For founders and small teams
Merlin is an AI project director. It turns your goal into a task plan, runs AI agents to do the work, checks their output, and brings you the result to approve. You own the outcome — not the project management.
How Merlin compares
Chat-based AI tools make you the project manager. Merlin runs the whole loop — and does it cheaper.
Full context re-sent with every prompt
Progressive context compression — the fewest tokens per task
Forgets between sessions — you re-explain and re-instruct
Memory is captured continuously — context and instruct
One long chat thread, no visible progress
Spec-driven: epics, tasks, and a board — track progress at a glance
Blocks the chat with one question at a time
Async inbox — answer questions across dozens of tasks when it suits you
One expensive model for everything
A pool of models — the Director puts cheap models on simple tasks
Cheap by design
Merlin isn't another AI model — it's the harness on top of the agents and models you already have. Each task goes to the cheapest model that can do it well, and progressive context compression keeps every call lean.
AI agent harness
Treat Merlin as an AI agent harness: it gives agents a goal, the right context and memory, a contained worktree or run, checkpoints, review notes, handoff status, approval gates, and deploy-control decisions before sensitive changes ship.
A day with Merlin
Write 20 LinkedIn posts from our positioning and schedule the first week.
In plain language. Attach context — docs, links, your product. Merlin remembers it for every future goal, so you never re-explain.
The Director breaks the goal into tasks, puts AI agents to work, checks their output, and sends weak results back — without you prompting anything.
20 posts ready · 1-week schedule attached
Reviewed twice. 4 posts rewritten for tone.Finished work and important decisions land in your inbox. Approve it, or send it back with one note. Nothing ships without you.
Use cases
Real goals founders give Merlin — each becomes a plan, agent work, and a reviewed result.
→ Reviewed lead list + outreach drafts + follow-up sequence
→ Posts reviewed for voice, with a publishing schedule
→ Competitor teardown + gap analysis + recommended angle
→ Clustered insights + prioritized spec, ready to build
→ Task breakdown + reviewed pull requests
→ Drafted report with data pulled from your project history
Your control
You decide how much Merlin can prepare on its own — per task. Routine draft, research, and verification work can run without extra prompting; anything sensitive waits for your approval. Every decision and every output stays visible in the task history.
FAQ
ChatGPT answers one prompt and forgets. Merlin takes a goal, plans it into tasks, runs agents across multiple steps, remembers your project context, and iterates until the result is ready for approval.
No. Merlin works alongside what you already use. Connect context from your docs, repo, or notes — no migration required.
Never by default. Final delivery and sensitive actions always pass through your approval. You choose which routine draft, research, or verification tasks can run without extra prompting.
Founders, solopreneurs, consultants, and small teams who need marketing, research, ops, and product work done — without becoming a full-time AI project manager.
Write the goal, attach your context, and watch Merlin turn it into a plan, agent work, and a result waiting for your approval.
No credit card. No setup. One goal.