Goals in · finished work out
Mrrlin plans it, coordinates your agents, sends weak output back, and stops at every step that needs a human.
“It cut a week of coordination down to an afternoon. I stopped being the bottleneck between five agent tabs.”
“Free tiers of Gemini and Grok now do the volume work. My paid agents only get the calls that actually need them.”
“I can see which model did what, what got rejected, and why. That transparency is the reason I trust it with client work.”
“More work ships, and it ships better — the review pass catches the things I used to catch at 1am.”
“Roughly half our tasks now close without anyone touching them. I only show up for the decisions.”
Workspace
Plan, execution, review, and the approval inbox live in one place — not across five chat tabs.
Active goals
Approval inbox
Marketing site is ready to publish
Copy, SEO, schema, and mobile checks passed. Two claims were rewritten after review.
6 weak personalization lines rejected
SEO opportunity map saved
Map support pages from ICP
Set up analytics events
Build landing page and demo section
Draft metadata for 6 pages
Check claims against evidence
Reject weak ICP matches
Positioning → homepage messaging
Schema markup verified
Inbox · 4
The launch package passed copy, SEO, schema, and mobile checks. Two claims were rewritten after review. Nothing is published until you approve.
Two positioning routes are plausible and the evidence is split. Mrrlin needs your judgment before the campaign brief is finalized.
Project memory · 24 docs
What changed
Homepage messaging now matches the canonical positioning: goal-based execution, coordinated agents, verification, retries, and human approval.
Review notes
Reviewer agents rejected three vague claims, tightened the CTA path, and confirmed that all publish steps wait for founder approval.
Next suggested goal
Create the first SEO support page and draft outreach to founder-led SaaS teams.
Model routing
Always ask me before
Subscription usage
1w window · resets in 7d · no metered API calls
Connected agents
Chat tools vs Mrrlin
Chat-based AI tools make you the project manager. Mrrlin runs the whole loop — and does it cheaper.
scroll to follow the loop →
Without Mrrlin
paste the context again
one opinion, one pass
you are the only check
The chat ends and the context goes with it. Next time you paste it all again.
With Mrrlin
plan ▸
You
One sentence, in your own words.
Mrrlin · Director
Context is pulled from memory, not re-explained by you.
Gate · review panel
Independent reviewers, rounds until they converge.
no consensus → rerun
Isolated
Its own git worktree. Every artifact recorded as it goes.
Gate · proof required
A live address, a green build, or your own sign-off.
no proof → stays open
Written back
What was decided, and why.
◂ remember
Execution layer
Everything that usually falls on you after the prompt — clarification, routing, review, retries, handoff. Six steps, and twice the work has to prove itself before it moves on.
One plain-language business goal becomes scope, constraints, acceptance checks, and visible tasks.
If context is missing, Mrrlin asks before agents waste time on the wrong work.
Routine work goes to cheaper models; hard or code-heavy work routes to Claude Code, Codex CLI, Sonnet, DeepSeek, or local agents.
Consensus review checks evidence, weak assumptions, brand fit, and acceptance criteria before delivery.
Rejected drafts, thin research, or shaky implementation are sent back with review notes instead of landing in your inbox.
Sensitive claims, sends, publishing, deploys, and final deliverables wait for human approval.
What lands in your inbox
Where to start
Founder-led SaaS
OutputPublished site, SEO checklist, launch plan, reviewed copy
AI-native agency
OutputCampaign brief, content pack, review notes, client-ready assets
Growth lead
OutputExperiment plan, copy, tracking plan, results memo
Technical agency
OutputDelivery plan, implementation tasks, reviewed report
Agent harness
Claude Code, Codex, Sonnet, DeepSeek, or your own local setup. Mrrlin plans, asks clarifying questions, sends important calls to consensus, then routes execution per task.
The state lives in Mrrlin and outlives every session. The model runs on your machine and is yours to swap. Nothing crosses between them without a receipt.
Mrrlin — the state that outlives every session
41 tables · the rules are database constraints, not team habits
Status, dependencies, and how much autonomy each one is allowed.
Specs and evidence, ranked by how current they are — not just by keyword.
The reasoning is stored with the decision, so the next session inherits it.
The proof each finished task produced, attached to the task itself.
Your machine — your models, your keys, your subscription
Mrrlin never holds a model credential and never resells tokens
What you actually meant, pinned down before anything gets built.
Architect, security and scope reviewers — each on its own model — have to converge.
An isolated git worktree, never your checkout. Artifacts saved as it goes.
A separate read — never the executor grading its own work.
A live address, a green build, or your own sign-off.
No receiptThe task stays open. A model saying it finished is not evidence, and the database will not accept it as one.
A resumed run ships a delta, not the whole history — measured on one real task: 641 KB → 25 KB of prompt.
The outside — nothing reaches it that you did not grant
Branches, pull requests, workflows.
Pages that return a real address.
Handoffs that actually arrive.
Read and change, per client.
Fig 0.1
Claude Code reads the repo and docs. The Director writes the plan, acceptance checks, and open questions before work starts.
Fig 0.2
Important decisions go through independent model review. Missing evidence and weak assumptions become specs.
Fig 0.3
Simple work goes to cheaper models. Hard tasks route to Sonnet, DeepSeek, Codex, or specialized agents with visible checkpoints.
Human control
Routine work continues without another prompt
Weak output is rejected and sent back for another pass
Evidence, reviewer notes, and model consensus stay attached
Publishing, sending, and client-facing claims wait for you
Goal
Launch founder-led SaaS outreach without sending risky claims.
Question
Use the competitor comparison angle, or switch to a safer productivity angle?
FAQ
A chatbot responds to prompts. Mrrlin turns a business goal into an AI workflow, plans the tasks, runs the work, remembers context, and keeps iterating until the outcome is ready.
Founder-led teams, early-stage startups, small product and growth teams, and agencies or consultants that need repeatable business workflows.
No. The goal is to work across the tools you already use, not force a migration to a new system.
Human review stays in the loop for final delivery and sensitive decisions, while the system handles the repetitive execution work.
Start with the site, the SEO base, the outreach, or the client work you keep postponing. Mrrlin will map the execution path.
No credit card · No setup · One goal
