Dev48
Language
  • About
  • Services
  • Industries
  • Technologies
  • Articles
  • Contacts
Book a call
    Home/Articles/Agent api now supports reusable agents
Dev48

© 2026 · All rights reserved.

Agent API now supports reusable agents

Источник: Perplexity AI

Agent API now supports reusable agents

Source: Perplexity AI

Teams create reusable agents with versioned Profiles, Skills, and managed connectors they can share across applications.

September 29, 2026

Teams get more from AI when their agents reuse the same tools, instructions, and authentication across applications. Finance teams, for example, want every monthly close and variance analysis to follow a consistent process. That happens when AI agents follow the same reconciliation instructions and reporting format across applications, with each application supplying the relevant financial data.

Today we’re making these agent workflows easier with updates to Perplexity Agent API that give teams a reusable, versioned configuration layer for agents. We’re adding custom Skills, managed connectors, and Profiles to the Agent API. Profiles assemble an agent by saving its model, instructions, tools, Skills, connectors, and run settings under one versioned ID.

The updated Agent API gives project administrators shared resources that members can reuse, while Profiles and Skills keep configuration versioned and repeatable. Project administrators can configure an agent once, save it as a versioned Profile, and make that setup available to authorized developers across the project.

Updates create shared project resources for agent workflows

The updates create a project-level place to manage approved external integrations. The Agent API lets project administrators share managed connectors, reusable Profiles, and versioned Skills with their teams.

Project members can reuse the same tools and instructions without rebuilding the configuration in every application.

With the Agent API updates, teams get:

  • Reusable Profiles: A Profile stores a versioned configuration for the model, tools, and run settings used by an agent workflow.
  • Versioned Skills: Teams can package instructions, procedures, and supporting files as versioned Skills that members invoke from the same project.
  • Managed connectors: Project administrators can configure approved integrations for Agent API requests without asking every developer to wire the service separately.

Users can also create their own customer connectors to give agents access to internal systems and services not covered by the managed connector catalog

Agent API creates consistent workflows across applications

Together, Profiles, Skills, and managed connectors let teams configure an agent once and reuse it across applications and workflows. Here are some example use cases:

  • An incident-response team can combine Datadog telemetry, a GitHub repository, a read-only investigation Skill, and a Slack update in one saved agent. The application supplies the incident details. The Profile supplies the model, instructions, tools, service connections, and run settings.
  • A platform team can package its deployment checks, rollback criteria, and reporting format into a versioned Skill. Project members can invoke the same procedure from an internal tool, continuous integration workflow, or service automation. The team can publish a new version of the Skill instead of changing the instructions in every application.
  • A project administrator can configure an approved GitHub or Slack Connector once and make it available to project members. Applications reference the shared Connector instead of storing separate credentials and tool definitions. When access changes, the administrator updates the connection in one place.

It’s easy to update shared resources

We introduced the Agent API as a managed runtime for agentic workflows in March, bringing search, tool execution, and multi-model orchestration into a single integration. Last month, we expanded it into one programmable surface for web search, URL fetching, code execution, MCP, and specialized search tools.

This update adds a reusable team layer to that foundation. Teams move from one-off workflows to agents that run across multiple applications. A shared Profile keeps the model, instructions, tools, and service connections together. Skills and connectors provide the procedures and access the agent needs.

Teams can update these shared resources as their workflows change instead of rebuilding each implementation. Developers get a consistent foundation for extending the same agent into new applications, services, and automations.

Resources can be configured from the API Console

Profiles and custom Skills are available now for Agent API projects. Managed Connectors are available in preview, with GitHub, Slack, Google Drive, Datadog, Linear, and Notion currently supported.

Project administrators can configure these resources in the API Console. Project members can then reference them in Agent API requests using an API key from the same project.

Get started with the Agent API today.

← All articles

More in AI & Machine Learning

All →
OpenAI sparked Hugging Face bids with early investment offer ahead of Nvidia's $13 billion dealПресса
OpenAI

OpenAI sparked Hugging Face bids with early investment offer ahead of Nvidia's $13 billion deal

OpenAI still doesn’t seem to have a handle on all of its rogue AI activityПресса
OpenAI

OpenAI still doesn’t seem to have a handle on all of its rogue AI activity

ElevenLabs’ new v4 speech model supports more expression control and 90 languages
Пресса
ElevenLabs

ElevenLabs’ new v4 speech model supports more expression control and 90 languages

Meta, Google, Amazon, Microsoft draw Sen. Warren questions about AI tax subsidiesПресса
Amazon

Meta, Google, Amazon, Microsoft draw Sen. Warren questions about AI tax subsidies

LangSmith Custom Apps: Build custom interfaces around your agent data
LangChain

LangSmith Custom Apps: Build custom interfaces around your agent data

New in LangSmith: Engine v2, Managed Deep Agents, Fine-Tuning, and more
LangChain

New in LangSmith: Engine v2, Managed Deep Agents, Fine-Tuning, and more

More from Perplexity

AI Alignment: Current Research and Debate
Perplexity

AI Alignment: Current Research and Debate

Introducing Portable Computer
Perplexity

Introducing Portable Computer

Portable Computer comes to AMD-Powered Agentic PCs
Perplexity

Portable Computer comes to AMD-Powered Agentic PCs

Scrunch vs. Peec AI: Choosing the right AEO tool [2026]
Perplexity

Scrunch vs. Peec AI: Choosing the right AEO tool [2026]

AEO checker tools that measure answer engine visibility [2026]
Perplexity

AEO checker tools that measure answer engine visibility [2026]

Keeping Commerce Weird Podcast: New Tech Cannot Rewrite Human Nature
Perplexity

Keeping Commerce Weird Podcast: New Tech Cannot Rewrite Human Nature