work / pip · On hold
Pip
A JARVIS-style assistant that ran my day, by voice.
I wanted a secretary I could just call. The moment that made it concrete was driving: that is when I actually reflect, and exactly when I cannot write anything down. Iron Man gets the credit for the shape of it. Then I widened the brief, because my real frustration was that habits, goals, tasks and events each lived in a different app.
Calendar, tasks, habits, goals and notes scattered across apps, and admin that drains the energy real work needs.
A chat and real-time voice assistant sharing one brain and one memory. It reads and writes Google Calendar and Tasks, tracks habits with streaks, links them to goals, takes notes, auto-schedules work around no-go zones, and stages every AI write behind an approval queue. The interface is a command-deck HUD with an audio-reactive voice orb.
- Next.js
- Supabase Postgres
- Drizzle
- Auth.js
- Gemini 3 Flash
- ElevenLabs realtime voice
- Vercel
- One brain for chat and voice: a single tool loop over a single memory, so the two channels never disagree
- The LLM is a config seam: the model was swapped four times (Gemini, Kimi, Groq Llama, back to Gemini) with roughly a one-file change each time
- Voice time-to-first-word cut from about five seconds to under two through streaming, batched reads and warm-up calls
the rest of the build notes(3)
- An approval workflow stages calendar, task and note writes for review, so the assistant earns trust instead of assuming it
- A single header prefix cost weeks, disguised as random voice disconnects: distrust the symptom, read the wire
- On hold deliberately: Google Spark now delivers the same assistant layer inside a Workspace plan I already pay for, and the per-summary API spend could not justify itself against a free-to-me alternative
Knowing when to pause is part of the job. Pip worked and I used it daily, but the summary API cost real money every day and Google Spark shipped the same assistant layer inside a subscription I already hold. Putting it on hold was the right call, not a failed one. The voice architecture went straight into the next build.
A pure-frontend replica with mock data: nothing is saved, nothing leaves the page. The real app is richer; this is the feel of it.
loading the demo…
The thinking
user · pains · gap · tools
Me first: a founder with an ADHD-I brain juggling two businesses, study and eight builds. By extension, anyone whose working memory needs an external system that talks back.
Target user
- Calendar, tasks, habits, goals and notes live in five apps that never talk to each other
- Typing and admin drain the exact energy the real work needs
- Assistants forget: what you told the chat, the voice call never heard
Pain points
Consumer assistants are either chat-only or voice-only, and their memories are split. Nothing affordable runs one brain across both channels with real write access to your life, gated by trust controls.
Gap in the market
- Gemini 3 Flash
- ElevenLabs realtime voice
- Claude Code (build)
- Supabase Postgres
- Next.js
- Web Push
AI + tooling
- Jobs-to-be-done: an external working memory that talks back
- Fogg behaviour model: nudges land at the moment of ability, not just intent
- Habit loop design (cue, routine, reward) with streaks linked to goals
- Progressive trust: staged approvals before the AI writes to your life
- One-brain architecture (single tool loop, single memory) + idempotent tools
Frameworks
- My daily driver through the build period
- Voice time-to-first-word under 2s (from ~5s)
- Model swapped 4x with ~one-file changes
- Demoed to a group of friends: the feedback was consistent, they wanted the event scheduling and planning layer most, because their real pain was a pile of undone tasks and no sense of when to do them
- On hold since Aug 2026: per-summary API cost did not justify it once Google Spark shipped the same assistant layer inside a Workspace plan I already pay for
Success metrics
My own SEO lab: keyword graphs to client-ready reports.