Trust center
Hand over the work.Keep the guardrails.
Exactly what Taseli can touch in your account, and what it never does.
- You
- A rule
- Autopilot
- An AI agent
Change Service
Google AdsJournal · monitoring · restore
One way in
Every change goes through one checked path. Enforced in code.
Tokens never leave the vault
Encrypted, isolated, never sent to AI models, logs or analytics.
Your data isn't a product
No selling, no sharing, no training AI on your account.
You can trust it with your money.
Six safeguards on every change, enforced on our servers.
Approval by default
Nothing changes in your account until you approve it, unless you turned on autopilot for that change type.
Caps enforced on the server
Daily amount and percentage caps are checked before every write. Out-of-range changes are rejected.
A dry run before every change
Every change is validated with the platform first. If it would fail, it never runs.
A journal of every write
What changed, who approved it, why, and what happened next. Kept for your records.
Honest restoration
Supported settings are restored automatically when results drop. When an exact restore isn't possible, we say so.
Quiet windows
Shabbat and holidays for the account's location: no changes, and ads paused and resumed on time.
OAuth scopes, per platform, and why
We ask for the least each platform allows, and our servers decide what actually runs.
Google sign-in
Liveopenid, email, profileTo sign you in
Google Ads
Livehttps://www.googleapis.com/auth/adwordsRequested separately, after sign-in. Google's only Ads API scope; the free audit only reads
Meta
Coming soonads_read, ads_management, business_management, read_insights, pages_read_engagementRead performance, apply approved changes, agency access
TikTok
Coming soonTikTok Business CenterAdvertiser authorization. Reading performance and applying approved changes
We never request Gmail, Drive or other Google data. Disconnect any time, in Taseli or your Google account.
How your data is protected
- Encryption
- Tokens in a dedicated vault. All data encrypted in transit and at rest.
- Isolation
- Row-level security per workspace, tested on every change.
- No human access by default
- Staff see nothing unless you grant time-limited, logged access.
- Application security
- Firewall, rate limits, secret scanning, encrypted backups, yearly penetration test.
AI, and what it never sees
- AI explains and drafts. It doesn't detect anomalies, and it can't approve anything.
- Anthropic and TypeSafe process under signed DPAs: zero retention, no training on your data. Models never see tokens.
- Content from your account is treated as data, never as instructions.
Retention
| Performance metrics | Up to 25 months while your workspace is active |
|---|---|
| Change journal and restore data | Restore for 30 days; journal kept while active |
| Cancelled workspace | Purged within 90 days, sooner on request |
| Financial records | As required by law, separately from account data |
Compliance
Google API Services User Data Policy
Including Limited Use: Google data powers only features you see.
Israeli Privacy Protection Law, Amendment 13
Security procedures, access logs, incident runbooks and data-subject rights.
GDPR
A customer DPA, a public subprocessors list, transfer safeguards.
Accessibility
WCAG 2.2 AA and Israeli Standard 5568 as the design target.
Responsible disclosure
Found a vulnerability? Email the steps to reproduce. We acknowledge within two business days and don't pursue good-faith research.
security@taseli.co.ilFrequently asked questions
Can Taseli spend my money without asking?
Why does Google's consent screen say “manage your campaigns”?
Can your staff see my account?
What happens to my data if I leave?
An AI campaigner. You just approve.
Taseli does the work. You make the call.
- Free
- Nothing changes without your approval