Which AI agent should run your life admin — or your business?
In seven weeks, Grok Bot, Muse and Dots all shipped the same idea: a named agent with its own cloud computer that keeps working after you close your laptop. ◐ partly verified Here’s how to pick one without handing the wrong one your inbox.
Six free tools
Agent Picker
Answer 7 quick questions. Get a ranked shortlist with reasons and caveats.
Side-by-side
Filter and sort 10 agents by surface, price, region and lab compute.
Cost & allowance
Tasks per week × Gartner’s 5–30× agent multiplier vs each plan.
Trust-zone checklist
Score your setup before you hand over your inbox.
Compute → agents
Why the compute map predicts who can keep an agent running.
Timeline
Every launch since Dot shut down, and 2026–30 scenarios (forecasts).
The compute map doesn’t tell you who has the best agent today. It hints at who can afford to keep one running for you, 24/7.
Agent tasks use 5–30× the tokens of a chat reply ✓, and inference is heading for about two-thirds of all AI compute in 2026 ✓. Free tiers, weekly allowances and $100+ plans are how labs share out scarce chips. That’s the brief’s reading, and it’s an opinion.
See the animated explainer →Short answers ◇ forecast
The brief’s segment verdicts are forecasts and opinion, not facts. Use the Picker for your own situation.
Mass consumer default
Muse (free tier, WhatsApp). Gemini Spark is the sleeper: Google owns the most compute.
Knowledge work breadth
Dots: 4,000+ apps and ChatGPT as a base, if safety holds.
Parallel builder teams
Grok Bot: a fleet, not a friend.
Trust & regulated work
Claude and Microsoft Autopilot, by not being consumer companions.
Startups (Instinct, Cue)
Best UX experiments like phone calls and agent wallets, but the weakest compute position.
Before you connect your email
Your inbox can reset most of your passwords. Meta pays up to $130,000 for one zero-click prompt injection against a single Muse user ✓. Score your setup in two minutes.
Run the trust-zone checklist Read the guideGo deeper
Agent Selection Kit
- All 8 SKILL.md templates
- Full 40-question procurement & security questionnaire
- Weighted evaluation scorecard (CSV)
Team / Business Kit
- Everything in the Agent Selection Kit
- Vendor comparison workbook for Dots / Autopilot / Grok Bot / Claude
- Rollout & approvals policy template
Agent Tracker
- Monthly update of prices, allowances and regions
- Changelog of what each vendor changed
- Updated scorecard defaults
Agent selection call
- Your workflows mapped to 1–2 agents
- Trust-zone plan for your accounts
- Written follow-up
Free tools stay free. Prices are proposals awaiting approval.