Ideas
Nine packaged, AI-delivered offers for Equals Five clients that make them go aaaaah. Each one is aimed at a problem a board already argues about — and none of them is AI-first.
Board-level problems first. AI is how we deliver, not what we sell.
Every offer below starts with a sentence an MD or a finance director has already said out loud. The AI is in the delivery — it is what lets a small embedded team do work that used to need a department — but it is never the pitch, never a login, and never a platform we ask them to maintain.
We do not sell software. We sell the answer to the thing they're arguing about.
The shape
- The Strategy & Planning Model — £15–25k, 3–4 weeks, fixed scope. The paid front door: the client's go-to-market plan and a ranked list of what to execute.
- Execution is made of named offers, not a vague retainer — each solving one recognisable problem, starting small, and measured in a number.
- Embedded specialists deliver inside the client team, as they always have. The package describes the work; it does not replace the people doing it.
The Strategy & Planning Model
£15–25k · 3–4 weeks · fixed scope.
- Who buys it: the MD, founder or commercial director of a B2B business with unclear growth priorities or weak pipeline contribution. Not a marketing manager looking for campaign ideas.
- What it does: diagnose the growth position → define the target market and ICP → map the buying committee → sharpen the proposition → find the pipeline and conversion gaps → set objectives → prioritise channels → build budget scenarios → hand over to execution.
- What the client gets: a go-to-market plan they can act on, a ranked execution list, and a model showing what each budget level returns.
- Why it's a product, not a proposal: fixed scope, fixed method, standard discovery, standard templates, standard board-ready output, three price points with real differences between them.
- How it's judged: it converts into retained execution work.
The nine offers
Each one leads with the sentence the client says, not the technology we use. Each has its own page.
Gold Digger Pro
"A year of revenue is sitting dead in our database."
Join the CRM to invoicing to find who actually stopped spending, rank the base by value, and hand their people a ready opener per contact. AI prepares; a person sends — no automated blast, which is the part senior buyers bin.
Status: built, live engagement behind it. Evidence and delivery plan →
AI Visibility
"Buyers vet us through AI now. Are we in the answer?"
Measure share of answer across the engines on their real buyer questions, find where they're absent, fix it with structured, machine-readable content. No ad spend. Compounds.
Status: live client programme with measured data. The problem and the proof →
Ground Truth
"They check us out before they ever contact us, and we have no idea what they find."
Two stages. First, outside-in: inventory every trust signal a buyer can see — reviews, directories, ratings, employee pages — benchmark it against three named competitors, and close the gaps that lose deals. Then, inside-out: mine the whole review corpus for the questions customers actually ask, which becomes the comms programme, and the service failures that keep recurring, which go to operations. Those same third-party sources are what the AI engines now cite, so the work lands in visibility as well as in service.
Status: stage one designed; stage two run on a live corpus — 1,000 reviews labelled, full book ~37,000. The evidence and the worked output →
Profit X-Ray
"Thousands of SKUs, and no idea which ones make money."
A read-only margin map by product, customer and channel. The blended gross margin on the P&L is an average of winners and losers; this shows which is which, and what the loss-makers cost.
Status: designed, not built. The case and the evidence →
The Install
"Everyone here uses AI. Nothing in the P&L has moved."
Most AI implementations in B2B fail, despite everyone using it. We model the workflows as they actually run, pick the best-in-class tools for each, install them with a KPI agreed before we start, and bring the client's own team up to run it. AI mostly exposes the blockers and the gaps that were already there. We start in marketing, then accelerate outward.
Status: the strongest fit with the model — services, no software, no client login. Why implementations fail, and the method →
The Growth Model
"What does my marketing return, and what growth hits my exit number?"
Budget in, growth curve out — and what that curve does to the value of the business. Definitions agreed with the finance director before any modelling, so the output survives a board meeting. It plans the year inside the planning engagement, then becomes the monthly predicted-versus-actual that proves the return.
Status: v1 exists; needs templating before it's sold as standard. The pressure and the method →
The Delivery Multiplier
"We can't grow. Everything routes through three senior heads."
Take the one job a professional-services firm is buried under — proposals, pitch production, reporting — and multiply the output with AI on top of their own proven method, so capacity stops being a headcount decision. The partner approves rather than produces, and the highest-return change is usually bidding less, better.
Status: proven on our own delivery first. The bottleneck and the method →
Agents that don't suck
"We want AI answering customers. We do not want to be the next screenshot."
Almost everyone has met a bad customer-facing agent, and the reasons they are bad are documented and consistent: no source of truth, no scope, no escalation, no testing, and a deflection metric that rewards failure. We build to a standard that fixes all five — internal first, narrow slice second, widened only on the numbers.
Status: standard defined; only built on top of a real knowledge layer. The evidence and the standard →
CompanyOS
"Why does the fifth AI project cost as much as the first?"
The layer that gets built once so everything after it is cheap: one canonical record per customer, product and person; the proposition, claims and pricing rules written in a form machines can use; governed retrieval; and a test set that proves a change was an improvement. The category is real and the components are commodities — what nobody sells is your business modelled into them. That modelling is the work, and it is what Palantir bills at £150k per person per quarter.
Status: the ambitious one — designed, components proven separately, not yet sold as a sequence. Who delivers it, what's in it, what you can do with it →