Industries
You automate everyone else’s business but your own
Consulting firms run on partner hours and past work. Both get spent rebuilding things the firm already has, somewhere, in a folder.
What we build here
- A searchable memory of every engagement
- Proposal drafts from your own past work
- Research synthesis that shows its sources
- Utilisation, pipeline and bench in one live view
Boutique consultancies where knowledge walks out the door and proposals eat partner time. We build the internal systems you’d build for a client if you had the hours.
- less partner time typically spent on a first proposal draft
50–60%
less partner time typically spent on a first proposal draft
- of utilisation typically recovered by seeing bench risk weeks earlier
5–10 pts
of utilisation typically recovered by seeing bench risk weeks earlier
- typical time to answer “have we done this before?” (down from days of asking around)
1 hr
typical time to answer “have we done this before?” (down from days of asking around)
Ranges from engagements and process maps in this sector. Your numbers are your numbers. We measure them before we quote, not after.
The work
Where the week actually goes.
None of this is exotic. It’s the ordinary friction that never makes it onto anyone’s priority list because it’s everyone’s job and no one’s project.
- 01
Every proposal starts with “who has done something like this?”
A partner posts in Slack. Three people answer with half-remembered engagements. Someone finds a deck from two years ago that’s close enough. The proposal takes eleven partner hours to write, and most of them go on rediscovering work the firm already did.
- 02
The firm’s knowledge leaves with the senior who wrote it
Every engagement produces research, models and frameworks worth reusing. They live in one consultant’s Drive folder, named “v7 final actual”. When they leave, it’s gone — and you’re paying to rebuild your own thinking every year or two.
- 03
Utilisation numbers that arrive too late to act on
Time entry is late, the roll-up is a spreadsheet, and the picture arrives mid-month. By the time you can see someone was at 40% for three weeks, the three weeks are gone. Pipeline and bench are tracked in different places by different people who disagree.
- 04
Deliverable assembly done by the most expensive person available
The night before a readout, a senior manager is formatting slides, chasing a number and rewriting an exec summary. The thinking was done days ago. What’s left is assembly, and it’s being done at $400 an hour because there’s no one else at 11pm.
The fix
What we build, and what it replaces.
Every one of these runs in your accounts, against your real data, with a named person owning the exceptions. Nothing here needs your team to learn a new tool.
A searchable memory of every engagement
We index proposals, decks, models and research into a private, permissioned library over SharePoint or Drive. Ask it in plain English (“what have we done in mid-market distribution”) and you get the actual engagements, the people who ran them, and what was reusable. Claude answers with citations to real files, so nothing is taken on faith.
Proposal drafts from your own past work
Paste the brief. The system finds relevant past engagements, drafts the approach, team and pricing structure from your real precedent, and flags where you have no comparable work. A partner edits. The first version takes an hour instead of a day, and it’s built on what the firm actually did.
Research synthesis that shows its sources
Interview transcripts, filings and market data get summarised into a working brief where every claim links back to where it came from. Your consultant starts from a structured draft instead of forty tabs. Nothing is asserted without a source a client could check.
Utilisation, pipeline and bench in one live view
n8n pulls time entries, HubSpot pipeline and staffing into one Postgres-backed view refreshed nightly. Bench risk and over-allocation flag before they cost you a month. Partners see the same numbers, which ends most of the argument.
How it’s delivered
The services behind this work.
- Custom AI SolutionsAI built on your own documents, your own history, and your own rules, so it gives real answers instead of confident guesses.
- AI ConsultingA written plan that says which three things to automate first, what each one costs, and what you get back.
- Business AutomationYour systems talk to each other. Quotes, invoices, and follow-ups happen without anyone retyping anything.
Questions
What people in this sector ask first.
Client work is confidential. We can’t pool it into one searchable index.
Permissions are built in from the start. The library respects your existing access rules: engagement walls hold, restricted work stays restricted, and searches only return what the person asking is already allowed to open. Client-identifying material can be excluded or redacted at index time. We agree the rules before indexing a single file.
Our IP is our thinking. Does using AI on it give it away?
We configure so your content isn’t used to train anyone’s model, and the index lives in your tenancy. Your frameworks stay yours. If a workflow would require sending IP somewhere you’re not comfortable with, we say so instead of quietly doing it.
We’re 30 people. Doesn’t this only make sense at scale?
Thirty people is where knowledge loss hurts most, because you have enough past work to be worth searching and nobody whose job is curating it. Larger firms have a knowledge management team. You have a partner doing it at midnight.
Won’t drafting proposals from past work make them generic?
It would if the draft went out as written, and it doesn’t. The system does the recall, finding the right precedent and structure, and your partner does the judgement. The generic version is what happens today when someone copies the nearest old deck at 9pm.
Let’s put a number on it.
Thirty minutes. We’ll ask where your week goes, tell you which of the above we’d automate first, and what it’s worth. You leave with a written summary either way.