Content
AI compliance checker that reviews marketing copy for Turkey before it goes out
Compliance Checker reads a post, SMS, WhatsApp message, e-mail or ad before it is published or sent, returns a pass, warn or block verdict, lists every issue with the exact excerpt and a fix, and writes a compliant version of the text. It is not legal advice; it is the last check before sending.
- As a workflow node
agent.complianceChecker- Default model
- Claude Sonnet 5
- Economy model
- Claude Haiku 4.5
- Tools
- 0
- Report
- COMPLIANCE_REPORT
What the agent does
The agent first sets the message's İYS class: anything that promotes goods or services is commercial (ticari), while a message only about an order, shipment, appointment or payment is informational (bilgilendirme). A single discount code added to a shipping message makes it commercial. Commercial SMS, WhatsApp and e-mail to consumers need İYS consent, the sender's identity and a free opt-out; merchants and tradesmen (tacir / esnaf) need no prior consent.
It then applies the amendment to the Commercial Advertising Regulation in force from 1 August 2026: sponsored content needs a “Reklam” or “Tanıtım” label visible at first sight and repeated on every part (#işbirliği alone is not enough), human-looking AI characters must be disclosed, deepfake endorsements are banned, the previous price in a discount must be the lowest price of the last 10 days on the same channel, and “best” or “cheapest” claims need proof. Comparative advertising rules are checked too.
In health services, advertising, prices, campaigns and patient testimonials are not allowed; visual posts go out with interaction switched off and before/after images only under conditions. Alcohol promotion is banned, restaurants included, and public posts carry no personal or health data of third parties (KVKK). In a workflow the result stays one item, so an If step can stop sending with {{ $json.verdict === "pass" }}.
Inputs / Outputs
Inputs
- Text to check
- Paste the text, or leave it empty in a workflow; the agent checks the body, caption, subject and hashtag fields of the items from the previous step. Write
{{ $json.body }}to point it at one field. - Channelrequired
- Social media, SMS, WhatsApp, e-mail or ad. İYS consent and the opt-out line are checked for SMS, WhatsApp and e-mail; label and claim rules apply on every channel.
- Sector
- General, health services, restaurant / café, finance or alcoholic beverages. The health promotion and alcohol rules switch on by sector.
- Recipient
- Consumer or merchant (tacir / esnaf). Commercial messages to consumers need İYS consent; merchants need no prior consent but can still opt out.
- Sponsored / influencer content
- Turn on for paid collaborations, gifted products or affiliate codes; the agent checks where the “Reklam” or “Tanıtım” label sits and that every part repeats it.
- Allow web search
- Off by default. When on, the agent may run up to two web searches to check a factual claim such as an award or a ranking.
Outputs
- verdict
- The verdict: pass (can go out), warn (fix or verify first) or block (must not go out as written). One block-level issue makes it block.
- messageClass
- The İYS message class: ticari (commercial) or bilgilendirme (informational).
- issues
- Each issue: a rule id (e.g.
iys.promo_in_transactional,ad.sponsored_label,health.testimonial), a severity, the excerpt from the text and a concrete fix. - revisedText
- The compliant version of the text, keeping the channel format and tone; the original text when nothing needs to change.
- requiredLabels
- Labels and lines the text must carry, such as “Reklam”, the opt-out line or the AI disclosure.
Saves a report to the Reports page on every run: COMPLIANCE_REPORT.
Tools
Integrations it uses
No integration is required.
Example tasks
- “Check this campaign SMS for our café: is the opt-out line there, and does the discount wording follow the previous-price rule?”
- “Check our cosmetics brand's influencer post and tell me where the Reklam label must go on each story frame.”
- “Check our dental clinic's Instagram post; remove everything that counts as advertising and rewrite it as information only.”
- “Check our shipping notification template; if it contains promotion, split it into an informational and a commercial message.”
Tips for good results
- Put the agent right before the sending step and branch with If: passing texts are sent, the rest go to approval.
- Set the channel and sector correctly; the health and alcohol rules only apply when the sector says so.
- Have a person read every block result; the agent gives no legal opinion, it points at the risky parts.
- In high-volume flows switch off Save report on the node; the verdict already travels on the item.
Model advice
Default model
Claude Sonnet 5
Per 1M tokens: $2 input · $10 output
Economy model
Claude Haiku 4.5
Per 1M tokens: $1 input · $5 output
The default balanced model is reliable for subtle calls such as commercial versus informational and for writing a natural rewrite. For bulk checks of short, templated SMS the economy model is enough; stay on the balanced model for health and finance copy.
Templates that use this agent
Frequently asked questions
Does it replace legal advice?
No. The agent is an automated pre-check before sending; it shows the rule and the fix but never says the text is legally approved. Ask an expert for block results and unclear cases.
Can it verify the previous price?
It does not see your price history. When it finds a discount it adds a warning asking you to confirm the previous price is the lowest of the last 10 days on the same channel.
Does it know whether recipients gave İYS consent?
No, it does not check recipient lists; it tells you whether the text is commercial. In campaign sends, recipients without consent are skipped by the platform's İYS check anyway.
Get started with OrqLabs
Build your first workflow today
Start on the free plan: connect your accounts, pick a template and stay in control with approval steps. No credit card required.