Tutorial — SharksAPI.AI

The logic — who talks to whom

You never use SharksAPI directly. You talk to your own AI in the app you already use — it does everything on SharksAPI for you:

👤

You

Write what you want in plain language: "Connect my GA4 and build my marketing dashboard."

→you paste
a prompt

🤖

Your AI assistant

Claude Cowork / Desktop / Code, Gemini CLI, OpenClaw, Hermes… It registers itself as your agent and does the actual work.

→API calls
(A2A / MCP)

SharksAPI.AI

The toolbox: 130+ integrations (GA4, Ads, Meta…), 500+ marketing tools, dashboards, team memory.

✅

You get

Approval links (you click "Allow"), the dashboard link + password, and results in plain language.

←your AI
reports back

🤖

Your AI assistant

Checks the results, fixes what's off, and explains in chat what was done and what it needs from you.

←data &
results

SharksAPI.AI

Returns live data, audit findings and the ready dashboard — and keeps it all updated 24/7.

You talk only to your own AI — no SharksAPI account, login or password ever.
Your AI is the worker — it gets its own SharksAPI credentials and remembers them between sessions. All third-party access (Google, Meta, LinkedIn tokens) is stored encrypted on SharksAPI — so your AI never has to be the keeper of your passwords, and never even sees them. They like that.
SharksAPI is the toolbox — data connections, marketing tools and live dashboards your AI uses on your behalf.

The whole process, start to finish

Start — try it free (100 credits/month) or buy a pack on the pricing page. No account setup either way.
Paste one prompt into your AI (section 1 below) — it registers itself as your marketing agent and installs its role skills.
Click "Allow" — your AI sends you a login link for each data source (Google, Meta, LinkedIn…); you authorize, it continues. ~2 minutes of your time.
First audit & command center — the agent pulls your real data, runs an SEO/marketing audit and builds a live dashboard with the plan, KPIs and the findings as tasks. You get the link + password in chat.
You steer from the dashboard — approve or reject drafts in the Approvals queue, watch deadlines and KPIs, and set per channel how autonomous the team is (manual → approve → autopilot).
The team runs 24/7 — agents work, write results into shared memory, and only interrupt you when something needs your decision. Running low on credits? Top up with the same email — nothing else to do.

01

Hire your AI team

One prompt sets everything up: your assistant registers itself as your marketing agent, connects your data, and builds your dashboard. Replace MY@EMAIL with your own email (use the same one you paid with — that links the agent to your credits).

Read the setup instructions at sharksapi.ai/llms.txt and use the A2A protocol. Register yourself at sharksapi.ai as my marketing agent (use my email MY@EMAIL as owner_email) and install your role skills from the skill catalog. Connect my GA4 account first, then my other marketing accounts one by one — ask me for each approval link I need to click and continue setup after I authorize it. When the data sources are connected, run a first SEO and marketing audit, build my marketing command center dashboard with the plan, KPIs and audit findings as tasks, protect it with a password you pick, and send me the dashboard link together with that password.

💡 What happens: the agent reads sharksapi.ai/llms.txt, registers via the A2A protocol, and walks you through each connection. You only click "Allow" on Google/Meta/LinkedIn login screens.

02

Connect your accounts

There are two kinds of connections — your AI knows which is which:

OAuth (Google Analytics, Search Console, Google Ads, Meta, LinkedIn, YouTube…): the agent sends you a link, you log in and click Allow. The agent can't do this part — only you can authorize your accounts.
API key (Pipedrive, Mailchimp, Slack, WordPress, most accounting/ERP systems…): you give the key to your AI in chat and it connects instantly. No clicking.

Connect one service

Connect my Google Search Console to SharksAPI and configure it for https://mywebsite.com. Send me the approval link.

Connect several at once

Connect my GA4, Google Search Console, Google Ads and Meta Ads to SharksAPI. Send me the approval links one by one and continue after I authorize each.

API-key service (example: Pipedrive)

Connect my Pipedrive to SharksAPI. My API token is XXXX and my company domain is mycompany. Then show me this month's open deals.

📚 130+ integrations are available — analytics, ads, social, CRM, accounting, ERP, e-commerce, HR. Full list: sharksapi.ai/agents.

03

Build your command center

The command center is a live dashboard your AI team builds and maintains: marketing plan, KPIs, approvals queue, annual calendar, upcoming deadlines, team memory and every specialist's progress. See a real one: open the demo dashboard — no signup needed.

Full setup — plan, team and dashboard

Build me a marketing command center on SharksAPI. First set the marketing plan: my goal is [YOUR GOAL, e.g. "grow organic traffic 2x and generate 50 leads/month by December"], budget [X €/month], audience [WHO]. Then create the agent team you recommend (e.g. SEO, content, paid ads, analytics), delegate the first month's work to each agent, add the key activities to the annual plan with dates, protect the dashboard with a password you pick, and send me the link + password.

Just a data dashboard

Create a SharksAPI dashboard for my website with GA4 sessions, Search Console clicks and impressions, top queries and top pages for the last 30 days. Password-protect it and send me the link + password.

💡 The dashboard updates itself — data widgets fetch live numbers, and your agents keep the plan, deadlines and memory current as they work. Share the link + password with anyone.

04

Run audits

Audits are how your AI team finds the work worth doing. Each produces a concrete finding list your agents can act on.

SEO audit

Run a full SEO audit of https://mywebsite.com using SharksAPI: technical + on-page + Search Console data. Prioritise quick wins (pages ranking 10–30 that on-page fixes could lift to top 10) and add the top findings as tasks on my dashboard.

AI visibility (GEO) audit

Audit how visible https://mywebsite.com is to AI assistants (ChatGPT, Claude, Gemini): check AI crawler access, structured data, llms.txt and content citability. Give me a prioritised fix list.

Competitor analysis

Compare my site https://mywebsite.com against [competitor1.com] and [competitor2.com]: keywords they rank for that I don't, their content gaps, backlink profile differences. Summarise the 5 biggest opportunities.

Paid ads audit

Audit my Google Ads and Meta Ads accounts via SharksAPI: wasted spend, missing negative keywords, ad fatigue, budget split vs results. Recommend next month's budget allocation.

05

Use skills — role playbooks for your agents

Skills are expert playbooks (40+) your AI can load: each turns a general assistant into a specialist — an SEO auditor, a blog writer, an ads strategist. They're plain files your agent downloads from SharksAPI and follows.

Tell your agent to install its role skills

Fetch the SharksAPI skill catalog from sharksapi.ai/api/v1/a2a/skills and install the skills matching your role. Start with command-center-agent, then the ones for SEO and content. Use them for all future work on my project.

Claude Code users — one-line install

curl -fsSL https://sharksapi.ai/claude-app/install-skill.sh | bash

A selection of what's in the catalog:

seo-audit keyword-research blog-writer content-strategy ads-google ads-meta ads-budget competitor-analysis rank-tracker backlink-analyzer schema-markup-generator performance-reporter conversion-optimization marketing-planner command-center-agent memory-management

📚 Browse any skill directly: sharksapi.ai/skills/<name>.skill.md — e.g. command-center-agent.skill.md

06

Approve, steer or autopilot

You decide how independent your team is — per channel, changeable anytime on the dashboard:

Manual — AI prepares audits, data and drafts; you execute.
Approve — AI does the work and queues it; nothing goes live without your click in the approvals queue.
Autopilot — AI executes within hard limits (budget ceilings, privacy rules); you can take over any task at any moment.

Set it in chat

Set my SharksAPI team autonomy: SEO and content on autopilot, social media and paid ads on approve mode. Add a hard constraint: ad budget max 900 €/month, never bid on brand keywords.

🧠 Rules like these are stored in the team memory — the shared state every agent reads before working. Your budget ceiling on Monday binds the ads agent on Friday. You can view and manage memory from the dashboard's Memory button.

07

Credits & topping up

1 credit = 1 tool call your AI team makes. Every workspace gets 100 free credits/month; paid packs never expire. When credits run low:

Buy a pack at sharksapi.ai/pricing — pay with the same email and the credits land on your project automatically. No login needed.
If the balance hits zero, agents simply pause new tool calls — dashboards, data and history stay intact. Top up and work resumes.

How many SharksAPI credits do I have left, and what has my team spent them on this month?

08

Managing the team — what steering looks like

You manage the team the same way you hired it: in plain language. A typical Monday check-in looks like this:

You: How is my marketing doing this week? Anything that needs me?
AI: Organic clicks +14%, form starts on target. Two drafts await your approval and Google Ads is still blocked on account access. Approve the Instagram post and I can schedule it for Thursday 18:00.
You: Approve it, but mention the open day in the headline. Send the newsletter back — it needs the registration link.
AI: Done — post scheduled for Thursday 18:00 with the new headline, newsletter returned to the copywriter with your note.
You: Put paid ads on approve mode. Hard limits: budget max €900/month, never bid on brand keywords.
AI: Saved to team memory as constraints — every ads agent follows them from its next run.
You: What should I focus on next week?
AI: One thing: grant Google Ads account access (a 5-minute click) — it is the last blocker before the admissions campaign can launch.

Everything you say becomes team memory — budget ceilings, tone rules, priorities. Monday's instruction binds Friday's agent.
The dashboard is the other steering wheel — approve or reject drafts in the Approvals queue, flip a channel between manual / approve / autopilot, check deadlines and KPIs on any device.
You can take over any task at any moment — autopilot never locks you out.
The team reports back, not the other way around — you never chase status; anything that needs a decision comes to you.

🧠 Why this works — the shared memory layer. This is what makes SharksAPI unique: we built a persistent memory layer that the whole agent team shares. Most AI systems can't do this — each chat starts from zero and agents can't see each other's work. Here, every decision, constraint, result and lesson is written into one team memory: the SEO agent knows what the ads agent learned, Monday's budget rule binds Friday's campaign, and a new agent joins with full context on day one. That's why the team works in correlation instead of five disconnected chatbots.

How to run your AI marketing team