The question I get asked most often about AI in consulting is whether it will replace consultants. I think that is the wrong question. The more useful one, for anyone whose work is operational rather than advisory, is what AI frees a consultant to do better. For me, the honest answer is direct. AI handles more of the screen-bound work so I can stay in the field longer, run more projects on the ground, and put my hours where they actually move things.
That framing sits at a different angle to most of the current AI consulting story. The Big Four have collectively committed more than $10 billion to AI initiatives since 2023, with Anthropic and OpenAI partnerships embedded across their platforms, and dedicated forward-deployed engineering teams placed inside large client organisations. That is one workable model for very large clients. The Deloitte report refunded by the Australian government after fabricated references and a hallucinated court quote were discovered is the cautionary footnote. It is also not the model I am interested in building, and probably not the one most operational sustainability work actually needs.
This article is about a different shape. An AI-augmented consultant in the operational sense. Not a tech vendor with a sustainability sticker on it. Not a strategy advisor with a chatbot. Someone whose practice is field work, audits, capacity building, and on-the-ground project delivery, with AI quietly compressing the desk work behind the scenes. Some call this being a freelancer with superpowers. The phrase fits.
Key Takeaways
- AI is plumbing, not product. Used well, it disappears into the workflow. The client sees better, faster, more rigorous work, with the methodology available if they want to inherit it.
- The valuable hours are operational. Walking the site, training the team, building stakeholder networks, supervising actual interventions. AI does not help much with those, and it does not need to.
- Where AI does help is the back office. Research synthesis, market analysis, reporting, fundraising drafts, business modelling, data analysis. The work that used to consume thirty to fifty percent of an operational consultant’s time.
- Capacity transfer is half the engagement. The skill travels with the project. Client teams inherit the methodology, not just the report. Less dependency, more compounding.
- The direction of travel is clear. AI is closing the gap between expertise and action. The consultants who restructure around that will spend more time in the field, not less.
What “AI-Augmented” Actually Means Here
The term gets used loosely. Almost any consultant who has opened ChatGPT now calls themselves AI-augmented. That is not the meaning I am using. The version I mean is specific, and it has four properties together. Drop any one of them and the model breaks.
Years of doing the actual work. Circular economy projects in Indonesia and Africa, ESG research on CSRD and ESPR, systems thinking applied to waste, water, and energy. The tools do not create this. They amplify it.
Fluent use of the tools at the architecture level. Skills, sub-agents, prompt caching, programmatic outputs. Reproducible work, not one-off ChatGPT conversations.
The output is the client deliverable, not a slide deck about it. Audits, regulatory analyses, business models, grant proposals, training programmes. Real artefacts that get used.
The methodology travels with the work. The CLAUDE.md, the skill, the workflow document. The team that hired the consultant ends up able to run the next analysis themselves.
Without domain expertise, you get a technologist asking sustainability teams to define their own requirements. Without AI fluency, you get a traditional consultant who can use ChatGPT but has not restructured around the tools. Without delivered work, you get a strategy advisor with a more sophisticated slide deck. Without capability transfer, you get a black-box service that creates dependency rather than independence. The combination is what makes the model different.
The Field-Computer Ratio
The version of sustainability consulting that makes it onto LinkedIn carousels is mostly slide decks and strategy frameworks. The version that gets actual results is messier and more physical. Walking the harbour to map where fishing nets accumulate. Spending weeks with municipal teams and informal waste collectors. Sitting in a workshop while the team iterates on a bamboo treatment process. Training, supervision, network building, behaviour change.
The frustrating hours are usually the screen ones. Synthesising frameworks, formatting reports, building business models, regenerating the same brand voice instructions every time I sit down to write something. The on-the-ground work is what nobody can outsource. The screen work is exactly what AI can now help with.
Industry analysts have started to name this shift directly: AI is freeing valuable technical human resources to maximise report quality and customer experience. Drones and satellite imagery with AI overlays are starting to handle ecosystem monitoring that used to mean physical surveys. The category that environmental and sustainability consultants are moving towards is one where the human time goes where it adds most value, and the rest gets compressed.
The structural shift that matters more than any feature announcement is not “AI replaces consultants.” It is something more specific. AI quietly takes a chunk out of the screen-bound work, so the operational work grows into the space it should always have occupied.
Three Real Projects, with AI in the Plumbing
The cleanest way to show what this means is to take three past projects and walk through where AI tools, used at the architecture level I work at now, would have changed how I spent the hours. None of these ran with the tools I have today. The “what if” is illustrative, not a claim that the projects were inefficient. They produced real results.
Project STOP, Indonesia (2019-2020)
Multi-stakeholder programme to build plastic recovery infrastructure in coastal Indonesian cities. My role covered innovation and knowledge management. Much of the work was in the field with municipal teams, informal waste collectors, and regional government counterparts.
Where AI would have helped: the waste composition analysis, the regulatory comparison across Indonesian provinces, the literature review on circular economy interventions in similar contexts, the donor briefing notes, and the case study writeups for international partners. Probably a third of my screen time across that year. What I would have done with the time back: spent more weeks in Yogyakarta and Pasuruan, sitting with operational teams while they ran the systems we had helped design.
Seychelles fishing net recovery (2021)
End-to-end study identifying how to recover discarded fishing nets in Victoria’s port and route them into productive reuse, alongside the Sinerxia consultancy and the Seychelles government. The result was a small recycling company with five jobs and a meaningful reduction in nets going to landfill.
Where AI would have helped: the European market analysis for nylon recycling capacity, the business model iterations, the stakeholder mapping, the formal report. The European scan could have been days rather than weeks. The business model spreadsheet, which iterated maybe ten versions, could have been a structured prompt cycle. What I would have done with the time back: more hours in the port talking to fishing crews about how they actually decide whether to bring a damaged net back or dump it at sea, which is where the actual policy levers live.
East Bali Poverty Project bamboo research (2016)
I worked with the East Bali Poverty Project, a charity supporting upland Balinese communities, on identifying viable bamboo-based social businesses. The work covered market research across European and Australian niche markets, visits to bamboo workshops and processing facilities, and conversations with local craftspeople. The research informed what eventually became East Bali Bamboo Bikes, a social enterprise still operating today.
Where AI would have helped: the market and competitor research, the fundraising prospects list, the early grant proposal drafting, the reporting back to the charity board. What I would have done with the time back: more time in the highlands with the workshop teams, more visits to the bamboo plantations, more conversations with the cooperative members who would eventually run the business.
The pattern across all three is the same. The screen-bound work, which used to consume thirty to fifty percent of an operational consultant’s hours, is exactly what AI is genuinely good at. The field-bound work, which is where the value comes from, is not getting easier or faster. It is just becoming the larger share of where my time actually lands.
Where AI Genuinely Helps
A more concrete inventory of where the architecture earns its place in operational sustainability work. None of this is novel as a list of capabilities. The combination is what changes the unit economics.
Cross-referencing CSRD, ESRS, GRI, sector-specific guidance. Days instead of weeks, with structured outputs that update cleanly when regulations change.
Scanning recycling capacity, bidding for niche markets, mapping supply chains. The kind of desk research that supports a field intervention but never replaces it.
Iterating on revenue models, cost structures, breakeven scenarios. Useful for early-stage social enterprises and recovery schemes where the model has to flex during design.
Donor reports, board briefings, grant proposals, formal recommendations. Drafted with the methodology consistent across documents, edited by hand for the parts that matter.
Material flows, waste composition, energy consumption, monitoring datasets. Structured analysis pipelines that handle the synthesis without losing the underlying numbers.
Training materials for multilingual teams. Stakeholder communications. The cross-cultural workload that has always added a tax on international operational projects.
The technical detail behind this is worth one sentence. The architecture combines prompt caching, reusable skills, sub-agents with isolated context, and programmatic outputs to change the ratio of structural overhead to actual analysis. The longer read on the methodology, if you want it, sits in the Claude Code for sustainability research piece.
Capacity Building Is Half the Point
What makes operational consulting different from advisory consulting is that the goal has always been to leave the client more capable than you found them. Train the municipal team, leave the methodology, hand over the system. Field work without capacity transfer creates dependency that nobody wants.
The AI multiplier changes what capacity transfer can look like. It used to mean a training session, a written methodology, a checklist. Now it can also include a working Claude Code skill, a CLAUDE.md file with the team’s domain context, a worked example the team can adapt. The capability is in the artefact, not just in someone’s memory.
Several of the consulting skills I use day to day are now open source on the public skills hub. The research-writing-assistant skill, the grant-proposal skill, the GSC audit skill, the chart-rendering library. None of them are products in their own right. They are the working methodology of a practice, exposed so other people can use it. The fact that I do not sell them is the point. They are part of the work.
Where This Is Heading
The interesting question is not how AI changes consulting tomorrow. It is what the work looks like in three years, when the architecture has matured and the back-office compression has compounded. A few directions that feel directionally credible based on what is already moving.
- Field-aware AI assistants. Tools that take field notes, sensor data, drone imagery, and stakeholder transcripts and turn them into structured analysis without losing context. The consultant supervises and edits rather than producing from scratch.
- Leverage-point identification. AI structured to apply systems thinking to large datasets, identifying intervention points that humans would miss because they are buried in volume. Not autonomous decision-making. Augmented diagnostic capacity.
- Continuous monitoring instead of episodic studies. The current model is a consultant flies in, runs a study, writes a report, flies out. The next model is continuous data flow, with the consultant intervening when patterns deserve attention.
- Capability transfer at infrastructure scale. Skills and methodologies travel further. A circular economy framework built once can underpin work in twenty cities, with local consultants adapting rather than rebuilding.
The framing the WEF used recently is the cleanest one I have seen. AI is starting to close the gap between expertise and action. The consultants who restructure around that close will spend less time producing paperwork and more time changing actual things on the ground. That is the direction worth building toward.
What This Isn’t
Honest limits. AI does not substitute for years of operational experience, or for time spent in the field, or for the slow trust that real interventions depend on. It does not produce infinite output from one practitioner. It is not appropriate for everything, and it comes with environmental and disclosure trade-offs I have written about elsewhere. The model only works when the constraints are named honestly.
Three Ways to Engage
Three project shapes, each suited to a different starting position. The shapes intentionally read like projects rather than service tiers, because that is what they are.
A regulatory analysis, a circular economy diagnostic for a specific material stream, a business model for a recovery scheme, a grant proposal under deadline. I deliver, you get the artefact, the engagement closes.
A field project where I work alongside operational staff on the ground, with the back-office synthesis and reporting running on the AI-augmented architecture. The team learns the methodology by doing the work.
I work with your team to translate their existing methodology into reusable skills, document the workflow, and walk them through running a real piece of work end to end. The team owns the capability when we close.
All three sit under the same AI for Sustainability Research consulting practice. If any of them fits the situation you are sitting in, that is the conversation to start.
Why This Matters Now
The regulatory volume coming through CSRD, ESPR, and CBAM is rising faster than existing consulting capacity can absorb. Two models are emerging for handling it. The very large firms are scaling by spending heavily on AI platforms and packaging them for enterprise clients. That is one workable answer. It is not always the right shape for operational sustainability work, and the Deloitte case is a reminder that AI rigour without domain rigour produces expensive errors.
The other model is smaller practices, restructured around the tools, doing the operational work that has always been the hard part of sustainability consulting, with the screen-bound work compressed enough to free up the field time that actually moves a project. That is the shape I am building, and the shape I think most sustainability work will need over the next two to three years.
If any of that resonates with the kind of project you are thinking about, the simplest next move is to talk through your project. I read every message myself, write back honestly about whether the model fits, and only take on work I think I can deliver well. That filter is part of how this stays operational.
Related reading
- Claude Code for sustainability research: a practitioner’s workflow. The methodology behind the back-office speed.
- How Claude Code skills cut AI energy use. The technical and energy economics of the architecture.
- Is Claude sustainable? What Anthropic discloses, what it doesn’t. The honest position on the tools themselves.
- AI usage in sustainability consulting: Anthropic Index analysis. What the data says about how sustainability professionals are actually using AI.
- The Claude Code skills hub. The open-source consulting skills that travel with the work.