> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gravitygtm.com/llms.txt
> Use this file to discover all available pages before exploring further.

# What is Gravity Agents Lite?

> Autonomous Claude-based agent fleets for pipeline, meeting prep, fundraising, and more — deployed per-customer, configured through a self-serve MCP server.

Gravity Agents Lite is a suite of autonomous, Anthropic-Claude-based agents
that run your day-to-day GTM motion for you — reading your inbox and
calendar, watching your pipeline, prepping you before meetings, and
drafting the outreach and digests a GTM team would otherwise write by
hand. Each agent runs on a schedule (or in response to an event like an
upcoming meeting), reads from your own tools, and posts its output to
Slack or writes it straight into your pipeline — a fleet that works in
the background and shows its work, whether or not anyone's watching.

As of 2026-08-14, that background fleet is also a live conversation when
you want one. A Slack assistant responds to a direct message or an
@mention for on-demand web, company, and people research, and reads Slack
itself to help answer what it's asked. The same approval cards the fleet
already posts for an email or a LinkedIn-post draft are now interactive
too — tap **Approve**, **Edit**, or **Cancel** right on the card instead
of acting on it somewhere else.

<Note>
  This conversational layer is read-tier today, not a replacement for the
  autonomous fleet. It can research on demand and act on a card already in
  front of you, but it can't yet search or read Gmail, Sheets, Drive, or
  Calendar as part of a live conversation — those stay automated-agent-only
  for now — and most write actions still happen through the autonomous
  fleet or an approval card rather than a free-form chat request.
</Note>

<Tip>
  Looking for the on-demand LinkedIn/company-data enrichment API instead? See
  [Horizon](/horizon/quickstart) — a separate product on this same docs site.
  Agents Lite can *use* enrichment data as part of a workflow, but the two
  ship, price, and deploy independently. See
  [How this differs from Horizon](#how-this-differs-from-horizon) below.
</Tip>

## What the agents actually do

The fleet is modular — you turn on the agents relevant to your team, not
all of them. Typical modules include:

* **Pipeline Strategy** — a daily digest of where every deal stands and
  what needs attention next, posted to Slack.
* **Meeting Prep** — a brief before every meeting, pulled from your
  calendar, past notes, and CRM/pipeline history.
* **Fundraising Digest** — investor pipeline status, outreach state, and
  next steps, for teams actively raising.
* **Email Follow-up**, **Market Intelligence**, **GTM Signal Capture**,
  **LinkedIn Engagement**, **Investor Relations**, and others, depending on
  what your team needs.

Every module reads from the tools you already use (email, calendar,
meeting notes, a data store, Slack) rather than asking you to adopt a new
system of record. See
[Available integrations & modules](/agents-lite/integrations-and-modules)
for the full, current list of what's built and what each module requires,
and [Agent reference](/agents-lite/agent-reference) for exactly what each
module does, reads, and writes — trigger and schedule, inputs, outputs, and
what's configurable today versus onboarding-time-only.

## The three deployment tiers

Every agent fleet needs four pieces of infrastructure underneath it: a
place the agent code runs (GitHub Actions), a Claude/Anthropic account the
agents think with, somewhere the Slack app is hosted, and a small
key-value store the agents use to avoid double-processing the same event.
Who owns each of those four pieces is a spectrum, not a single switch —
and, as of 2026-08-12, the default isn't the same for all four:

* **Claude/Anthropic, app hosting, and the key-value store** default to
  **Gravity-managed** — the fewest accounts for you to create, hand-
  provisioned by a Gravity operator on request. Moving ownership of one of
  these three to you makes sense when you have a specific reason to hold it
  yourself — infrastructure you already have in place, or a concrete
  compliance or data-residency requirement — not as a general goal to work
  toward.
* **Code hosting (GitHub) defaults the other way — to customer-owned.** A
  Gravity-owned repository holds your actual business configuration (your
  ICP, deal sizes, Slack IDs, content corpus), and a Gravity-hosted default
  here would mean that data sits in a repo under one shared Gravity
  credential alongside other customers' — a security review found that
  concentration risk isn't worth the small amount of setup it saves you
  (creating a private repo and adding Gravity's deploy bot as a
  collaborator). Ask your Gravity contact if you'd specifically like
  Gravity to host your repo instead — it's still available, just not the
  starting point.

The small set of enrichment APIs stays Gravity-owned regardless of tier
either way (see [Security &
credentials](/agents-lite/security-and-credentials#3-secrets-that-stay-entirely-with-gravity)).
Tier 2 is available as a stepping-stone for teams who want to take on
just one or two pieces without going all the way to tier 3:

1. **Gravity-hosted** — Gravity provisions and owns all four: your
   agents' code repository, the Anthropic environment, the hosted Slack
   app, and the key-value store. Available on request, but **not the
   default** — GitHub specifically defaults to customer-owned per the
   security rationale above, so choosing this tier means explicitly
   asking Gravity to take on your repo too.
2. **Hybrid (mixed ownership, the default for a new deployment)** — you
   own some of the four pieces while Gravity manages the rest, and which
   pieces is a genuinely independent choice per piece, not a fixed second
   option. **The starting-point mix is: Gravity manages Claude/Anthropic,
   hosting, and the key-value store; you keep your own GitHub repo.**
   Talk to your Gravity contact if you'd like a different mix — e.g.
   taking on your own Anthropic environment, or handing GitHub to
   Gravity too.
3. **Fully customer-managed** — you own all four: your own GitHub repo
   (with Gravity added as a collaborator to push agent code), your own
   Anthropic account and API key, your own hosting for the Slack app, and
   your own key-value store. Almost every secret you generate stays
   entirely in your own accounts; the two exceptions (your Anthropic API
   key and your key-value store credentials) are shared with Gravity once,
   securely, since Gravity's provisioning step needs them — see
   [Security & credentials](/agents-lite/security-and-credentials) for the
   full trust model. **Appropriate when you have a specific reason to own
   Anthropic/hosting/KV yourself** — existing infrastructure investment
   already in place, or a concrete compliance or data-residency
   requirement — not as a default starting point (GitHub, as always,
   stays customer-owned by default regardless of tier). See
   [Getting started](/agents-lite/getting-started) for the onboarding
   conversation every new deployment starts with, and the account-setup
   walkthrough this tier uses once you submit it.

<Note>
  **"Zero Gravity"** is an automated, fully self-serve version of the
  Gravity-hosted tier above — provisioning all four pieces without a
  Gravity operator doing it by hand, eventually eliminating per-customer
  GitHub repos entirely (Gravity's or the customer's), which is what
  makes today's GitHub-customer-owned default a deliberate stepping stone
  rather than a permanent architecture. It's still in development and not
  yet generally available. Until it ships, choosing the Gravity-hosted
  tier means a Gravity operator sets the four pieces up on your behalf,
  not a one-click automated flow.
</Note>

Whichever tier you're on, day-to-day configuration works the same way —
through the self-serve config MCP server described in
[Self-serve configuration](/agents-lite/self-serve-config).

## How this differs from Horizon

[Horizon](/horizon/quickstart) is an on-demand enrichment API: you (or an
agent) ask it a question — find this person's email, look up this
company — and it answers in one API call, billed per call from a prepaid
credit balance. Gravity Agents Lite is the opposite shape: long-running,
autonomous agents that work on a schedule without being asked, deployed
once per customer, configured rather than called. They're built and sold
independently, though a deployed agent fleet is free to use Horizon (or
any other tool) as one of the resources it reads from during a run.

## Next steps

* [Getting started](/agents-lite/getting-started) — the onboarding
  conversation every deployment starts with, and the account-setup journey
  and go-live steps that follow
* [Self-serve configuration](/agents-lite/self-serve-config) — how the
  config MCP server works, day to day
* [Security & credentials](/agents-lite/security-and-credentials) — the
  trust model for every secret involved
* [Available integrations & modules](/agents-lite/integrations-and-modules) —
  the full current list of what's built
* [Agent reference](/agents-lite/agent-reference) — what each module does,
  reads, and writes, module by module
* [Onboarding reference](/agents-lite/onboarding-reference) — the exact
  field-by-field detail behind the onboarding conversation