Now in early access

The agent backend for your app

General Augment gives your product a secure backend for in-app chat, messaging channels, memory, tools, and proactive agent workflows - without building the agent platform yourself.

General Augment
Agent backend
Identity & user linking
Memory & embeddings
Tool execution
Channel delivery
Approvals & audit
Usage & cost controls
Secure runtime
Configure once. Use everywhere. Govern centrally.
Open Responses-compatible APIGCP-first deploymentMulti-tenant isolationpgvector memoryVertex AI models
The hard part

Adding AI chat is easy for a demo. Shipping it in production is not.

The hard parts begin after the demo: user identity, memory, tools, approvals, message delivery, retries, costs, logs, security reviews, and model upgrades. General Augment gives your app the agent backend those features need.

user identitymemory & embeddingstool validationapproval gatesmessage deliverydelivery retriescost controlsaudit logssecurity reviewmodel upgradeschannel routingcredential vault
How it works

Three steps to a production agent

01

Connect your app, users, tools, and channels

Link General Augment to your existing app backend. Bring your user accounts, APIs, data sources, and the messaging channels your users already use.

02

Configure one secure agent with memory, policies, and model preferences

Define tools, memory scope, safety policy, identity rules, usage budgets, and model routing. One configuration governs the agent across all surfaces.

03

Use that agent from your backend, chat, messaging, and jobs

The same tenant agent answers in-app chat, runs backend workflows, delivers via WhatsApp/SMS/Telegram, and executes scheduled briefings - all governed centrally.

General Augment - agent backend
In-app chat
WhatsApp
SMS
Telegram
Scheduled jobs
Backend workflows
One agent, every surface

One agent across every surface your users are on

Configure a single tenant agent with your tools, memory, and policies. That same agent handles in-app chat, runs background workflows, and delivers through WhatsApp, SMS, and Telegram - identity linked back to your app users throughout.

No per-channel agents to maintain. No drift between what your app does and what your messaging bot says.

Feature pillars

Everything a production agent backend needs

Add an agent without rebuilding your app

Keep your existing product and backend. General Augment adds the agent layer around it: chat, memory, tools, identity, observability, and messaging.

bring-your-own-app

Open Responses-compatible API

Use POST /v1/responses so your backend can adopt General Augment with a familiar OpenAI-style interface. No vendor lock-in.

responses-compatible

Memory without another vector database

Store and search user memory through managed pgvector-backed memory and provider-agnostic embeddings. No separate infrastructure to operate.

managed-pgvector

Tools, MCP, and approvals

Connect APIs, generate tools from OpenAPI specs, use MCP servers. Apply validation, timeouts, approval gates, audit logs, and credential boundaries.

mcp - openapi - audit

Messaging channels built in

Deliver the same agent through in-app chat, WhatsApp, SMS, and Telegram. Channel identities are linked back to your app users automatically.

whatsapp - sms - telegram

Usage, cost, and safety controls

Track agent turns, tokens, tool calls, channel delivery, and costs per tenant. Enforce budgets before model dispatch. Full audit trail included.

budget-gates - audit-logs
Developer API

Works like an API you already know

General Augment exposes an OpenAI Responses-compatible endpoint. Your backend, jobs, and product chat all call the same API - with memory, identity, tools, and channel delivery handled automatically.

POST /v1/responses - Responses-compatible
curl
curl https://api.generalaugment.com/v1/responses \
  -H "Authorization: Bearer $GA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "balanced",
    "input": "Build a morning briefing for this user",
    "user": "app-user-123",
    "metadata": {"source": "spark"}
  }'
Use cases

Built for agent-powered products

Productivity

Proactive personal assistant

Daily briefings, meeting prep, smart alerts. One agent answers in-app and pushes to the channels your users prefer.

in-apptelegramsms
Health & Care

Care navigation

Secure patient messaging with escalation paths, task follow-up, and policy-aware tool calls. Full audit trail.

in-appwhatsapp
Finance

Policy-aware support agent

Marketplace or fintech app support with approvals, audit logs, and budget enforcement baked in before dispatch.

in-appbackend
Vertical SaaS

Tool-using product assistant

Connect your product APIs as tools. Let the agent answer questions, run operations, and surface results in-app or via message.

in-appslacksms
Operations

Scheduled briefings & alerts

Background agent jobs that run on a schedule, pull from product data, and deliver briefings via Telegram, SMS, or WhatsApp.

telegramsmswhatsapp
Agencies

Multi-client agent infrastructure

Multi-tenant isolation means you can run agent features for multiple clients on one platform without shared state or credential exposure.

multi-tenantall channels
Security & governance

Built for teams that need to ship with confidence

General Augment was designed with multi-tenant isolation, credential safety, and audit requirements from the start - not bolted on afterward. GCP-first deployment with GKE, Cloud SQL, Redis, KMS, Secret Manager, and Vertex AI.

Project-scoped API keys & tenant isolation
Every tenant is isolated. Credentials, memory, and audit events never cross tenant boundaries.
User identity linked across channels
Link a user's app account to their WhatsApp, SMS, or Telegram identity. One user profile, every surface.
Credential vault & KMS
Long-lived credentials stay in the platform vault and KMS. Tool calls never expose secrets to the model.
Tool validation, rate limits & approval gates
Tool calls are validated, rate-limited, audited, and optionally gated on human approval before execution.
Budget gates before model dispatch
Usage budgets are enforced at the platform level before any model call is made - not after the fact.
Pricing

Simple, usage-based pricing by agent turns

Developer
$49
per month
  • 1 app
  • 1,000 monthly active users
  • 10,000 agent turns / mo
  • 5 tools
  • In-app chat channel
Get started
Startup
$299
per month
  • 1 app
  • 10,000 monthly active users
  • 100,000 agent turns / mo
  • 20 tools
  • All messaging channels
Start building
Growth
$999
per month
  • 100,000 monthly active users
  • 500,000 agent turns / mo
  • Custom & private MCP tools
  • All messaging channels
  • Priority support
Get started
Enterprise
Custom
contact us
  • SLA & security review
  • Data residency controls
  • Optional dedicated deployment
  • Custom onboarding
  • Design partner access
Talk to us

SMS/WhatsApp delivery and complex model usage may be billed separately. Telegram has no provider delivery fee but still counts as agent turns.

Your product should not build its own agent platform

Start shipping AI features your users actually experience - not infrastructure your team maintains.