How it works Features Pricing Docs Log in Start free
For founders & freelancers who ship with an AI agent

AI already writes your code. Let it run the work, too.

You're running multiple projects across several ventures or clients — with an AI agent as your dev team, but you're still the one tracking status, juggling priorities, and pretending Thursday has 14 hours. KnightLeap hands the program-management layer to your agent: it drafts the week from your backlog, picks up stories, ships, and keeps every project's status and OKRs current — over a hosted MCP server, against capacity that refuses to lie. You make the calls; the busywork stops being yours.

Connect Claude Code, OpenAI Codex, or any MCP agent in two minutes. Nothing to install.

How it works

Your agent plans, executes, and reports. You make the calls.

1 · It plans the week

Monday morning, KnightLeap's AI drafts the sprint from your backlog up to your real capacity — prioritized by WSJF, capped at three goals per project, across every venture in one pass. You review and commit. One shared sprint per project, enforced by the system, not by discipline.

2 · It executes while it codes

Your coding agent picks up a story, marks it in progress, ships, and sends it to review — over a hosted MCP server, with one token and exactly your permissions. The board updates itself as the work happens, so the plan never drifts from reality.

3 · It reports without you asking

Status, velocity, and OKR roll-ups stay current on their own; your daily brief and retrospective are drafted from your real workspace — retrieved live over SQL and the same API your agent uses, so they reflect what you actually did, not a generic prompt.

Built for the technical, multi-project week

Honest capacity. Real roll-ups. An agent that does the PM busywork.

Portfolios = your ventures or clients

Group projects under each business or client — and keep Family alongside the LLCs, because your week doesn't care about the difference. Roll-ups at every level.

you → portfolios → projects → weeks

Capacity vs. your real hours

Allocate 112 hours across five projects on your 80-hour reality and KnightLeap flags it on Monday — before you promise three clients the same Thursday.

112h / 80h → Overcommitted!

Quarterly objectives

Weeks are for execution; quarters are for direction. Set OKRs per business, scoped to 13 weeks — and watch each weekly sprint roll up into them. No annual goal theater.

quarter → 13 weeks → done

Recurring routines

Your operating cadences as templates: days, start time, checklist, assignee, reviewer. Instances stamp into every week automatically; history stays untouched.

edit once → all future weeks update

Your week on a calendar

Days × hours, only your assigned work, sized by estimate, placed at its start time. The 6:30 school run and the 10:00 client call coexist honestly.

one view · every project · your hours

Tiny-team handoffs

Assignees and reviewers on every story. Your VA executes, you review; your AI agent ships, you approve. Ready-for-review pings the right person.

execution + review, even at n=2

Your AI runs the board — live

Connect Claude Code, OpenAI Codex, or any MCP client with one command — nothing to install. Your agent picks up a story, marks it in progress, ships, and sends it to review, so the sprint tracks itself while the code gets written. One token, your exact permissions, revocable anytime — 46 typed tools plus a REST API under /api/v1.

RAG-like, without the vector store. The in-app AI grounds every brief, story breakdown, and OKR check-in in your live workspace — retrieved over structured SQL and the same MCP/REST tools your agent calls. Real stories, real hours, real history; no generic prompts, no hallucinated status. It drafts; you confirm.

Hosted MCP server · 46 tools · REST API · grounded in your data · read the docs →
$ claude "fix the onboarding bug and ship it"
→ list_my_stories("onboarding") KDP-214 "Onboarding rework · 6h"
→ start_story("KDP-214") · in_progress
… writes code, runs tests, commits …
→ complete_story("KDP-214") · ready_for_review
✓ SaaS Product sprint updated — 6h burned down, reviewer notified
Pricing

Priced per operator, not per seat.

14-day free trial — no credit card. Then one flat price: your agent's work is never metered, and collaborators (VA, client stakeholder, teammate) are free.

Operator

$49 / month

For the solo operator running up to 5 projects — across your own startups or client builds. Unlimited AI-agent activity, every feature, free collaborator seats.

Start free trial
FAQ

Frequently asked questions

Do I need to build with an AI agent?

That's who it's for. KnightLeap is built for founders and freelancers who already drive a coding agent — Claude Code, Codex, Gemini CLI — day to day. The planning stands on its own, but the payoff is letting your agent run the board over MCP. If you don't code with AI yet, this probably isn't for you, and that's okay — we built it opinionated on purpose.

Does the AI just make things up?

No. Its context is retrieved from your live workspace over structured SQL and the same MCP/REST tools your agent uses — RAG-like grounding without a vector database. Every brief, story breakdown, and OKR check-in references your real stories, hours, and history. And it drafts; you confirm before anything ships.

What can the coding-agent integration actually do?

A hosted MCP server exposes 46 typed tools (plus a REST API under /api/v1), authenticated by a personal token that acts as you, with exactly your permissions. Your agent can read the sprint, start and complete stories, manage tasks and acceptance criteria, and check in on OKRs — addressed by human-readable keys like KDP-13 and KDP-2026-W27.

Why weekly sprints?

Quarters are too long to stay urgent and days are too short to move anything. A week is the unit you actually plan your life in — KnightLeap just makes it explicit, per project, across every venture.

Why only 3 goals per sprint?

Because you run five projects. Three goals per project per week is what commitment looks like; ten is a wish list. The constraint is the feature — same reason there's exactly one shared sprint per project.

How is this different from Notion, Linear, or Jira?

Those are blank canvases (or heavyweight trackers) built for teams orbiting one product. KnightLeap is opinionated software for one technical operator across many ventures or clients: enforced weekly sprints, capped goals, capacity that calls your bluff — and an AI agent treated as a first-class participant, not a chatbot bolted on the side.

Can I really mix business and personal?

That's the point. Your Thursday doesn't distinguish between a client deadline and the school run — so your planning tool shouldn't either. Portfolios keep them organized; capacity keeps them honest.

What happens after the trial?

You get 14 days free with no credit card. After that it's a flat $49/mo (Operator, up to 5 projects) or $99/mo (Studio, unlimited) — your agent's activity is never metered and collaborators are free. If you don't subscribe, your account is paused (your data is kept) until you do.

How is my AI API key stored?

The in-app AI is bring-your-own-key (Anthropic, OpenAI, or Google) and optional. Keys are encrypted at rest and shown only once — after that we display just the last four characters, plus a "last used" timestamp. Revoke anytime; scope a dedicated key with a monthly spend cap if you want. Skip the key entirely and the agent/API integration still works.

Get started

Your AI writes the code.
Let it run the work, too.

14 days free, no credit card. Then a flat $49/mo — unlimited agent, no per-seat tax. Connect your agent in two minutes.

Start free trial