AI Policy | NEO GTM

Effective date: 1 October 2025 · Revision date: 21 December 2025

This policy explains how GTM Intelligence Ltd (“GTM”, “we”) uses artificial intelligence in the NEO GTM platform, which providers process data on our behalf, and the commitments we make about your data. It supplements our Terms & Conditions, Privacy & Cookie Notice and Data Processing Agreement.

1. Where AI is used

AI-assisted features in NEO GTM are limited to the following:

Drafting outbound messages — generating email, LinkedIn and call-script drafts for a named recipient, using your own positioning, messaging hooks and case studies.
Summarisation and classification — condensing source material into signals and assigning them to categories.
Enrichment — deriving structured attributes (such as a role archetype, or the category a contract sits in) from text we already hold.
Account and opportunity scoring — ranking accounts and contacts against your addressable market.
Next Best Move and go-to-market recommendations — suggesting the action most likely to advance an account, and surfacing what is working so you can do more of it.
Assistant and connector — answering questions about data you are already entitled to see, including through our connector for Claude.

This list is not fixed. We continue to add AI-driven intelligence designed to help you be more commercially successful; new AI features operate under this policy.

2. Model providers

We do not train or host our own models. AI features are delivered through third-party model providers acting as sub-processors. The providers in use today are:

OpenAI — the primary provider for drafting, summarisation, classification and enrichment.
Anthropic (Claude) — assisted features, and the NEO GTM connector for Claude, which lets an authorised user query their own workspace’s data from Claude.

We choose the model that produces the best result for each feature, and we may switch, add or remove model providers as models improve — the field moves quickly and we keep our options open to serve you better. We will not name a provider we do not use. Any material change to the sub-processor list is notified under the change process in our Data Processing Agreement, so you can review a new provider; the commitments in this policy apply to whichever provider is used.

3. Your data and model training

Your data is never used to train AI models, and never leaves your workspace to benefit anyone else.

Data you upload or sync is not used to train, fine-tune or improve any model — ours or a provider’s. We use our model providers under terms that do not permit data submitted through their APIs to be used to train their models.
No pooling. One customer’s data is never combined with another’s. Each workspace is logically segregated, and an AI request is scoped to the data the requesting workspace and user are entitled to see.
Minimum necessary. Content sent to a model provider is limited to what the feature needs — the recipient’s professional details, your own messaging assets, and the relevant source text.
We do not sell your data, and we do not use it to build or enrich our own datasets.

4. How NEO improves with use

NEO gets more useful the more you use it — but that improvement stays inside your own workspace and works for you alone. When NEO sees which messages earned replies, which accounts engaged, or how a deal progressed in your connected CRM, it uses those signals to sharpen the recommendations and personalisation it gives you.

This is tuning to your activity, not training a model. Your data is never used to train shared AI models, is never pooled with other customers’ data, and never influences what any other customer sees. One workspace’s results never leak into another’s.

Separately, we continuously improve the platform itself — our methods, models, data operations and intelligence — and those improvements benefit every customer. That product-level improvement is built from our own research and from aggregated, de-identified operational learnings about how the system performs — performance signals such as reply and engagement rates, not the content of your messages or records; it never draws on the content of your workspace, and your confidential data is never the raw material for a feature another customer receives. Making the service smarter for everyone and keeping your data isolated are not in tension — we do both.

5. Human oversight

AI output in NEO is a draft for a person to review, not an action taken on your behalf. Generated messages are written into your outbox and are not sent automatically unless you explicitly enable scheduled sending. A user can edit or discard any draft before it is used.

6. Scoring and profiling

NEO scores and ranks organisations and business contacts to help you prioritise who to approach. This is profiling in the sense of the UK/EU GDPR, and we describe it plainly rather than deny it. It is used only to order a sales list.

NEO is not designed to make, and must not be used to make, decisions producing legal effects concerning a person or similarly significantly affecting them — for example decisions about employment, credit, admission to a course, or access to a service.

7. Accuracy

AI output can be wrong. Generated text may contain inaccurate statements about an organisation or an individual, and derived attributes are inferences rather than verified facts. Some contact details in our datasets are inferred from common address formats at an employer and may be inaccurate — this is described in our Privacy & Cookie Notice.

Output must be reviewed by a person before it is relied on or sent. The Services are provided without warranty of accuracy or completeness, as set out in the Terms.

8. Your own data, CRM and email

NEO does not need any confidential information from you to activate or to deliver value. You do not have to share client lists, contracts, your CRM, pricing or any internal document to switch the platform on. NEO is built on market intelligence it already holds and generates; everything in the previous sections works before you connect a single system of your own.

Some customers choose to connect a CRM or upload internal information — client lists, account notes, deal records — so NEO can also reason over what you already know. This is purely optional and additive, and NEO delivers value without it.

When you do, that information is your data, processed by us only as your processor under the Data Processing Agreement, strictly on your instructions.
We connect with read-only, least-privilege access, scoped where supported to the objects and fields you choose, keep it isolated to your workspace, never pool it, never use it to train models, and return or delete it on termination. We do not write back to your systems unless you explicitly enable it.
Outbound email activity tracking (open, reply and send signals) is an optional part of NEO Core and the only feature that needs access to a connected Gmail or Outlook mailbox. It is enabled per user, only when you choose to connect one, through a verified, GDPR-compliant email integration provider we engage for this purpose, requesting the minimum access needed to track messages sent through the platform.
If you would rather work with your data directly, NEO exposes an API: pull NEO’s enrichment, signals and scoring into your own systems and build intelligence on top of it, under the same access and isolation controls.

9. Security and location

Data sent to model providers travels over encrypted connections and is subject to the same access controls, workspace segregation and sub-processor safeguards described in our DPA, including EU Standard Contractual Clauses and the UK IDTA where a transfer requires them. Our system of record is hosted in the United Kingdom by default, with EU or US regions available by agreement.

10. Questions

Questions about this policy, or requests for our current sub-processor list, can be sent to tiago@neogtm.com.