Skip to content
gentic.news — AI News Intelligence Platform
Connecting to the Living Graph…

Replit Agent

product stable
Replit

Replit Agent is an AI-powered coding tool from Replit that autonomously builds, deploys, and manages software projects, enabling rapid development with minimal manual intervention.

1Total Mentions
+0.40Sentiment (Positive)
0.0%Velocity (7d)
Share:
View subgraph
First seen: May 21, 2026Last active: May 21, 2026
Profile✓ Researched · 2026-06-24high confidence

Replit Agent is the autonomous AI coding agent built into Replit's browser-based, cloud development platform. From a natural-language prompt it plans, writes, tests (clicking through the live app in a real browser), debugs, and deploys full-stack web and mobile apps end to end, with no local setup. It is one of the leading "vibe coding" / app-builder tools, helping drive Replit to ~$253M ARR and a $9B valuation (Series D, March 2026) with 50M+ users.

Built by
Replit
Category
Autonomous web/mobile app-building agent (cloud AI IDE)
Open source
No
Models
Runs primarily on Anthropic Claude (Sonnet/Opus 4.x family) with a Google Gemini 3.1 Pro option; users can select among frontier third-party models per task (Pro/Enterprise unlock the most capable models). Inline code completions use Replit's own Replit Code v1.5 model.

Available on

Web (browser-based cloud IDE)iOS / Android mobile appsAvailable inside Claude (Anthropic) as an integration

Pricing

Starter: free (free daily Agent credits, 1 published app). Core: $20/mo ($18/mo billed annually) with $25 in monthly credits, unlimited apps, up to 5 collaborators, up to 2 parallel agents. Pro: $100/mo ($90/mo billed annually) with $100 in monthly credits, up to 15 collaborators / 50 viewers, up to 10 parallel agents, most powerful models, Turbo mode, 28-day DB restore, credit rollover (1 month). Enterprise: custom (SSO/SAML, single-tenant, VPC peering). Usage beyond credits is pay-as-you-go via effort-based per-checkpoint pricing.

Key capabilities

  • Autonomous build-test-fix loop: the agent periodically tests the running app in a real browser preview (clicking buttons, forms, APIs) and self-corrects in a reflection loop
  • Extended autonomy: runs up to ~200 minutes per task with a Max Autonomy mode; Agent 4 adds plan-while-building and parallel agents working simultaneously on different components
  • Design Canvas: an infinite board to prototype UI/visual variants before code is written, plus multi-artifact output (web apps, mobile apps, slides, data apps, animations)
  • Agent Customization (June 2026): always-on Custom Instructions injected into every project plus reusable on-demand Skills, set in Workspace Settings
  • Package Firewall (June 2026, with Socket): blocks malicious/compromised packages at install time inside the agent loop; on by default, blocking ~8,000 packages/day
  • Effort-based credit pricing with Economy / Power / Turbo modes, integrations (Slack, Linear, Notion, Telegram), and one-click deployment with managed database

Strengths

  • +Fastest 'idea to deployed app' path: no local setup, builds and hosts full-stack apps from a prompt, accessible to non-developers
  • +Genuinely autonomous: self-tests in a real browser and fixes its own bugs over long (hours-long) runs
  • +Strong model lineup (frontier Claude models plus Gemini option) and broad scope now covering web, mobile, slides and data apps
  • +Built-in supply-chain security (Package Firewall) and security tooling baked into the build loop

Limitations

  • Effort/credit-based billing is unpredictable and can climb fast; users are charged per checkpoint even for failed or no-op attempts, and Core credits don't roll over
  • Best suited to greenfield app generation rather than working inside large existing/enterprise codebases or precise issue-patching
  • Cloud-only and proprietary — no self-hosting, and reliability/cost of long autonomous runs varies; Replit's headline speed/cost claims (e.g. '10x more autonomous', '3x faster') are self-reported, not independently benchmarked

Best for

Founders, designers, and non-developers (and developers prototyping) who want to go from prompt to a deployed full-stack web or mobile app fast, without managing local tooling.

What's new

Replit Agent 4 ("Built for Creativity") launched March 11, 2026, alongside a $400M Series D at a $9B valuation — adding the Design Canvas, parallel agents, plan-while-building, single-shared-project collaboration, and multi-artifact output (web/mobile/slides/data apps). June 2026 added Agent Customization (Custom Instructions + Skills) and the Socket-powered Package Firewall (on by default, blocking ~8,000 malicious packages/day).

Sources (13)
  1. 1. replit.com
  2. 2. replit.com
  3. 3. replit.com
  4. 4. replit.com
  5. 5. replit.com
  6. 6. replit.com
  7. 7. docs.replit.com
  8. 8. developer-tech.com
  9. 9. blog.replit.com
  10. 10. blog.replit.com
  11. 11. infoq.com
  12. 12. anthropic.com
  13. 13. cloud.google.com

Signal Radar

Five-axis snapshot of this entity's footprint

live
MentionsMomentumConnectionsRecencyDiversity
Loading radar…

Mentions × Lab Attention

Weekly mentions (solid) and average article relevance (dotted)

mentionsrelevance
01
Loading timeline…

Timeline

1
  1. Product LaunchMay 20, 2026

    Replit reported 50% QoQ growth in AI-assisted coding in Q1 2026.

    View source
    growth:
    50% QoQ

Relationships

No relationships mapped yet.

Recent Articles

No articles found for this entity.

Predictions

No predictions linked to this entity.

AI Discoveries

No AI agent discoveries for this entity.