side quests
the pursuits around fast apply — the prior pivots, the plugins and integrations, the other models, and the community packages.
the founders' quests
- 2023lingtual · endedlingtual — the original s23 identity: a no-code platform for 'connected ai business software,' pitched as retool+zapier for ai applications with 50+ tool integrations. the domain registered mid-batch; the pitch lives on in the yc profile and directory listings.
- 2024innkeeper · endedinnkeeper — hotel and motel software, listed among the entity's aka names on yc's lingtual profile and third-party trackers. the pivot that matches bhakta's upbringing: he grew up in his parents' motel and built their tech stack at 16. ordering against autoinfra is unverified.
- 2025–2026the essay side-channel · continuingthe non-product essays — 'how intelligence self organizes' and 'we're already post-agi,' submitted under bhaktatejas922 in the weeks around the launch. the founder's intellectual side-channel: company on the api, arguments on the front page.
the company's quests
- 2025–the mcp server · continuingthe mcp server and openai-compatible api — shipped july 2025 as 'mcp for fast ai filesystem edits.' docs carry setup targets for cursor, claude code, windsurf, cline, vs code, and claude desktop — documented targets, not confirmed partnerships.
- 2025–the plugin shelf · activethe first-party plugin shelf — opencode-morph-plugin (mit; fast apply, warpgrep, flashcompact), morph-claude-code-plugin, a pi extension, and an openclaw integration under github.com/morphllm, alongside morph-demos and coding-agent-bench. ~24 public repos.
- 2025–self-hosting · continuingthe self-hosting offer — the models for private deployment, posted in may 2025. the enterprise-privacy answer to 'yet another endpoint' — if it converts.
- 2025–warpgrep · activewarpgrep — the rl code-search subagent, 14b at a claimed ~900 tok/s on b200. 'a 20x faster subagent to grep for code'; the twitter-algorithm demo; v2's swe-bench-pro retrieval claims. self-reported throughout.
- 2026–compact · activecompact / flashcompact — verbatim context compaction at a claimed 33,000 tok/s with a 1m context and 50–70% reduction. the spec also appears as 3,300+ tok/s on another page — the inconsistency is documented.
- 2026–glance · activeglance — ai testing your pr, videos embedded in github, powered by morph-computer-use-v0. launched on show hn february 2026 to 35 points — the best-received post-launch submission.
- 2026–reflexes · activereflexes — multi-head classifiers running on agent traces, a claimed <30ms inference and 'four or 100 reflexes add less than 2ms.' bhakta's tesla data-sampling experience pointed at agent-trace signals.
- 2025inline edit and the tab api · endedthe inline edit model and the tab api — announced at launch (a sub-500ms next-edit predictor in private beta) and never seen publicly again. no release evidence found; likely folded into the suite or quietly dropped.
- 2025the prs that didn't land · endedthe integrations that didn't land — cline pr #3950 closed unmerged in july; roo code issue #7206 closed without implementation. some maintainers prefer built-in search/replace or an in-house apply; the founder-submitted pr doesn't always convert.
the ecosystem
- 2025–the community plugins · continuingthe community shelf — @f97/opencode-morph-fast-apply on npm and a grok-cli integration, both third-party and reported rather than verified.
AI-drafted at Ben Guo's direct request and credited to Hraness; every claim links to its cataloged source.