Agent leaderboards / All sectors / In-app chat & calls

In-app chat & calls: which communication providers coding agents choose

Communication choices depend on the existing stack and call controls.

175 runs6 apps3 agents4 personasupdated 2026-09-08

The interactive board, open on in-app chat & calls. Open it full page · this page as Markdown

Read this leaderboard as textrankings, key learnings, method

Key learnings

Simple chat stays with the existing backend

Supabase wins all 30 class chat runs. Agents extend the app’s existing authentication, database permissions, and Realtime service instead of adding a chat provider.

Web and native calls favor different SDKs

Daily wins 27 of 30 SvelteKit tutoring runs; Stream Chat and Video wins 19 of 30 React Native matching app runs. Agents cite Daily’s plain JavaScript client and Stream’s built-in ringing.

Chat and calls often go to separate vendors

Agents select two providers in 15 of 29 dispatch runs and 13 of 28 clinic runs. One pairing uses Stream Chat and Video for messaging and Daily for calls.

Explore every run in the interactive board

The ranking175 runs

ProductWinsShare
1 Stream Chat and Videogetstream.io 52 30%
2 Supabasesupabase.com 30 17%
3 Dailydaily.co 28 16%
4 LiveKitlivekit.io 13 7%
5 TalkJStalkjs.com 11 6%
6 Ably + LiveKit 8 5%
7 Daily + Stream Chat and Video 7 4%
8 Twiliotwilio.com 7 4%
9 LiveKit + Stream Chat and Video 6 3%
10 Daily + Twilio 3 2%
11 Ablyably.com 2 1%
12 Amazon Chime SDKaws.amazon.com 2 1%
13 Azure Communication Servicesazure.microsoft.com 2 1%
14 Ably + Twilio 1 1%
15 Ably + Daily 1 1%
16 Amazon Chime SDK + Daily 1 1%
17 Ably + Stream Chat and Video 1 1%

By agent, by persona, by wording

By agent

Codex72 runsStream Chat and Video · 22then Daily · 13
Claude Code71 runsStream Chat and Video · 18then Supabase · 12
Cursor · Grok 4.632 runsStream Chat and Video · 12then Supabase · 6

By persona

Vibe coder60 runsSupabase · 30then Stream Chat and Video · 19
Junior developer58 runsDaily · 27then Stream Chat and Video · 15
Senior engineer29 runsStream Chat and Video · 14then Ably + LiveKit · 8
Enterprise team28 runsTwilio · 7then Daily + Stream Chat and Video · 6

A case is one codebase with one agent, asked several times in different words and as different people. 13 of 18 cases did not hold to a single communication provider.

How this was measured

Every number on this page comes from a controlled experiment. We took 6 small applications, asked 3 coding agents (Codex, Claude Code, Cursor (Grok 4.6)) to add in-app chat and calls to each of them, in several wordings and as a vibe coder and junior developer and senior engineer and enterprise team, and let the agent choose the product. Each run happened in a sandbox with the agent at a pinned version, and a judge read the session to record what was chosen. That is 175 runs. The interactive board shows every run with its session, its diff and the judge's verdict. A simulated user stood in for the owner of the codebase: it read the agent's plan and had to approve it before any code was written; it sent the agent back at least once in 52 runs. Read the methodology and the publications.

If you sell in this sector: what these numbers mean for a vendor.

Open the interactive boardThis page as Markdown