Tools & utilities
Stuck in a revision loop? There's a tool to get you out.
Write it, check it, distribute it, or set something up once and let it run. Every card below opens to show the tool in action.
Open a category to see its tools.
GrantWriter
GRWTurns a grant submission into a structured, standards-compliant application.
See the demo ›Write it: GRW
A grant application has to answer every funder criterion precisely and still make a compelling case. Holding both in your head while you draft and cross-check takes hours.
- Cross-checks your submission against every funder question and requirement
- Flags every gap and catches data that doesn't add up
- Drafts the application section by section, ready for your review
Here's a fictitious grant application, checked against a real funder's criteria.
BriefWriter
BRWDrafts the brief from what you have, flags what's missing, and recalibrates when things change.
See the demo ›Write it: BRW
You never have all the information when you need to start writing a brief.
- Reads everything you give it, including what's buried in a notes section
- Finds the gaps and carries them into the brief as flags, so nothing gets lost
- Run it again when things change. It recalibrates the whole brief in moments
Here's a fictitious project intake.
Feature Entry
FTETurns tickets into drafted tracker rows, checked against your writing standards.
See the demo ›Write it: FTE
Reads your tickets, creates tracker rows with draft text, and applies writing standards. It loops until you approve, so nothing goes out without your sign-off.
Five fictitious tickets for a fictitious visitor management product.
You Said Yes
YSYTurns a commitment into a starting point, with continuity between sessions.
See the demo ›Write it: YSY
You committed to something and now you're staring at a blank page.
- Interviews you to draw out what you know, what you've promised, and the constraints you're working with
- Turns the conversation directly into finished outputs, not just notes
Here's a fictitious speaking commitment. Watch You Said Yes turn it into a poster and speaker notes.
AI at Home
Are you ready?
ClearStory
CLSTurns a messy conversation thread into one neutral, structured record.
See the demo ›Write it: CLS
Messages spread across platforms. Reading them and stitching them together mentally eats time you don't have.
- Reads a thread and returns a fixed structure every time: what's happened, where each person stands, what's still open
- Works to a defined framework, not a mood, and strips out flattery and insults equally
Here's a fictitious WhatsApp thread from a community garden committee: six people, three tangled topics.
Document Health Check
DHCFlags every rule break, inconsistency, and typo. Nothing gets through unnoticed.
See the demo ›Check it: DHC
A generic AI check will catch typos, but it won't catch a broken cross-reference or a vague step.
- Runs against your rules: structure, clarity, consistency, instruction quality
- Returns a punch list, one pass, no misses
Below is a fictional user guide with deliberate errors: some obvious, most not.
Fidelity Checker
FDCTraces every claim back to source and flags what can't be verified.
See the demo ›Check it: FDC
A generic AI check reads for tone. It doesn't check whether the draft still says what the source said.
- Traces every claim in a draft back to its source, sentence by sentence
- Flags what drifts, overstates, invents, or goes missing before release
Below are a source article and a draft. The draft looks close, but it isn't.
RedCheck
RDCTurns reviewer comments into one structured, attributed punch list.
See the demo ›Check it: RDC
Reviewer comments come back scattered across emails, tracked changes, and sticky notes. Nothing tells you what's actually been fixed.
- Reads annotations from one or more marked-up documents
- Attributes every comment to its reviewer, turns it into one action
- Verifies each correction and gives it a verdict. Runs in cycles until everything is fixed
Below is a document annotated by two reviewers, Sam and Susan: 13 comments between them, corrections made, one round done.
Release Packager
RLPAssembles approved copy into release-ready output, section by section.
See the demo ›Distribute it: RLP
Takes approved tracker rows and assembles the release output: section updates, datasheet entries, and release notes, ready to place into documents.
Five approved tracker entries from a fictitious visitor management release, ready for packaging.
| Feature | Ticket | Type | Status |
|---|---|---|---|
| Visitor pre-registration | VM-1042 | New feature | Approved |
| Configurable badge expiry warning | VM-1058 | Enhancement | Approved |
| Recurring visit schedules | VM-1063 | New feature | Approved |
| Bulk visitor export: CSV | VM-1071 | Enhancement | Approved |
| Check-out time not recorded on manual sign-out | VM-1089 | Fix | Approved |
Social Media Creator
SMCTurns one brief into on-brand posts for up to five platforms.
See the demo ›Distribute it: SMC
You know what you want to say. Every platform wants it in a different shape.
- Writes from your Org Profile, so every post already matches your tone and style
- Checks against what you've already posted, so nothing repeats
Here's a fictitious brief. Set your brand colours and see Social Media Creator draft the post.
ClassifyMe
CLMClassifies bank and card transactions into a reconciled income and expenditure report.
See the demo ›Utilities: ClassifyMe
The problem
Consolidating data from separate sources is messy no matter how careful you are. Every source has its own format and its own quirks, and getting them lined up accurately, period after period, is a job in itself.
The workaround
I use Power Query, and I'm good at it. But every structural change on the source side, a column moved, a file renamed, or new location means another tweak to the query; another half-hour spent on checking and testing the code. Catering for various changes in working practices or source data requires real engineering capabilities, not spreadsheet skill. And not to put too fine a point on it, they're skills and experience I don't have.
How I built it
That's exactly the gap the LLM fills. I know what the data means and what I need to see in it. Claude runs on an LLM that can pull data from different sources and build something that doesn't break the same way twice. I send Claude my disparate files monthly and I get this report.
What matters, at a glance
The report is styled so totals and variances stand out immediately, instead of being buried in rows I'd have to check line by line.
Profiler
ORGCaptures your tone, writing standards and brand guide once, so every tool applies them automatically.
See the demo ›Utilities: Profiler
The problem
Teams that write need to sound like each other and use the same writing style and vocabulary. Templates can help frame the look and feel of a document, but framing the actual content is difficult. The problem compounds as team members change and the shared history of how the writing style was conceived fades. Add to that multiple hands touching a document, and it isn't long before coherence is lost, requiring more rework.
The workaround
Writing guides, training sessions, and peer reviews all have their place, but the results are not outstanding. LLMs help, but only up to a point. I can set the guidelines for a document and a colleague can paste in the same prompt, but by the time it's time to edit, those guidelines are usually lost in a chat somewhere. They get harder to find as time passes.
How I built it
That's why Profiler exists. It's a single profile that stores the writing patterns, specific vocabulary, and grammatical dos and don'ts you upload. Whoever uses that profile to edit documents or create new ones applies the same patterns and guidelines automatically. It also stores your brand colours, so every output reflects you.
What matters, at a glance
Set up once, applied everywhere. Users never have to repeat the patterns and guides whenever they use a tool. The profile is available to every tool without extra setup, so it's built-in consistency no matter what you're creating, reviewing, or editing.
Integration-setup-guide.docx
Configure the connector before you enable sync.
Enable sync1
Enable sync once configuration finishes.2 Data updates every fifteen minutes. No manual refresh is required.3
Troubleshooting1
Check your connection first if sync stops. Restart the integration if the issue continues.
once
One file in your Drive. Every tool fetches it by Org ID before it runs.
Tone: direct, instructional, present tense
Vocabulary: "enable," "configure," never "turn on," "set up"
1Lone subheadings: heading stands alone, never folded into the sentence below
2Section purpose: opening line states the action the section enables
3No filler language: no "simply," "just," "easily" or similar
Sentence structure: one idea per sentence, never joined with a comma and an "-ing" word
Brand colours:
FindMe
FMESearch, rename or move files as singles or in bulk; plus a safety bucket before anything's deleted.
See the demo ›Utilities: FindMe
The problem
It's not about being organized. You can start that way. But years pass, files pile up across folders, naming drifts, and even a tidy system slowly stops being one. Going back to find something, or check whether you already have it somewhere else, gets harder every year, not easier.
The workaround
I patched it together with a combination of different utilities. It worked, but it wasn't slick. I was always hunting for "the" tool that would actually sort me out, never quite finding one that did the whole job.
How I built it
That's the gap the LLM fills. I could learn PowerShell, or local listener handling, or hash-based duplicate detection. But knowing enough to get by isn't the same as knowing enough to build something robust, that handles edge cases, that doesn't quietly break. I don't have that level, and I don't need to. The LLM does. I supplied what no AI can: which files actually matter to me, what counts as a duplicate worth flagging, and which folder a category belongs in.
What matters, at a glance
The LLM also built the report to work the way I actually look at things: colour-coded flags, panels grouped by what needs attention, so I see what matters before I've read a single filename.
Try it below: search, rename or move files
SafeShare
SHRShares a working view of your data without exposing the source file behind it.
See the demo ›Utilities: SafeShare
The problem
Giving sharees viewer access to a live Power BI report means paying for premium per-seat licensing, and the costs can mount up.
The workaround
I lived with it for years because I didn't see another way for others to view the data without exposing it to the internet or sharing the source file. Neither was a safe option, and my internet searches for alternatives yielded nothing.
How I built it
I asked Claude for a way to share the finished report without the Power BI Service licensing at all. It gave me a scheduled, one-way file sync inside my own Drive: robocopy copies the finished file from a source folder into a separate folder that the sharee has Viewer-only access to. If they change anything on their end, the next sync overwrites it with the source version.
What matters, at a glance
Sharees only ever have Viewer access to a separate folder in your Drive. They never get access to your source folder, and anything they do to their own copy doesn't last past the next sync.
scheduled copy
access
In short
Speed and trust. Not one or the other.
Every tool thinks with you and for you, but never instead of you. It weighs up the options and flags what needs your judgment. You decide what's best.
Questions
Get in touch.
Not sure which tool fits your situation? Tell me what you're working on.