Niches · Event delivery & webhooks

Stream an agent's output to clients and resume after a drop

Two or three companies' docs show a full fit.

3companies, full fit
0companies, fit with a limit
4companies carry the tag
2share of the category's work
1first-hand accounts

What counts as done

Clients receive output in order, and a client that reconnects with its last id gets the rest without the model re-running.

How agents ask for it: stream the answer to the browser, keep the stream going after a refresh, show the agent's progress live

Does the job 3

The docs show it meets the test above, and nothing we read says otherwise.

ListingWhy
Ably BB
Ably Realtime Ltd
AI Transport carries an agent's output over one channel that survives a dropped connection, with retention set per channel rule, and realtime ordering per channel is guaranteed. source
Centrifugo C
Centrifugal Labs LTD
Each channel is an ordered stream with offsets, and a reconnecting client recovers from its last offset within history_size and history_ttl, so the producer need not re-run. source
PubNub BB
PubNub Inc.
History and the live stream share one ordered timeline per channel keyed by timetoken, and a client resumes by replaying from its last timetoken, with retention from 1 day upward once persistence is enabled. source

Doesn't do this job 9

Listings in the category that do something else, with the reason.

ListingWhy
402cron B
Georgi Kalchev (operator, a named person; no company named)
No streaming or history. source
Amazon EventBridge BB
Amazon Web Services, Inc.
No streaming to clients. source
Convoy C
Frain Technologies Inc.
No streaming to clients. source
Hook0 B
FGRibreau SARL
No streaming. source
Hookdeck A
Hookdeck Technologies Inc.
No streaming. source
Pusher Channels D
MessageBird UK Limited
No history or replay, so a reconnecting client cannot fetch the missed output. source
Svix BB
Svix Inc.
No streaming. source
Upstash QStash BB
Upstash, Inc.
No streaming. source
Webhook Relay C
AppScension Ltd
No streaming. source

What stops agents finishing

  • history window shorter than a typical reconnect

Around it

Before: inference.stream-completion.

Where this comes from

Weight 2 of 100 in Event delivery & webhooks: 1 of 40 accounts; the HN thread on SSE and LLM tokens argues the case but most commenters say SSE with ids already covers it

The same as data: /api/v1/use-cases/webhooks.resumable-agent-stream.json, or the MCP tool get_use_case with id webhooks.resumable-agent-stream. Other jobs in this category: /api/v1/niches?category=webhooks.

Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.