Roadmap

What is next

Everything in the v1 launch lives in the changelog. This page is what is on the way next, grouped by area. Click any row to open the details. As items ship they move out of this list and into the changelog.

Modes

Voice agents

Run any Platos agent through a voice provider. Same agent config, new transport.

NextQ3 2026

→

Messaging agents (SMS, WhatsApp, Telegram)

Twilio + Meta + Telegram bots backed by the same agent runtime, persistent threads.

Next

→

Email agents

Inbound email triggers a turn; outbound goes through a verified domain.

Next

→

Slack and Teams native agents

Bot users that participate in channels and DMs with full thread continuity.

Next

→

Plugins

Consume external MCP servers

Plug any MCP server (Notion, GitHub, weather, your own) into a Platos agent. Tools auto-imported, tool-calls forwarded.

NextQ3 2026

→

Plugin marketplace

Browsable, installable catalog of community-contributed skills and tools.

Next

→

More official skills

Web research, codegen, image generation, document parsing, browser automation.

Next

→

Versioned plugin distribution

Pin a plugin to a version. Stage updates. Roll back on regression.

Next

→

Memory & RAG

RAG hybrid retrieval

BM25 plus semantic retrieval, fused. Reranker on top.

Next

→

Citation artifact in chat UI

First-class citations with source spans rendered inline in the chat panel.

Next

→

Memory as a product

Public REST and SDK over the unified memory primitive (Mem0 parity).

Next

→

Shared eval corpus

Open eval suites that any Platos user can pull and contribute back to.

Exploring

→

Platform

Helm chart and horizontal scale

Multi-replica agent runtime, Postgres read replicas, ClickHouse cluster.

Next

→

Alternate vector stores

Qdrant and Milvus as drop-in replacements for pgvector.

Next

→

Multi-region deployment

Recipes for active-active across regions with conversation pinning.

Next

→

Agent cluster builder UI

Visual canvas for composing sub-agents and shared memory pools.

Next

→

Observability

Trace export to LangSmith and Langfuse

OTel-based exporter so trace tooling teams already use sees Platos traces.

Next

→

Cost forecasting

Predicted spend per agent and per user based on the last 30 days.

Next

→

DX & Embedding

@platos/embed web component

Drop-in chat widget you embed on any site, points at any Platos agent.

Next

→

Platos CLI

Scaffolding tool for agents, skills, entity backends.

Next

→

VSCode extension

Edit prompts, run evals, inspect traces without leaving the editor.

Next

→

Studio mode

Visual prompt editor with eval runs side-by-side.

Next

→

Hot reload for skills in dev

Edit a local skill source, see changes in the running agent without restart.

Next

→

Out of scope

Self-hosted models

Run inference inside the Platos VPS rather than via API providers.

On hold

→

Platos Cloud (managed hosting)

Hosted Platos with us running the runtime, monitoring, and model layer.

On hold

→

Talk to Platos

Powered by the Platos runtime

Greetings. I am Platos — the docs, made conversational. Ask me anything about the runtime, and I will answer from what is written, not what I imagine.

Powered by Platos →