Asana
Asana's AI Teammates stopped answering questions and started writing the automation themselves.
A side-by-side editorial comparison of Kagi Search and AFFiNE — release velocity, themes, recent moves, and the top alternatives to consider.
| Feature | Kagi Search | AFFiNE |
|---|---|---|
| Sector | Collab | Collab |
| Velocity score | 0.0 | 7.5 |
| Sparks · 30d | 0 | 1 |
| Top themes | search-api, assistant, mobile-apps, user-control | byok, copilot, self-hosting, dual-train |
| Last editorial update | 2h ago | 51m ago |
| Website | — | Visit → |
Kagi is unbundling itself: search as an API, Assistant as an app, AI as a switch.
The dated part of this feed stops in April on housekeeping releases — reliability work, refinements, a Small Web expansion. Everything since sits in undated rows and is more substantial: the Search API went to public preview in May carrying each account's own lenses and blocklists, June started rebuilding search widgets, July added a global off-switch for AI in search, and July 30 put Kagi Assistant on iOS and Android as a standalone app.
AFFiNE just cut its AI loose from a fixed provider — bring your own keys, your own models, per workspace.
AFFiNE publishes from two trains simultaneously — dated canary builds and 0.27.x betas — so the same pull request often appears twice with different version numbers. Most of the recent output is dependency automation: setup-python, actions/cache, RevenueCat SPM bumps. The substantive work sits on the server: blob sync with presigned URLs for S3 and R2, document garbage collection reliability, and now a rebuild of how Copilot gets its models.
The dated part of this feed stops in April on housekeeping releases — reliability work, refinements, a Small Web expansion. Everything since sits in undated rows and is more substantial: the Search API went to public preview in May carrying each account's own lenses and blocklists, June started rebuilding search widgets, July added a global off-switch for AI in search, and July 30 put Kagi Assistant on iOS and Android as a standalone app.
Kagi is separating what used to be one subscription into distinct products with distinct surfaces: a search index other people can query, an assistant that lives on a phone, and a search page where the AI layer is optional. The through-line is user control — preferences that follow the API key, widgets that can be switched off individually, threads that can be exported or deleted in bulk — which is the one thing an ad-funded competitor cannot match on. Growth now depends on those surfaces rather than on convincing people to change their default search engine.
The Search API is the piece with launch details still outstanding, so expect it to leave preview with firm pricing and the beta users migrated; the mobile Assistant is described as a first step, which points to the missing platform features arriving next.
AFFiNE publishes from two trains simultaneously — dated canary builds and 0.27.x betas — so the same pull request often appears twice with different version numbers. Most of the recent output is dependency automation: setup-python, actions/cache, RevenueCat SPM bumps. The substantive work sits on the server: blob sync with presigned URLs for S3 and R2, document garbage collection reliability, and now a rebuild of how Copilot gets its models.
The server is where AFFiNE is investing, and the shape of that investment is self-hosting. Presigned upload URLs configurable per storage provider, doc GC hardening, and now workspace-level model profiles all point at the same customer: someone running AFFiNE on their own infrastructure who wants the AI features without routing content through AFFiNE's vendor of choice. Removing the legacy provider-specific settings in the same change says this is a replacement, not an option.
With provider catalogs and managed tiers now in the model, the next moves are likely on the governance side — surfacing which models a workspace permits and what they cost — since the change also stripped token-cost and provider metadata out of public responses.
Other Collab products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Kagi Search or AFFiNE.
Asana's AI Teammates stopped answering questions and started writing the automation themselves.
Read the Docs deploys weekly, and almost all of it is rebuilding how builds run.
Zoho Projects makes time tracking the product's center of gravity, and lets Zia build the schema.
SiYuan's 3.8 alpha turns the note vault into an agent surface: embeddings, reranking, MCP.
Discord is turning the chat client into a storefront and Nitro into a rewards program.
monday.com is building the control room for AI it already sold customers on.
See all Kagi Search alternatives → · See all AFFiNE alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. AFFiNE is currently shipping more aggressively (velocity 7.5 vs 0.0), with 1 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. AFFiNE is currently shipping more aggressively (velocity 7.5 vs 0.0), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Collab products to evaluate alongside.
Top Kagi Search alternatives in Collab are ranked by recent ship velocity. Browse the "Kagi Search alternatives" section above for the current picks, or visit /alternatives/kagi for the full list with editorial commentary on each.
Top AFFiNE alternatives in Collab are ranked by recent ship velocity. Browse the "AFFiNE alternatives" section above for the current picks, or visit /alternatives/affine for the full list with editorial commentary on each.