The agentic network is the layer of the GTM Decision System that turns context into action. The Customer Context Graph establishes what is true about an account. This layer decides what to do about it, and produces the guidance, analysis and assets your teams work from.
What sits in this layer
Four kinds of component do the work, and most workflows use more than one.
Coaches are interactive and user-facing. A coach is a guided, multi-step workflow built for a specific sales motion, such as preparing for a meeting, building a prospecting sequence or developing messaging. Base coaches cover common motions out of the box, and custom coaches are configured to your own motion, assets and methodology.
Agents run in the background with no user input. They continuously enrich account context, detect change and surface risk, so that something arrives in front of a seller without anyone having thought to ask for it.
Assets are generated documents in your own branding, produced as PowerPoint, Google Slides or Word. Your team requests one against a target account, and can refine the result conversationally rather than editing it by hand.
Inline guidance is produced inside a workflow when a downloadable document is not what you need: a recommendation, a risk, a piece of analysis delivered in place.
Why the reasoning holds together
Components in this layer chain, so the output of one becomes the input to the next without the context degrading along the way. A signal an agent detects can trigger a coach, and the coach can produce an asset, all reasoning over the same account context rather than starting again from scratch at each step.
Two consequences matter in practice:
The same reasoning is applied the same way every time. What a seller gets does not vary with their skill at asking, so the quality of an account point of view no longer depends on who built it.
The reasoning stays intact wherever the component runs. A coach invoked from your CRM, from an AI assistant or from a front end you built reaches the same conclusion it would in the Databook application. See Deployment for how that works.
What coaches and agents draw on
Everything in this layer reasons over the Customer Context Graph, which means both the verified third-party intelligence about your customer and the first-party context you have connected. In practice a coach can use:
Category | What it covers |
Company identification | Industry, size, headquarters and other basic company facts |
Company strategy and intelligence | Strategic priorities, business challenges and leadership statements |
Financial and performance data | Revenue, growth metrics and opportunity scoring |
News and events | Recent announcements, earnings calls and conferences |
Technographics | Information about the company’s tech stack |
People and contacts | Executives and decision makers, with titles and background |
Products, solutions, use cases & product marketing | Your own value propositions, solutions and customer case studies |
Sales methodology and strategy | Your sales methodology, sales motion and strategy |
CRM, calendar, call transcripts and other first-party data | Deal stages, engagement history and upcoming meetings |
Where your team extends this layer
GTM AI Studio is the build environment for the agentic network, and it is designed so that the people who own the number can configure the workflows they depend on without an engineering ticket or a release cycle.
Coach Studio for coaches. You clone a base coach and edit it, which leaves the original live and unaffected for everyone else. Edits are described in plain language through the Composer, and you can add or remove steps, adjust the output format, preview against a real account and restore an earlier version.
Prompt Studio for reusable prompts. A prompt is a saved instruction that produces a consistent answer without the user having to write it themselves, and it can include variables such as account, industry or use case that the user fills in before running it. Active prompts appear in the Prompt Library.
SmartTemplate Studio for assets. You upload a PowerPoint or Word file containing placeholders, map content to those placeholders with the Composer, preview the result against a real account, and set it live for your team to generate on demand for any account.
Building and setting live are two separate things. Who in your organisation can do each is decided by the roles you assign, so you can open building up widely while keeping control of what becomes the organisation's standard. See the AI Studio collection for the detail on each Studio.
Example use cases
A base meeting preparation coach cloned and edited so that its output follows your own discovery framework, then set live for the whole sales team.
An executive briefing asset built once from your branded template, then generated self-service against any target account by anyone who needs one.
A prompt with a use case variable that lets a seller run the same structured research across different industries without rewriting anything.
A background agent that watches for management changes at your named accounts and surfaces them, rather than waiting for a seller to check.
What to expect
Outputs reflect the context you have connected. A coach that reasons over your own solutions and deal history produces something specific to you. The same coach without that context produces something more generic.
Generation is not instant for long outputs. A multi-slide asset reasons over the account before it builds, so expect it to take longer than a chat response.
Custom coaches diverge from base coaches once cloned. Improvements Databook makes to a base coach do not flow into a clone you have already edited.
Getting started
Every organisation starts with the base coaches and assets already available to it. Speak to your account team about which workflows to configure to your own motion, and which of your first-party sources to connect so the guidance reflects your position in the account rather than a generic view.
Read GTM Decision System for how this layer fits with the rest, and Deployment for where your teams work with it.