AI and data trust
AI does the heavy lifting. People make the calls.
Where we use AI, who checks it, how we protect your data, and what we think AI should be for. The same rules apply whether you work with us directly or under your agency’s brand.
- Companies: a DPA with you, before we touch anything
- Agencies: a DPA with you, for every client you add
How we use AI
Where do you use AI?
On the repetitive parts: mapping systems, drafting workflows, checking data quality, writing documentation, first drafts of reports. Never to make a decision about a business, and never to send something to a customer without a person checking it.
Who checks the work?
A named person reviews every change before it goes live. The review is logged on the board, with the Loom.
Do AI agents talk to customers?
Not unless you ask for it. If an AI agent ever sends a message, it has a name and is labelled as an AI agent in the signature. People always know. (This also follows the EU AI Act transparency rules for AI that interacts with people.)
How we protect data
Does AI see personal data?
As little as possible. Before any AI step, our workflows strip out the fields the task doesn’t need and replace names, emails, phones and company names with placeholders, wherever the task allows. The real values go back in only inside our own system. Some tasks need the real details, like researching a company or matching duplicates. For those, our AI providers are on paid business plans that never train on your data, with zero data retention where they offer it.
Where is the data stored?
In the EU, in Germany and Ireland. We access each tool with our own named logins, stored in a password manager with a separate vault per customer, and we remove our access when the work ends.
Which sub-processors do you use?
Can you sign our DPA or security questionnaire?
Yes. Send it over before you start.
The impact part
AI should give people time back, not take their jobs.
When a system we build frees up a team’s time, someone will ask: can we cut two people now? We’ll push for better uses first.
Move people up.
When the manual work goes, the people doing it can learn the work they actually want, like selling instead of copying data.
Give time back.
Shorter weeks, fewer late nights. A team with a life outperforms a team that is always on.
Grow with the same team.
Keep everyone and let the system multiply what they do, instead of cutting the people who know your customers.
Honest about what’s AI. You always know which parts of our work are AI, which are human and which are both. And we push for the same honesty towards your customers.
No numbers we haven’t measured. We won’t tell you AI saves “10 hours a week” until we’ve measured it in your system, against your month 1 baseline.
For agencies
Under your brand, under the same rules.
- One DPA, with you. Signed before we touch any client data. Your client signs theirs with you.
- Same review on every client. A named person signs off every change, whoever the client is.
- Nothing hidden from you. Your client sees your brand. You see exactly what was AI and what was human, so you can answer their questions.
- Access per client. A separate vault for each of your clients, removed when that client leaves.
What won’t you do?
- ✕Use your data, or your clients’ data, to train AI models
- ✕Share data between customers, including inside our template library (templates are built from patterns, never from your records)
- ✕Let AI make a decision about your business, or message a customer without a person checking it
- ✕Make claims about your system we can’t prove