Shared memory that follows you across AI models.

Sovant gives your apps and agents consistent, model-neutral memory so context moves with you across GPT, Claude, and Gemini.

Inspired by a Fortune 500 engineer's pain: "Every time I switch models, I have to start from scratch."

WORKS WITH:

GPT
Claude
Gemini
More models coming soon

See Shared, Model-Neutral Memory in Action

Watch how Sovant carries context and facts across different AI models.

Sovant Playground
OpenAI · gpt-4o
Message Sovant

How It Works

Three simple steps to give your AI shared, model-neutral memory.

⬇️
01

Install SDK

Drop in the TypeScript or Python SDK. No setup, no vendor lock-in.

🗂️
02

Capture memories

Persist facts, events, and conversations with one call. Sovant keeps them across sessions and models.

🔎
03

Query memory

Retrieve the right context in any app, with any LLM. Memory stays shared and model-neutral.

What Developers Get

Everything you need to capture, inspect, and recall shared context across models.

Dashboard

Dashboard

Monitor, manage and track your project's memory activity in one place.

Demo Chat

Demo Chat

Conversations that don't forget. Context persistence across models.

Memory Explorer

Memory Explorer

Inspect, search and refine memories for precise recall.

Built for Developers

A powerful API that makes AI memory shared, portable, and persistent.

Model-neutral: Works across GPT, Claude, Gemini, and more.

Seamlessly switch between models while keeping context intact.

Semantic Search: Instantly retrieve relevant context.

Find the right memories with AI-powered search.

Type-safe SDKs: Full TypeScript and Python support with auto-completion.

Build faster with intellisense and type safety.

index.ts
npm i @sovant/sdk

// index.ts
import { Sovant } from '@sovant/sdk';

const client = new Sovant({
  apiKey: process.env.NEXT_PUBLIC_SOVANT_API_KEY!,
  baseUrl: 'https://sovant.ai',
});

// Create a memory
await client.memory.create({
  content: 'User prefers dark mode UI',
  type: 'preference',
  tags: ['ui', 'theme'],
});

// Semantic search
const results = await client.memory.search({
  query: 'user theme preference',
  limit: 5,
});
console.log(results);

Simple, Transparent Pricing

Start free, scale as you grow

Free (Beta Access)

$0/month

Perfect for testing and early experiments

  • 2,000 requests per month
  • Shared memory core
  • Dashboard & Memory Explorer
  • Standard persistence (testing)
  • Community support

Founding Developer

$5 one-time

Support Sovant's early development

  • Early supporter badge
  • Priority feedback channel
  • First access to Pro features
  • Recognition in community page
  • Optional upgrade (no feature gating)

Pro

Coming Soon

For production applications and teams

  • Unlimited API requests
  • Advanced memory indexing
  • Priority support
  • Custom analytics
  • 99.9% uptime SLA

Enterprise

Coming Soon

For organizations with compliance & scale requirements

  • Custom deployment (VPC / on-prem)
  • SOC 2 / GDPR readiness
  • Role-based access controls
  • Dedicated support & SLAs
  • Custom integrations & audit logs

A 'request' counts any API call that reads, writes, or patches memory.

Building in Public

We're actively developing Sovant with your feedback. Here's what's coming next.

Product Roadmap

Beta

Current

Core shared memory API with semantic search

Smarter grouping

Automatic context clustering and tagging

More model connectors

Extended model compatibility

Ecosystem

Integrations, plugins, and developer tools

Progress25%

Community & Feedback

Join us on GitHub

Contribute, report issues, or follow along with Sovant's progress building portable, model-neutral memory.

Share your Feedback

Help shape Sovant during beta — tell us where portable memory helps (or hurts) in your workflow.