Question 1

How much does Volt26 do on its own, versus just recommending?

Accepted Answer

Volt26 agents do real work, not just advice. In conversation they act, saving research, capturing ideas and drafting campaigns, and overnight the team reviews your signals and business numbers on a schedule and queues proposed work for your morning sign-off. Every action is logged to an audit ledger: low-stakes work executes on its own, and anything costly or outward-facing waits for your explicit Authorize. External actions (email, ads, integrations) are the next stage of the same spine, behind the same gate. You set the dial per agent and per project; on the Starter plan every email or post waits for your OK, and every tier above gets the full autonomy dial.

Question 2

Where is Volt26 customer data stored?

Accepted Answer

Customer data at rest is stored in AWS ap-southeast-2 (Sydney, Australia) in a Supabase-managed Postgres database. Application delivery runs through Vercel's global CDN, but all persistent reads and writes resolve to the Sydney database. The full sub-processor list is published at volt26.ai/subprocessors.

Question 3

Is my data used to train AI models?

Accepted Answer

No. Volt26 does not use your inputs or outputs to train any AI model, and Anthropic, Volt26's sole AI provider, does not train on API customer data. Volt26 calls Anthropic's Claude models directly with no AI broker or middleware in between.

Question 4

What security controls does Volt26 have?

Accepted Answer

Every organisation is isolated with PostgreSQL Row-Level Security enforced at the database layer. Data is encrypted in transit (TLS 1.2+) and at rest (AES-256). Direct database access is allow-listed, and any support impersonation is recorded to an append-only audit log. Volt26 publishes a live list of controls still maturing at volt26.ai/trust.

Question 5

What happens to my data if I leave Volt26?

Accepted Answer

Your data is yours. Ask and we export your organisation's data in full (self-serve export is on the roadmap), and organisation data is deleted within 60 days of termination.

Question 6

Could I just assemble this myself with Claude and a few agents?

Accepted Answer

Yes, and some founders should. Claude gives you skills, schedules and webhooks, which is most of the raw material. What you would be taking on is the assembly: deciding the roles, writing and maintaining the prompts, plumbing your business data into every call, deciding what each agent may do unattended, building the approval step and the audit trail, and keeping the whole thing running while you also run your company. Volt26 is that assembly, kept up to date, with a bill you can predict.

Question 7

Does every Volt26 customer get the same advice?

Accepted Answer

No. Volt26 agents work from your organisation's vault, where every conversation, research run, and decision is stored, not from generic best practice. Two founders asking the same question get different answers because their context differs, and recommendations keep diverging the longer you use it. This compounding memory is Volt26's core design.

Question 8

What does one action buy, and what does it cost you?

Accepted Answer

One action is one agent turn: a question answered in the room, a signal read, an idea captured. Heavier jobs are charged as a bundle of actions so the number stays predictable: a play is 5, a deep business audit 20, a prototype 20, a full six-agent research run 30. Starter includes 250 actions a month, Founder 1,500, and extra packs are 100 for $10, 500 for $45 or 1,000 for $80. What those cost us in model time, measured on the last 30 days: a plain turn 3 to 5 cents, a Board meeting about 21 cents, a research run about 34 cents at the median. We publish that because almost nobody in this category says what a credit actually buys. A reply cut short at the length limit is flagged on the message and you can ask for the rest.

Service	Purpose	Data location
Supabase (on AWS ap-southeast-2)	Database, auth, storage	Sydney, AU
Anthropic	AI model inference	US (see AI data handling below)
Vercel	Application hosting, CDN	Global (static/edge only)
Resend	Transactional email	US
Stripe	Payment processing (card data stays with Stripe)	US
PostHog	Product analytics	US
Userback	In-app feedback (page metadata + console logs)	Australia

Layer	Approach
Data in transit	TLS 1.2 or higher on all connections (browser to Vercel, Vercel to Supabase, Supabase API calls)
Data at rest	Encrypted at rest via AWS/Supabase default encryption (AES-256)
Backups	Encrypted at rest; managed by Supabase
Payment card data	Not stored by Volt26, Stripe handles all card data under their PCI DSS compliance programme

Topic	Position
Training on customer data	No. Anthropic does not train on API customer inputs or outputs. Volt26 does not use your data for model training.
Anthropic data retention	Standard: approximately 30 days (safety/abuse monitoring). Content that Anthropic's safety systems flag can be kept for up to 2 years. We are actively applying for Zero Data Retention (ZDR) with Anthropic; will update this page when confirmed.
Data path	Your inputs go directly from Volt26 to Anthropic's API and back. No intermediate storage of raw prompts/outputs by Volt26 at this time (see gaps below).
Autonomous agent actions	Agents do work inside Volt26 on their own: they save research and artefacts, capture ideas, start research runs, draft campaigns and support replies, and queue proposals overnight. Anything outward (an email through your connected sending account, a social post, a campaign send) waits for your explicit Authorize, and every action is written to an audit ledger. Agents do not execute code, make purchases or write to your external systems.
Safety screening	We run no content filter of our own. Anthropic's usage-policy safeguards apply to every call. Content from outside your own account (feeds, uploaded files, customer messages, search results) is labelled as untrusted in the prompt, and any action an agent proposes on a turn that carried it is held for your Authorize.

Item	Status
Formal least-privilege staff access controls and access review process	In progress
Prompt and retrieved-context logging for incident reconstruction	In progress: currently no persistent log of AI inputs/outputs for forensics
Self-serve data export	Not yet available; contact legal@volt26.ai to request an export
Deletion-on-termination flow	In progress: account deletion currently requires a manual request
Published data retention policy	In progress
SOC 2 Type II	Not yet started; planned as enterprise demand requires it

Trust & Security

Data residency

Sub-processors

Encryption

Tenant isolation

Access control and admin auditability

AI data handling

Availability

Privacy and legal framework

What we're still building

Responsible disclosure

Contact