Recommend a response per topic: original data or analysis a summary cannot give, a live page, newsletter or app capture, or less investment in commodity queries.
6. Approved by Audience editor and Section headsThe audience editor and section heads approve which recommendations go to the newsroom planning meeting.
7. Track the chosen topics for direct, newsletter and app visits over the following weeks.

Guardrail. Recommendations inform editorial planning only; what gets covered and how stays with editors, and the agent never rewrites published journalism to chase AI answers.

AI crawler and licensing ledger for the archive

AI companies crawl news and entertainment archives, and publishers now choose between blocking, licensing and pay-per-use access through tools such as Cloudflare’s crawler controls or the IAB Tech Lab’s Content Monetization Protocols (CoMP), first proposed as the LLM Content Ingest API. Business development owns licensing; the agent keeps a ledger of which bots take what, under which terms, and which AI answers cite the publisher.

the workflow, in one room

1. ai teammatespeople
2. TriggerRead CDN and server logs weekly and classify AI crawler requests by operator and declared purpose.
3. Compare each crawler against the robots.txt file, CDN rules and any signed licence in the contract register.
4. Sample AI assistants and answer engines on the publisher’s core topics and record citations, attribution and links back.
5. Flag crawlers taking content without terms and answers that reproduce reporting without attribution.
6. Approved by Head of business development and Legal counselThe head of business development and legal counsel approve any change to crawler access and any evidence used in a licensing negotiation.
7. Publish a monthly ledger of crawl volume, citations and referral traffic per AI company.

Guardrail. The agent never changes access controls or contacts an AI company itself; licensing terms and any legal claim are decided by people.

Preferred Sources and direct-follow prompts for loyal readers

Google’s Preferred Sources, launched in the US and India in August 2025 and rolled out globally in all supported languages in April 2026, lets readers pick outlets that appear more often in Top Stories. The audience team owns direct relationships; the agent runs prompts asking loyal readers to add the publication as a preferred source, follow the WhatsApp Channel or install the app.

the workflow, in one room

1. ai teammatespeople
2. TriggerIdentify loyal readers from analytics and the subscription platform: frequent visitors, registered users and newsletter readers.
3. Draft prompts for newsletters, article footers, app banners and social posts using the Preferred Sources link Google provides to publishers.
4. Set up variants by audience and placement so the team can see which ask works.
5. Approved by Audience editorThe audience editor approves prompt copy, placements and frequency caps so loyal readers are not over-asked.
6. Publish the prompts and track preferred-source link clicks, WhatsApp Channel follows and app installs.
7. Report which loyal-reader segments now arrive directly or through Top Stories.

Guardrail. Prompts go only to readers whose consent covers the channel used, and frequency caps keep the ask from crowding out the journalism.

Cancellation and billing conversations through Convexa

US subscription rules are in flux: the Eighth Circuit vacated the FTC’s click-to-cancel rule in July 2025, the FTC reopened rulemaking with an advance notice in March 2026, and ROSCA still requires a simple way to stop recurring charges. Customer care owns cancellations; Convexa handles billing and cancel requests in chat and voice, makes one approved save offer and cancels if the person still wants to leave.

the workflow, in one room

1. ai teammatespeople
2. TriggerTrigger when a subscriber opens chat, messages on WhatsApp or calls about billing, a price change or cancelling.
3. Verify the account and explain the plan, next charge date and price from the billing system.
4. If the person asks to cancel, make one save offer from the approved list matched to the reason given.
5. Cancel straight away if the person declines, confirm in writing and stop the next charge.
6. Approved by Customer care leadThe customer care lead approves the save-offer list, scripts and escalation rules before they go live, and reviews disputed cases.
7. Log reason codes and outcomes so pricing and editorial teams see why people leave.

Guardrail. Cancelling stays as easy as signing up, with no repeated offers or forced calls, consistent with ROSCA in the US, the UK subscription contracts regime under the DMCC Act due in January 2027, and the opt-out the RBI e-mandate framework gives Indian customers.

Failed renewal recovery under RBI e-mandate rules

Many subscribers lapse because a payment fails, not because they chose to leave. Under the RBI’s Digital Payments: E-mandate Framework, 2026, recurring debits need a one-time registration with additional factor authentication and a notification at least 24 hours before each charge, so lapsed mandates and expired cards are common. The subscriptions team owns involuntary churn; the agent finds at-risk renewals and drafts recovery messages.

the workflow, in one room

1. ai teammatespeople
2. TriggerPull renewals due in the next two weeks with mandate status, card expiry and last payment outcome from the billing platform.
3. Flag renewals where the mandate has lapsed, the card expires before the charge, or the amount needs fresh authentication.
4. Draft reminder and re-authorisation messages for email, SMS and WhatsApp with a link to the gateway’s secure update page.
5. Approved by Subscriptions leadThe subscriptions lead approves the message templates, timing and any grace-period offer.
6. Send reminders, retry failed charges within the payment gateway’s rules and stop retries when the grace period ends.
7. Report recovered renewals and failure reasons by payment method.

Guardrail. Messages never ask for card numbers or one-time passwords; customers update payment details only on the gateway’s authenticated page, and reminders do not replace the bank’s own pre-debit notification.

OTT churn-risk saves after a season finale

Streaming subscribers often cancel once the show they joined for ends, then return for the next big release, a churn-and-return pattern Deloitte’s Digital Media Trends research tracks. The retention team owns this; the agent spots viewers going quiet after a finale and builds a personal next-watch plan before their renewal date.

the workflow, in one room

1. ai teammatespeople
2. TriggerTrigger when a subscriber finishes a series or franchise and has not started another title within a set number of days.
3. Read their viewing history, profile language and renewal date from the subscriber data platform.
4. Build a next-watch shortlist from the catalogue and release calendar, including titles arriving before their next renewal.
5. Draft in-app, push and email messages with the shortlist and trailers already cleared for promotion.
6. Approved by Retention leadThe retention lead approves segment rules, message templates and any discount, which is held back for subscribers who have said they will cancel.
7. Send the messages, then measure whether viewers start a new title and renew.

Guardrail. Messages use viewing data only where consent and the privacy notice cover personalised recommendations, and every promoted title carries its age rating.

Consent-checked audience activation for ads and sponsors

First-party data now carries more of the advertising business, and the rules on using it are tightening, with India’s DPDP Rules, 2025 phasing in through May 2027. The ad operations team owns audience products; the agent checks consent before any reader segment is synced to an ad platform or offered to a sponsor.

the workflow, in one room

1. ai teammatespeople
2. TriggerTrigger when a campaign or sponsor package requests a first-party audience segment.
3. Build the segment from registration, subscription and reading data in the customer data platform.
4. Remove anyone whose consent record does not cover advertising or the named purpose, and anyone on a paid tier promised fewer or no targeted ads.
5. Check the segment is large enough that no individual reader can be identified before it is shared.
6. Approved by Data protection leadThe data protection lead approves each new segment definition and its purposes before first use.
7. Sync the segment to Google Ad Manager, Google Ads or Meta and log the consent basis against the campaign.

Guardrail. No reader data is used for advertising without a recorded consent that covers that purpose, as the DPDP Act, GDPR and the UK’s PECR require.

Creator collaborations briefed and checked for disclosure

Audiences, younger ones especially, increasingly find news and entertainment through creators on video platforms, and publishers are partnering with them for launches and sponsored work. The partnerships team owns these collaborations; the agent drafts creator briefs and checks every post for the required disclosure before and after it goes live.

the workflow, in one room

1. ai teammatespeople
2. TriggerTrigger when a sponsor deal or launch plan includes creator posts.
3. Shortlist creators whose audience matches the brief, with their past sponsored posts and how they disclosed them.
4. Draft the brief with key messages, only the claims the sponsor has substantiated, and the disclosure label and placement required.
5. Approved by Partnerships leadThe partnerships lead approves the creator list, the brief and each post before publication.
6. Check each live post for the label, its placement and the platform’s paid partnership tag, and flag any that are missing.
7. Report reach, engagement and sign-ups or tune-in per creator to the sponsor and the audience team.

Guardrail. Every paid post carries an upfront label such as “Ad” or “Paid partnership”, as the ASCI influencer advertising guidelines and the FTC Endorsement Guides (2023) require, and the platform tag alone is not treated as enough.

Frequently asked questions