AI Policy

Funnelfeedr is a sales platform built around a simple idea: your sales process should live in the system, not in your reps' heads. AI is how we make that practical — it puts the right knowledge in front of the right rep at the right moment. This page explains where AI runs in the product, what data it touches, and the rules we hold ourselves to.

We wrote this in plain language on purpose. If anything is unclear, ask us — contact details are at the bottom.

1. Where AI runs in Funnelfeedr

• Real-time call coaching. During live calls in the dialer, AI transcribes the conversation as it happens and surfaces the matching battle card or talking point from your team's own Knowledge Base — your USPs, objection handlers and playbook, written by you.
• Email intelligence and drafting. AI reads synced email threads to triage them, extract signals, and draft replies and outreach grounded in your Knowledge Base and the deal context. AI never decides on its own to send an email — a rep reviews drafts and hits send.
• Call and meeting summaries. After a call or a recorded meeting, AI produces a transcript, a summary and suggested next steps, so reps don't spend their evenings on admin. If the host has switched it on, the summary is also emailed to the external participants who attended (see section 3).
• Deal and conversation analysis. AI extracts action items, objections and qualification signals from conversations to keep deal data current.
• Prospect data quality. AI helps us keep contact and company data fresh — matching records against current sources, deduplicating, classifying job titles, and flagging records that should be updated or removed.

What AI in Funnelfeedr does not do: it doesn't decide on its own to send an email, it doesn't make or take calls without a rep, it doesn't score your employees for HR purposes, and it doesn't make any decision with legal effect about anyone.

2. What data the AI processes

Depending on which features you use: call audio and live transcripts from the dialer; meeting recordings and transcripts; your Knowledge Base content; CRM records you've connected (deals, contacts, activities); email threads synced through the platform; prospect contact and company data; and notes your team adds.

3. A human is always in the loop

AI output in Funnelfeedr is a suggestion, never an action. Coaching prompts are hints on screen — the rep decides what to say. Drafted emails are drafts — the rep edits and sends. Summaries and extracted deal data can be corrected before anyone treats them as fact. You control which AI features are enabled for your team.

Some emails do go out without a click at the moment of sending, but only because a user set them up in advance:

• Scheduled emails a rep has written and scheduled.
• Meeting reminders a user has configured for a meeting type, sent from the host's mailbox before each booked meeting. A meeting type has no reminders until someone adds them.
• Meeting recaps to external participants, if the host has turned on Automatically email external participants a meeting recap. It is off by default, only covers meetings the user hosted, only goes to confirmed attendees, is only sent when the recording holds a real conversation, and every recap carries an unsubscribe link.

See Email and Calendar Permissions for what Funnelfeedr does with a connected mailbox and calendar.

4. Where your data lives, and who processes it

Funnelfeedr is built and hosted in Sweden. Our platform — including your recordings, transcripts, emails and CRM data at rest — runs on Microsoft Azure in the Sweden Central region, inside the EU.

For AI processing, we use a small set of specialised providers under data processing agreements:

Our meeting notetaker runs on software we host ourselves on our own Azure infrastructure — no third-party notetaker service sees your meetings.

Storage and hosting are EU-based. Some AI providers process data on infrastructure in the United States; where that happens, transfers are covered by data processing agreements and EU-approved safeguards, and the processing is transient — data is not stored by the provider. Workloads we classify as sensitive are restricted to a vetted EU/US endpoint allowlist that excludes providers we can't verify.

One exception is worth stating plainly. Where you ask us to look into a problem in your own data and no anonymised copy will reproduce it, our engineers may use Anthropic's Claude to help diagnose it, in our centrally administered Claude Team workspace and under Anthropic's data processing agreement. This is not part of any product feature and never happens on our own initiative — only when you ask. Unlike the AI processing above it is not transient: the conversation stays in our workspace until we delete it, which we do once the investigation is finished. We take the narrowest extract that reproduces the problem, never include another customer's data, and record in the support ticket that AI-assisted troubleshooting was used.

5. Your data is never used to train models

Your calls, emails, Knowledge Base and CRM data are used to serve you — not to train or fine-tune AI models, ours or anyone else's. Our AI providers process data transiently under agreements that do not permit training on it, and where a provider offers a data-collection setting, we disable it on every request.

6. Sub-processors

Every provider above is a documented sub-processor under our Data Processing Agreement, and we give notice before adding or changing one.

7. EU AI Act

Under the EU AI Act (Regulation 2024/1689), Funnelfeedr's AI features are limited-risk AI systems: they assist salespeople with transcription, suggestions and summaries. They are not designed for, and must not be used for, high-risk purposes such as employment decisions. In AI Act terms, Funnelfeedr is the provider and you, the customer, are the deployer of these features.

8. Accuracy has limits

Speech recognition mishears words. Summaries can miss nuance. A drafted email can be confidently wrong. We work continuously on quality, but AI output is provided as-is and should be verified before you rely on it for anything important — pricing, legal terms, commitments to customers. That's another reason the human stays in the loop.

9. How long we keep data

Your recordings, transcripts, emails and CRM data are retained for as long as your account is active, so your team keeps its full conversation history. When data leaves the platform's working systems, it leaves quickly:

• Call recordings are removed from our telephony provider immediately after they're transferred to our own EU storage.
• Raw meeting audio from our notetaker is deleted as soon as processing completes; only the processed recording remains, in our own storage.
• AI providers do not retain your data after a request is processed.

You can request deletion of specific data or your whole account at any time via privacy@funnelfeedr.com, and any person in our prospect database can request erasure regardless of any customer relationship (see section 10).

10. Prospect data: fresh, verified — and removable

Funnelfeedr provides prospect and company data. Company-level data is anchored in official business registries. Contact-level data is collected and continuously verified by us — and here's the part most data providers won't say: we remove contacts we can no longer stand behind. Stale data wastes your reps' time and fails the people behind the records.

On our public company directory, personal contact details are masked — full email addresses and complete phone numbers are never shown.

If you're listed and don't want to be, email privacy@funnelfeedr.com and we'll act on it. Depending on what you ask for, we can redact your name from directory pages, suppress your record entirely, or erase it. Erasure is designed to stick: erased contacts are added to a cryptographic suppression list (we keep only a hash, not your details) that blocks the record from ever being re-collected — by import, by our data pipeline, or by our browser extension.

11. Coaching data and your team

Call coaching means AI processes what your reps say, and coaching results are visible to reps and their managers — that transparency is the point: the system exists to make every rep your best rep. Funnelfeedr does not automatically score employees or feed performance reviews, and we recommend being open with your team about how coaching data is used before enabling it.

12. Your responsibilities as a customer

You're the data controller for your calls, emails and CRM data. That means you:

• Ensure a legal basis for recording and analysing calls in your market. In Sweden, a participant may record their own calls without notifying the other party; other countries have stricter rules, and GDPR transparency obligations can apply regardless. Funnelfeedr leaves disclosure to you and your reps — configure your process to match the law where you sell.
• Configure permissions and features to fit your policies.
• Keep a human reviewing AI output before it affects a real person.

13. Security

All data is encrypted in transit and at rest. Access is role-based, credentials and API keys are held in a hardened vault, and everything runs on Microsoft Azure's EU infrastructure. If a breach affects your data, we'll notify you without undue delay in line with our agreement and GDPR.

14. No automated decisions with legal effect

Nothing Funnelfeedr's AI produces is an automated decision with legal or similarly significant effect under GDPR Article 22. Decisions about deals, prospects and people are made by people.

15. Changes to this policy

We'll update this page as the product and regulation evolve, and notify customers of material changes per our agreement.

16. Contact

Funnelfeedr Sveavägen 20, 111 57 Stockholm, Sweden privacy@funnelfeedr.com