What is Construct Computer?
Construct Computer is an AI employee for solo founders and small teams. It works from its own cloud computer with files, memory, a browser, and a terminal, and connects to Gmail, Notion, HubSpot, and more apps. It handles research, email, and scheduled workflows on its own while you focus elsewhere.
Top Features:
- Reusable workflows: encode a process once, then schedule and rerun it anytime.
- App builder: describe a tool and it writes and publishes a working app.
- Inspectable memory: review, correct, or delete anything the agent remembers about you.
Use Cases:
- Inbox triage: the agent reads new mail and drafts replies for you.
- Market research: gather sources on a competitor and receive a cited report.
- Recurring operations: schedule weekly reports or CRM updates that run without reminders.
Who Can Use Construct Computer?
- Solo founders: hand off research, email, and admin work before a first hire.
- Startup teams: share one AI employee and files with roles and permissions.
- Business owners: automate repeat tasks across Gmail, Calendar, and HubSpot without coding.
Pricing
- Lite ($9 per month): base model usage, two agents, and three scheduled tasks included.
- Starter ($59 per month): six times Lite usage, five agents, and ten scheduled tasks.
- Pro ($299 per month): fifteen agents, 1,000-step runs, your own keys, and a 7-day trial.
Pros and Cons
Pros:
- Persistent workspace: files, memory, and schedules persist so work continues between sessions.
- Clear oversight: an audit log records every action, and you can pause runs.
- Many channels: message it from the web, Slack, Telegram, Discord, or email.
Cons:
- No free plan: only Pro has a trial, and plans start at nine dollars.
- Steep top tier: the jump from Starter to Pro is large for tiny teams.
- Usage limits: steps, scheduled tasks, and runtime are capped on lower plans.
FAQs:
1) How is it different from ChatGPT?
It keeps files, memory, and schedules, and runs jobs on its own.
2) Which apps does it connect to?
Gmail, Calendar, Notion, Linear, Jira, GitHub, HubSpot, Airtable, and more.
3) Does it train on my data?
No, your workspace is never training data, and you own the output.
4) What if a job fails midway?
Finished work is saved as it happens, so reruns continue from there.
5) Is there an enterprise plan?
Yes, Enterprise adds SSO, private deployment, and custom MCP, scoped on a sales call.