DEEPGRAIN · AN OPERATING CONSULTANCY BUILT FOR THE AI ERA.
Read. Craft. Scale. In that order, always.
The strategic, the functional and the individual move together, or the work stalls at a pilot like every other AI rollout you've watched.
Walk one workflow through it with me.
Thirty minutes. Read, atomise, agree the first move.
The method
Three levels. You work at all three at once.
A function only becomes resilient when the strategic, the functional and the individual are moving together. Most programmes touch one, lean on another, and ignore the third.
Where the function is going.
What does the function look like in three years if AI keeps moving at this rate?
Strategy that ignores the technology shift, or hands it to a vendor.
Leadership owns the operating direction. AI is a lens on the plan, not a side track.
Where the work is done.
Which workflows compound, which leak, and which are theatre?
Tools pushed at roles. Pilots that never reach production.
Workflows mapped at the click level. Agents wired to the friction, not the title.
Where the craft lives.
Which people can build, and what stops them?
Mandatory training. Compliance metrics. No one shipping anything.
Champions inside the function. Air cover, time, and a small starting brief.
Worked example · a real people workflow
Mapped, then rebuilt. In that order.
⚖ The judgement step stayed human, by choice. Reclaimed time only counts if redeployed, not re-absorbed.
Map one workflow with me.
Thirty minutes. Leave with one move for Monday.
01 Read
Before we touch a thing, we understand.
Most consulting starts with the org chart. We start somewhere else. The org chart is rarely how decisions actually get made, and the strategy deck is rarely what drives the outcome.
Underneath all of it is a grain. The real pattern of how this organisation moves and decides. We read it first. We talk to the people doing the work. We watch where energy flows and where it stalls. We find the fractures forming before anyone has named them. Only then do we build.
“If you can't name where the work is actually getting stuck, every tool you buy will land in the wrong place.”
02 Craft
We build with the grain, not against it.
Human judgement. Machine precision. In the room together from day one. Agents that remove friction without removing thought. Systems designed around how this specific place actually works, not borrowed from a playbook that worked somewhere else.
The carpenter knows the wood will split if you cut against the grain. So do we. We build alongside your people, in your tools, with your context. Then we hand it over.
The partnership
Agents that partner. People who grow.
Agents take the repeatable, low-judgement work. Your champions learn to design, run, and extend them. The capability stays in the team, not in a vendor. See how enablement works →
It's a training programme first. The systems ship as a side effect, and the hours we reclaim go straight back to your people, for the work only they can do.
“The leaders who get this right don't lead the rollout. They lead the conditions that make the rollout inevitable.”
Build or hire
One mid-level hire. Or one well-built agent.
Same shape of work. Same expected output. Costed honestly: loaded salary on one side, build plus run on the other. The gap widens every year you keep the agent running.
New hire
Loaded annual costAgent
Amortised build + runSaving
Over 3 years, against the same output.
Ratio
The hire costs that many times what the agent does, at this horizon.
Indicative only. Hire assumes a mid-level London Ops salary (~£80k base) loaded with NI, pension, tooling and amortised onboarding. Agent assumes an £18k build amortised across three years, plus £12k/year for tokens, monitoring and periodic improvement. The agent removes a hire that would otherwise have been needed for the same workload.
03 Scale
We leave something that compounds.
Not a deck. A genuine capability. Teams who think well with AI. Structures that hold as you scale.
Months after we leave, the champions are still building. They have extended the work into places we never touched. That is the test: what you still have, and what you have added, long after the invoices stop. The guardrails that keep it trustworthy live with the team too, through governance designed for the people doing the work, not bolted on after the fact.
“You don't need engineers to build AI capability inside the function. You need three or four champions, given air cover and time.”
A directional model
What might this be worth in your function?
This is a partnership model, not a replacement one. Agents take the repeatable work; your people are coached to design, run, and extend them. Move the sliders to see what that shift could be worth in your function.
People in the function
Per person, on coordination & repeatable work
Salary plus on-costs, per person
Reclaimed hours go back to your people for higher-judgment work. This is a partnership model. Assumes ~60% of identified low-judgment hours are genuinely recoverable across a 48-week year. A directional model, not a quote. The actual number comes out of the diagnostic, usually within ten percent of this.
Common questions
What clients ask before they engage.
- How long does a Deepgrain engagement run?
- Most engagements run between three and nine months. We start with a 30-day Read phase to surface the operating reality, then move into Craft (typically 60–120 days of focused interventions paired with champion development), and finally a Scale phase that hands the practice over to your team. Some clients renew into a lighter advisory cadence after that.
- What does the first 30 days actually look like?
- The Read phase. Matt sits inside your operating cadence: standups, one-to-ones, leadership reviews. He runs structured interviews across the org. The output is a written diagnostic: where the operating story diverges from the operating reality, which interventions would compound, and which would break the grain. No slideware, no benchmarks. A document leadership can act on.
- What deliverables do we walk away with?
- Three things. First, the diagnostic document from the Read phase. Second, the interventions themselves: usually a small set of agentic systems and operating rituals built and shipped during Craft. Third, three or four trained champions inside the team who can extend, debug, and govern the work after we leave. The capability stays with you, not in a vendor.
- Who is the right fit for this work?
- Founders and operating leaders inside organisations worth getting right: typically AI-native, defence tech, financial data, transit and mobility, or climate. The common thread: leadership willing to look at the operating reality honestly, and a team capable of holding the practice once we hand it over.
- Do I need a technical team to make this work?
- No. The champion model is built around non-engineers: heads of People, ops leads, chiefs of staff, domain operators. They learn to design and run agents inside their own function. We bring the engineering muscle when something needs to be built deeper, but the day-to-day capability lives with operators, not coders.
- How does pricing work?
- Engagements are scoped per phase, not by retainer or day rate. If you want a fixed number before that conversation, start with the Grain Audit: two weeks, £2,000, one process mapped end to end. It's credited in full against a programme if you go further.
The Grain Audit
Don't take the method on faith. Run it on one process.
The Grain Audit maps one People Ops process end to end, ranks the highest-return automations, and hands you a 90-day plan you keep whether or not we work together. Two weeks. £2,000, credited in full against a programme. Three slots a month.