Tools
Free interactive tools for one-person businesses, built by Lachlan Coleman-Barrett — an AI systems consultant in Chiang Mai. Each one runs entirely in your browser: no signup, no account, and nothing you enter is submitted or stored anywhere.
The scope is narrow on purpose. These cover the solo operator's own operating system — what to automate, what to run with agents, how to structure what you know — and the part of it most builders avoid, which is selling the thing once it exists. Marketing ROI, ad spend and SEO scoring are deliberately absent: those are covered properly elsewhere rather than duplicated badly here.
Each tool publishes its own method in full on its page. A calculator that hides its weights can't be checked or argued with, and a number you can't interrogate isn't worth having.
The Leverage Audit
How much of your working week could run without you?
Log the hours you spend on 15 recurring tasks across five layers. Get a Leverage Score out of 100, the hours you could reclaim each week and each year, and a ranked list of what to automate first — with the actual approach for each. Every ceiling used in the scoring is published on the page.
Claude Code Skill Finder
What should I install to do X?
A searchable directory over 969 Claude Code skills, catalogued from a 1,045-skill installation and organised by task rather than by name. Nobody has curated a corpus this size publicly, which is the entire reason to build it. Skills referencing my own infrastructure are filtered out by the generator, on a full-body scan rather than a title check.
AI Stack Consolidator
Which of my subscriptions does one AI layer replace?
Paste your SaaS stack and see what a single AI layer genuinely absorbs, what it only partly covers, and what it can't touch. The rule underneath: a model absorbs generation, never proprietary data, distribution or runtime. Downgrades are never counted as savings.
Cold Email Teardown
Why is my cold email getting ignored?
Paste a cold email and get a score out of 100 with the specific problems named — a missing trigger, a buried ask, filler, jargon, length — and the fix for each. Rule-based rather than a model call, so the reasoning is inspectable and the same email always scores the same.
Subject Line Scorer
Is this subject line hurting my open rate?
Scores a cold email subject line as you type against length, spam triggers, vagueness and personalisation. Shows what is costing opens rather than just producing a number.
Follow-Up Sequence Builder
What do I actually say in follow-up three?
Builds a 4-touch follow-up cadence where every message carries a new angle instead of bumping the last one. Pick the scenario, get the spacing and the copy.
Value Proposition Tester
Does my one-liner actually say anything?
Scores a one-line value proposition on the five things that decide whether it lands: a named audience, a concrete outcome, jargon, length and specificity. Each check reports why it failed, not just that it did.
What Should I Charge?
What should I actually be charging?
Turns an income target, realistic billable hours and a positioning multiplier into an hourly rate, a day rate and a project minimum. The multiplier is shown rather than hidden, because it is the assumption doing most of the work.
Objection Response Builder
What do I say when they push back?
Pick a common objection and get an Acknowledge → Reframe → Question response. The structure matters more than the script: it keeps the conversation open instead of ending it with a discount.
Listed as what they are rather than dressed up as available. Nothing below is usable yet.
Second Brain Health Check
Is my note system actually working, or just large?
Point it at an Obsidian vault and get orphan-note percentage, link density, hub concentration and cluster balance — benchmarked against a real 688-node vault. Runs client-side without exception: a vault is the most personal file a person owns and it will never be uploaded anywhere.
Benchmarked against a 688-node vaultHow these are built
- Client-side, always.Nothing you type is submitted, stored, logged or analysed. That isn't a privacy policy, it's an architectural constraint — there is no server to send it to.
- Method published in full.Every weight, ceiling and threshold appears on the page next to the tool that uses it. If you think a number is wrong you can see it, which is the only way a number earns trust.
- Free and ungated.No email wall, no account, no drip sequence. A tool held hostage behind a form isn't a tool, it's a lead magnet wearing one.
- Narrow beats general.Each covers one specific question for one specific reader. A calculator that tries to serve everybody serves nobody, and the internet already has enough of those.
Looking for the systems I built for my own agency — dashboards, audit tools, content pipelines? Those are on the Apps page.