MARKETING STUDIO SETUP GUIDE
Set up Amazon Bedrock (AWS) for Marketing Studio
Open an AWS account, get the free credits, turn on Claude and connect Marketing Studio. Short steps, one at a time.
Before you start
Takes about 30 minutes. You need an email address, a phone and a payment card. AWS holds about 1 USD to check the card. The Free plan does not charge you.
What it does in Marketing Studio
Text captions: yes. AI photos: yes, in Regions with picture models (for example us-west-2). Product scenes: no. Video: no. Frankfurt has no picture models, so see the Pictures step, or pair Bedrock with an image provider such as OpenAI, xAI or Google Gemini. Marketing Studio has an “Amazon Bedrock” provider option in the version now being released. The default model will be openai.gpt-oss-20b, and the list of models comes from your own account.
Step 1. Create your AWS account
1.1 Go to aws.amazon.com. Choose “Create an AWS account”.
1.2 Enter your email and an account name. Choose “Verify email address”. Type the code from the email.
1.3 Choose a password.
1.4 Choose “Free (6 months)”. The button says “Choose free plan”.
1.5 Enter your contact information: your name, phone and address. Tick the AWS Customer Agreement.
1.6 Enter your billing information: your card details. Approve the check of about 1 USD in your bank app.
1.7 You land on a page that says “Congratulations”. Activation takes a few minutes, and can take up to about 2 hours.
1.8 Until then, Bedrock shows “Your account is currently being verified”. Wait, then try again.
AWS sign-up ↗ Choosing a plan (AWS) ↗Step 2. Sign in and protect the account
2.1 Sign in. Choose “Sign in using root user email”. Enter the code that AWS emails you.
2.2 Set up MFA. Open the Account menu, then Security credentials. Choose “Assign MFA device”. Choose “Authenticator app”.
2.3 AWS asks for MFA within 35 days. It is better to do it now.
Enable an MFA device for the root user (AWS) ↗Step 3. Earn up to USD 100 extra credit
3.1 The Free plan gives you USD 100 of credit straight away. You can earn up to USD 100 more. This step is optional.
3.2 Open Console Home. Find the “Explore AWS” box. Set its filter to “Earn AWS credits”. Five tasks are worth USD 20 each. Start each task from its link, using “Start activity”.
3.3 Task 1, Budget. Open Billing and Cost Management, then Budgets. Choose “Create budget”. Use the “Monthly cost budget” template. Set 5 USD and enter your email. Keep this budget even after the task. It is a safety alarm.
3.4 Task 2, Bedrock playground. Open Amazon Bedrock, then Playground. Choose “Select model”. Choose Amazon Nova Micro. Type a question. Press Run.
3.5 Task 3, EC2. Name the instance “Test Instance”. Keep Amazon Linux and t3.micro (“Free tier eligible”). Choose “Proceed without a key pair”. Untick “Allow SSH traffic”. Choose “Launch instance”.
3.6 The EC2 task only counts after you remove the instance. Open EC2, then Instances. Tick it. Open Instance state. Choose “Terminate (delete) instance”.
3.7 Then delete its security group. Open Security Groups. Tick “launch-wizard-1”. Open Actions. Choose “Delete security groups”.
3.8 Task 4, Database. The Free plan only allows “Express configuration” (Aurora PostgreSQL). If the orange “Create database” button in the task does nothing, go to Aurora and RDS, then Databases. Open the “Create database” dropdown. Choose “Express configuration”. Choose “Create database”.
3.9 Delete the database after the task counts. First delete the instance: Actions, then Delete, and type “delete me”. Wait about 7 minutes. Then delete the cluster. Untick “Create final snapshot” and “Retain automated backups”. Tick the acknowledgement. Type “delete me”.
3.10 Task 5, Lambda web app. Give it a name. Keep “Create default role” and the blueprint. Tick the Function URL acknowledgement. Choose “Create Event Function”. Open the function URL once.
3.11 Clean up the Lambda task. Open Actions and choose “Delete function”. Type “confirm”. Delete the role it made, in IAM, then Roles. Delete the log group, in CloudWatch, then Log groups.
3.12 The Explore AWS box sometimes switches its filter back to “Resources”. Choose “Earn AWS credits” again.
AWS Free Tier ↗Step 4. Choose your Region
4.1 Open the Region menu at the top right of the console.
4.2 Pick the AWS Region nearest to you and your customers.
4.3 If your data must stay in the EU, choose an EU Region such as Europe (Frankfurt) eu-central-1. Then use models whose ID starts with eu.
Step 5. Turn on Claude (optional)
5.1 Open Amazon Bedrock, then Model catalog.
5.2 Find the banner “Anthropic requires first-time customers to submit use case details”. Choose “Submit use case details”.
5.3 Fill in the company name, website and industry. Say whether the users are internal or external. Write one or two sentences about the use. Do not include personal data.
5.4 Choose “Submit use case details”. The details are shared with Anthropic.
5.5 Wait. AWS says it can take about 15 minutes before Claude answers.
5.6 Some models show “not available for this account”. This includes some OpenAI models on Bedrock. AWS decides this for each account. Pick another model.
Request access to models (AWS) ↗Step 6. Create the key
6.1 Open Amazon Bedrock, then API keys.
6.2 Choose “Generate long-term API keys”.
6.3 Choose an expiry. We suggest 90 days.
6.4 Choose “Generate”.
6.5 Copy the key now. It starts with ABSK. It is shown only once.
6.6 AWS says long-term keys are meant for trying things out.
Bedrock API keys (AWS) ↗ Generate a key (AWS) ↗Step 7. Connect Marketing Studio
7.1 In WordPress open Marketing Studio, then Settings, then AI. Or open the Studio and choose the gear “Settings” in the left rail.
7.2 Set Provider to “Amazon Bedrock”.
7.3 Pick the same Region as in Step 4.
7.4 Paste the key. Choose Save.
7.5 Choose “Load models from your account”. Pick a model.
7.6 Choose “Test this model”.
7.7 Marketing Studio stores the key on your own site and sends requests straight from your site to AWS. Nothing is sent to us.
Step 8. Pictures (optional)
8.1 Bedrock makes pictures only in Regions that have picture models. Today that includes us-west-2 (US West, Oregon). It offers Stability AI Stable Image Core, Stable Image Ultra and Stable Diffusion 3.5 Large. Frankfurt has none.
8.2 You can keep text in your own Region and use a separate Region for pictures.
8.3 AIPF tested Stable Image Core in us-west-2. It made a 1536x1536 picture in about 5 seconds.
8.4 Amazon Nova Canvas reached end of life on 30 September 2026, so do not plan on it.
8.5 In Marketing Studio, open Settings, then AI. Choose “Amazon Bedrock” as the picture provider.
8.6 Set its Region to us-west-2.
8.7 Paste the key. The same Bedrock key works if it is not limited to one Region. If it is, make a new key in us-west-2 (repeat Step 6 there).
8.8 Choose “Load picture models from your account”. Pick a model.
8.9 If a Region has no picture models, the list says so.
8.10 Picture prompts then go to the US Region you chose.
Stability AI models (AWS) ↗ Stable Image Core (AWS) ↗ Stable Image Ultra (AWS) ↗ Stable Diffusion 3.5 Large (AWS) ↗ Model lifecycle: Nova Canvas end of life (AWS) ↗Step 9. Spending
9.1 The Free plan cannot charge you. It uses credits only. AWS closes the account after 6 months or when the credits run out, unless you upgrade.
9.2 If you upgrade to a paid plan, keep the budget alarm from Step 3. AWS Budgets only send alerts. They do not stop spending.
9.3 Many Regions have no picture models. Use the Pictures step, or another provider such as OpenAI, xAI or Google Gemini.
Create a budget (AWS) ↗ Bedrock pricing ↗Stuck?
The filter reset. The Explore AWS box can switch back to “Resources”. Choose “Earn AWS credits” again.
A button does nothing. For the database task, skip the orange button. Go to Aurora and RDS and use “Express configuration”.
The account is still verifying. Bedrock shows “Your account is currently being verified”. This can last up to about 2 hours. Try again later.
Claude does not answer. Wait about 15 minutes after you submit the use case details.
A model says “not available for this account”. Pick another model.
Where your data is processed
AWS states that with Amazon Bedrock your content is not used to improve the base models and is not shared with model providers, and that AWS and third-party model providers do not use customer inputs or outputs to train models. Requests are processed in the Region you choose, but some models are offered through cross-Region inference profiles, such as those whose ID starts with global, and with those your data may be processed outside your Region’s geography. AWS says to check the destination Regions of the profile if you have data-residency requirements. If you make pictures in a Region with picture models, such as us-west-2 (US West, Oregon), the picture prompts go to that US Region.
Bedrock FAQs (data use) ↗ Data protection ↗ Inference profiles and regions ↗ AWS Customer Agreement ↗ AWS privacy notice ↗What it costs
Prices are per token and depend on the model and region. See the Bedrock pricing page.
Amazon Bedrock pricing ↗Checked against the provider’s own documentation on 17 October 2026. Provider pages and labels change; if a label differs, look for the nearest equivalent.
Back to all setup guides