Gavalytics / Gnome@Chief @Dev help prepare a landing page for a new product.
Gavalytics / Gnome
Meet Gnome. Your AI teammates.
A little help for the work ahead. Chat, build, and tackle everyday tasks with AI teammates, each with a role and workspace of their own.
Product concept · Your choice of models · You approve important actions
Say hello to the workspace · Interactive demo
GnomeLocal demo
Website Launch
3 BOTs · Local demo
Local demo · No AI, accounts, or tools connected.
Meet your team, switch conversations, and explore their workspaces.
All local demo. No AI, authentication, terminal execution, or network actions.
A teammate for each kind of task Everyone has a role.
Everyone has a role.
You bring them together.
Give each bot a role, instructions, a model, and the tools it needs. Let one focus on research while another writes code, then bring them together when it is time to share the work.
- Engineering
- QA
- Research
- Marketing
- Customer Support
- Finance
- Chief of Staff
- Personal Assistant
Pick up where you left off.
Keep your go-to bots pinned, find an earlier conversation, or mark a chat to come back to. Copy a setup that works, hide a bot for later, or delete one you no longer need.
One release. Three specialists.
See how the work comes together.
One teammate prepares the release checklist. Another tests it and gathers evidence. A reviewer brings the findings back to you, so you can decide what comes next.
Engineering → QA → Review → You
The right tools for each teammate.
Bring the AI you like working with
Choose an AI provider, an agent CLI, or a supported local model for each bot. Match the model to the task instead of asking one engine to do everything.
Examples include Claude, Codex, Grok, Cursor, Kimi, OpenCode, and Qwen. What works depends on your engine and account setup. Provider terms and charges still apply.
Help that goes beyond a reply
Computer use lets agents view screens, click, type, and work in apps. Choose your own machine or an isolated remote computer to give the task a place to happen.
Choose which folders and apps a bot can use. Keep unfamiliar tasks in an isolated environment, and require approval before destructive changes or commitments outside your workspace.
Keep your familiar tools close
Bring email, chat, code repositories, documents, project boards, spreadsheets, CRM systems, and internal tools into the workflow. Approve each connection and share it only with the bots that need it.
The examples here use Gmail, Slack, GitHub, Notion, Linear, and Google Sheets. Check which connectors are available, what access they need, and any account requirements as you set up.
A little less busywork
What would you like a hand with?
Pick a task that comes around often. Decide when it should run, which tools it needs, and where you want a person to check the result. These eight ideas are workflow examples, not live automations on this site.
01Every weekday
Inbox automation
Review incoming email, separate important messages from noise, and draft routine replies. Leave sensitive conversations for a person.
Tools: Gmail
02On schedule
Invoice follow-up
Find overdue invoices and prepare personalized reminders. Ask for approval, send only the approved message, and track the next follow-up.
Tools: Email · Billing system
03On every deployment
Release QA
Open the latest build like a customer, test important flows, capture screenshots, and report reproducible issues.
Tools: GitHub · Browser · Computer
04On every pull request
Pull request review
Review changes, identify meaningful issues, and summarize risks before a human makes the merge decision.
Tools: GitHub
05Every week
Competitor monitoring
Track selected competitors’ pricing, products, releases, and public updates. Turn meaningful changes into a concise, sourced report.
Tools: Web · Google Sheets
06Every Monday
Weekly business report
Collect revenue, refunds, advertising spend, and other business metrics into a consistent report ready for the weekly meeting.
Tools: Billing platform · Google Sheets
07Every 30 minutes
Support triage
Read incoming requests, prepare answers to common questions, identify technical issues, and route tickets with clear ownership under approved access rules.
Tools: Gmail · Linear
08When requested
Personal bookings
Search available appointments and compare suitable options. Ask for confirmation before completing a booking.
Tools: Browser
At home on your computerYour agents, your machine.
Your agents, your machine.
Know what stays and what leaves.
Gnome brings your agents and their activity into one workspace on your computer. Set them up, follow their progress, and talk to them without jumping between separate apps for each AI engine.
The local-first design keeps workspace conversations, credentials, and agent activity on your computer. Your AI accounts and subscriptions stay with their providers. Gnome brings them together, not replaces them.
Local-first does not mean local-only.
When you use a remote model, your prompts and relevant context go to that provider. Connected tools exchange data with their services, and remote computers handle data too. Before sharing confidential information, check what leaves your machine, who can access it, and how long providers keep it.
You set the limits. You give the go-ahead.
Give each bot only the access it needs. Require a person to approve sending messages, publishing, merging, deleting, paying, or booking. Check evidence and important claims, keep credentials safe, and decide when stored data should be deleted.
Find your lineup
Start with one teammate. Add help as you go.
Give your team a shared goal, decide who passes work to whom, and choose a person to review the result. Here are a few lineups to make your own.
Engineering team
Planner → Engineer → Reviewer → QA
Move from a development brief through implementation, review, testing, and release verification.
Connected tools: GitHub
Plan a team like thisSEO growth team
Researcher → Content strategist → Analyst
Find search opportunities, identify technical improvements, create briefs, and organize measurable actions.
Connected tools: Search Console · Analytics · Google Sheets
Plan a team like thisCompetitor watch
Monitor → Researcher → Brief writer
Watch the competitors you choose and turn meaningful changes into short business briefs.
Connected tools: Web · Google Sheets
Plan a team like thisLead research team
Researcher → Qualifier → Coordinator
Find relevant conversations and opportunities, explain why they matter, and prepare the next action for review.
Connected tools: Web · Spreadsheets
Plan a team like thisMarketing team
Strategist → Writer → Performance analyst
Research positioning, create campaign assets, and use performance data to recommend what comes next. Approve content before publication.
Connected tools: Slack · Google Sheets
Plan a team like thisA team in one Markdown fileFound a way that works?
Found a way that works?
Keep it in a playbook.
Save your team setup in a reusable Markdown file: agents, roles, instructions, who works with whom, shared rooms, tools, routines, and first tasks. Next time, you have a starting point instead of a blank page.
- Import the playbook into the agent workspace.
- Review roles, instructions, folders, and routines.
- Approve the connections and each agent’s access.
- Give the team a small first task with clear limits.
A playbook is a plan to review, not automatic permission to act. Read unfamiliar instructions and check tool access before you run anything.
Walk through the setup# Release readiness team
## Agents
- Engineer: prepare changes in the working folder.
- Reviewer: inspect the diff and summarize risks.
- QA: test key flows and capture evidence.
## Room
Release readiness: share findings and handoffs.
## Tools and access
GitHub: selected repository only.
Browser: staging environment only.
## Routine
After deployment, run the release checklist.
## Initial task
Review the latest staging build and report blockers.
## Approval
Ask before posting externally, merging, or deploying.Built around the way you work
For the things on your to-do list.
Developers
Get a hand with code, reviews, tests, release checks, docs, and issues. Keep a person in charge of review and release decisions.
Business teams
Spend less time on routine reports, support, follow-ups, research, and admin, using the tools your team already knows.
Founders
Give research, preparation, and monitoring a place in a small AI team, without explaining everything again in a new chat.
Power users
Make bots your own, connect your tools, and plan recurring tasks. Save the workflows you like as team playbooks.
Start small. Get to know your team.
Think of this as a checklist for getting ready in the agent app. This page is just a preview: there is no download, account connection, or installation here.
- 01
Prepare your workspace
In the agent application, choose the machine and working folders your team may use. This website does not install or run agents.
- 02
Choose your AI
Set up your provider account, agent CLI, or supported local model. Check costs and how your data is handled before connecting.
- 03
Create your first bot
Give it a name, role, instructions, model, working folder, and only the tools it needs.
- 04
Try one small task
Start with a draft or a task that only reads information. Check the evidence and approve important actions before they run.
- 05
Grow into a team
Add a specialist when you need one. Decide how bots pass work along, then add recurring tasks once you trust the workflow.
Stop starting from a blank chat.
Choose a workflow, give it an owner, and build a team around it.