Start here
Three things that will become visible
If you are still running AI as a peripheral experiment, 2026 will make that visible quickly. Not through one dramatic moment, but through a steady compounding of the things that matter most to a professional-services firm: win rate, margin, and team capacity.
- Your win rate. Clients already compare turnaround speeds across firms. If a competitor can respond to an RFP faster, with a better-structured first draft, you are not losing on quality — you are losing on pace.
- Your margin. Inefficiency does not hurt once. It compounds across every project cycle, eroding profitability in amounts hard to see individually but significant in aggregate.
- Your team workloads. Manual effort does not scale. The firms absorbing more work without proportionally more people are not larger or better funded. They have standardised the repeatable parts of their delivery.
None of this is because AI is a magic solution. It is because your competitors are systematically responding faster, with less administrative drag, on the same kind of work you are still doing by hand.
The context
What actually changed in 2025
2025 was not a year of marginal model improvements. Four things shifted in ways that matter for consultancy delivery.
- AI became genuinely usable for real consulting inputs. Models extended context windows significantly, reasoned more reliably through complex scenarios, and could interact with tools and systems without falling apart mid-task. That matters because consulting inputs are rarely clean — RFP packs arrive inconsistently structured, project folders are organised differently across teams, transcripts are messy. In 2025, AI could take those rough inputs and produce structured, reviewable outputs. Not perfectly, but reliably enough to be useful.
- AI moved into the tools consultants already use. It stopped being a separate destination. Microsoft 365, Notion, and a growing number of platforms embedded AI directly into email, documents, meetings, and search. Users now expect AI to be present where the work happens, not somewhere they navigate to.
- Larger firms industrialised their AI capability. McKinsey, Accenture, and others did not just pilot AI — they rolled it out at operational scale. Thousands of consultants now have AI accessible as a matter of course. Smaller and mid-size consultancies often underestimate this shift, but the competitive gap it creates is real and growing.
- Governance stopped being optional. For firms handling sensitive client data or working in regulated sectors, AI could no longer be an informal side experiment. Governance and compliance frameworks became a non-negotiable part of delivery capability, not a future concern to address eventually.
The shift
Operationalising AI is the differentiator now
Technology is no longer the differentiator. Successfully operationalising AI within controlled, auditable workflows is.
In the same way that no UK consultancy today would operate without the internet, AI-assisted delivery will become standard practice across the sector.
The question is not whether your firm will use AI. It is whether you will be ahead of that transition or reacting to it.
The year ahead
Five changes coming in 2026
- Scattered pilots will lose credibility. The firms pulling ahead stopped running multiple small experiments and picked a few high-impact workflows to transform end to end. Focused execution that shows up in actual client deliverables will increasingly separate firms that have made AI work from those still tinkering.
- Agentic workflows will become normal business practice. Not science fiction — practical systems that draft proposal packs, assemble client reports, update internal records, and flag outstanding actions, all with clearly defined human review points. The firms building these now will have a compounding advantage over those who wait.
- AI literacy will become part of the job description at all levels. Not coding or technical expertise, but knowing how to supervise AI outputs, validate what the system produces, and apply quality control consistently. That capability needs to be distributed across the team, not concentrated in one or two people.
- Governance will become a commercial advantage. Clients will ask harder, more detailed questions about auditability, transparency, and quality control. How your firm answers those questions will directly affect confidence in the relationship and your perceived maturity as an AI-capable organisation.
- Cost per output will start to matter. As AI usage grows across the organisation, individual decisions about how it is used compound over time. Unit economics — how much it actually costs your firm to produce a given output — will become a meaningful management metric for the first time for many consultancies.
Your role
Where to focus, based on what you own
The following is not a strategy document. It is a practical breakdown of where to focus energy depending on what you own inside your firm.
If you own commercial outcomes, treat AI as a tool for margin protection and senior-time preservation, not a technology initiative for the team to figure out. Identify two or three operational drains that consume disproportionate partner time relative to the value they produce — proposal development and internal coordination are the most common culprits. Fund one tightly controlled pilot that can demonstrate measurable time savings and a clear impact on profitability within a short timeframe. Make governance non-negotiable from the start. Your KPIs should not measure how often AI is used; they should measure partner hours redirected to higher-value activities, and reduction in write-offs and fee adjustments that erode margin. That is your return-on-investment story.
If you run operations, your job is consistency and predictability without introducing friction. First, standardise the sales-to-delivery handover into a single comprehensive pack everyone understands: scope definition, underlying assumptions, pricing rationale, project milestones, and explicit success criteria. Its absence is where scope creep starts and margin leaks. Second, reduce duplicate data entry across systems — manual handoffs between platforms are where time disappears and errors compound. Before you introduce any automation, document and measure the existing process. Unless you know what something costs to do today, you cannot calculate the return on automating it, or justify the expenditure.
The KWA framework applies directly: Knowledge first, then Workflow standardisation, then Agents to automate what has already been proven. Skipping steps is why most operational AI rollouts fail to deliver lasting value.
If you own growth and proposals, speed to quality is what the market will increasingly reward. The ability to produce a high-quality first draft rapidly, without compromising standards, is becoming a commercial differentiator. Build a controlled bid-assembly process that takes an RFP from initial input through to a reviewed, client-ready document. AI can get you to a strong first draft quickly if implemented correctly, but human experts review, refine, and approve every output before it goes to a client — without exception. Firms that get this right will not just win more bids; they will free up the senior time proposal development currently consumes and redirect it into relationship-building and commercial thinking.
If you own delivery and quality, your biggest opportunity is not better documents. It is reducing administrative overhead and getting earlier warning signals when projects drift. Automate routine reporting that consumes significant time but adds limited strategic value. Build systems that surface scope-creep indicators early in a project lifecycle, before they become client conversations, and mechanisms that flag margin drift before problems become difficult to resolve. None of this is possible if time tracking is done manually once a month. Automate data gathering and reconciliation first — that is the foundation that makes everything else measurable.
The controlling idea
Won by workflows, not by tools
2026 will not be won by the firms with the most AI tools. It will be won by firms that turn AI into controlled workflows that protect client trust and improve unit economics.
The 5 Steps for AI Leadership provide the right sequence: Align your firm on a written AI position, Activate your team on the right tools, Amplify the wins that stick, Accelerate deployment to new workflows, and Govern the whole thing properly. The order matters. Skipping Align and jumping to Activate is how you end up with inconsistent usage and no commercial story to tell. Skipping Govern is how you end up with a client incident.