AI for ecology consultancies: you already know the rule

Ecology consultancies are not short of work. They are short of hours. That changes what AI is for, because the useful question is not whether these tools cut your costs.

It is whether they hand back hours of the scarcest thing in English planning, which is a qualified ecologist's attention. Nine in ten small builders say their applications are held up by a lack of ecological expertise inside the councils deciding them, and nearly four in ten councils have no in-house ecologist at all.

On the risk side, ecology is in an unusual position. It is one of the few professions that has been using machine classification for years, and it already wrote the rule that makes AI safe.

The demand is not the problem. Your capacity is.

It is worth being precise about the market, because it changes what a practice should do.

The Home Builders Federation surveyed the first year of mandatory Biodiversity Net Gain and published the results in April 2025. The findings are blunt. 98% of small and medium builders find BNG a challenge. 94% have had planning applications delayed by it. 90% say the delay came specifically from a lack of BNG expertise or resource inside the local authority. Nearly 40% of councils have no in-house ecological expertise. A quarter of councils lost ecologists in the past year. Councils are now spending an average of £23,000 a year each buying that expertise back in from consultants.

Read that from the other side of the desk. Every one of those numbers is a queue forming outside the profession's door. The binding constraint in the whole system is the number of competent-ecologist hours available, and it is getting tighter, not looser. CIEEM's 2025 State of the Profession survey reports the same story from inside: a shortage of people experienced enough for senior roles, and the sharpest workload pressure landing in the April to October survey season.

So the honest framing of AI for an ecology practice is not cost-cutting. A two-person consultancy does not have a cost problem. It has a capacity problem, and every hour it spends formatting a report is an hour it is not spending in a field or on a judgement that only it can make.

Except at the bottom of the market, where the work is being legislated away

There is one part of the order book moving the other way, and it moves in three weeks.

From 6 August 2026, BNG will not apply to developments of 0.2 hectares or below. Government guidance is also already explicit that for small developments you do not need an ecologist: someone familiar with the site can do the survey and run the small sites metric themselves.

Put the two halves together and the strategy writes itself. The routine, low-value, high-volume end of the work is being deregulated and handed to laypeople with a spreadsheet. The end that is getting scarcer and more valuable is the licensed, competent, named judgement that a planning authority will actually accept. A practice that uses AI to shave a bit off its cheapest work is solving the wrong problem. A practice that uses AI to claw back hours from admin and drafting, and pushes those hours into survey capacity and judgement, is moving toward the part of the market that is growing.

That is the whole case for AI in this sector, and it is worth nothing at all if the tools damage the judgement. Which brings us to the part ecologists already understand better than most professions.

You already know the rule

Ecologists have been working with machine classification for years. Point an automated classifier at a night of bat recordings and it will tell you what it thinks it heard. Nobody in the profession treats that output as a finding.

They are right not to. A 2024 study in PLOS ONE tested three commercial bat call classifiers and found accuracy varies enormously by program and by species. The programs were good at avoiding false positives. What they were bad at was the thing that matters most: for the endangered Myotis species, the classifiers correctly picked up a real call as little as 1% to 52% of the time depending on the species and the software. A classifier that confidently reports nothing while missing half the calls of the species you are most legally exposed on is not a small problem. The authors' conclusion is the sentence to keep: a qualified analyst should verify automated classifications before species-specific conservation, regulatory and permitting decisions. That study is on North American species, so the numbers do not transfer to a British bat. The lesson transfers exactly.

So the profession already runs a working rule for machine output, and it is a good one:

The machine narrows the work. A competent person confirms. The confirmation, not the machine's output, is what goes on the record.

That is the correct rule for ChatGPT too. It is the same rule. The mistake practices are making right now is treating generative AI as a strange new thing that needs a strange new policy, when they have been living the right policy for a decade. You do not need to invent an AI position. You need to notice you already have one, and apply it to the tool that writes sentences instead of the one that reads calls.

The difference, and it is the only one that matters, is that a bat classifier tells you when it is unsure. It gives you a confidence score, and a file you can open and listen to. A language model does not do that. It produces the same fluent, assured prose whether it is right or inventing. The check has to be yours, because the tool will never flinch.

Where it earns its keep

Turning field notes into a first draft. This is the strongest use in the sector by a distance. A structured set of survey notes, target notes, conditions, times and records becomes a competent first draft of a Preliminary Ecological Appraisal in minutes instead of an afternoon. The ecologist then corrects and owns it. Nothing is being decided. The typing is being done, and the typing is a real cost.

The boilerplate that is genuinely boilerplate. Methodology sections, survey limitations, legislative background, the standard mitigation text that appears in some form in every report you write. Repeated prose with a known correct form is exactly where these tools are reliable and where checking is cheap.

Desk study and literature review, with a hard condition attached. Pulling together policy context, designations and species ecology is quick work for AI. The condition is that every citation and legal reference gets opened and checked. More on why below.

Client and planning correspondence. Explaining to an impatient developer why the bat survey cannot happen in February is drafting, not judgement. High volume, low stakes, easy to check.

The unbillable back office. Proposals, scoping responses, fee quotes, tender answers, the CVs and project lists that go into every framework submission. None of it is billable, all of it is necessary, and it is where a small consultancy quietly loses its evenings.

Where it bites

Fluency is not evidence. This is the failure mode to watch, and it is subtler than an AI writing your conclusion. It is an AI-drafted report that reads beautifully, describes the survey effort with total confidence, and quietly smooths over an ambiguity you actually noticed in the field. The draft sounds more certain than the survey was. In a backlog, in October, that reads like a finished report rather than a warning. The moment a draft is more confident than your evidence, the tool has made things worse.

The metric is arithmetic with rules, not prose. The statutory biodiversity metric is a defined calculation in a defined tool. Ask a language model for biodiversity units and it will give you a plausible number, because plausible text is what it makes. It has no grip on the trading rules or the spatial multipliers, and a wrong figure flows straight into the planning submission, the management plan and the section 106 obligations. Use the tool the calculation lives in. Do not ask a chatbot to do it, and do not ask it to check it.

Habitat condition is a field judgement. Condition is the input that most moves the numbers, and it comes from a competent person standing in the habitat applying published criteria. Nothing does that from a photograph and a description. Let AI infer a condition score and you have let it invent the foundation the whole assessment rests on.

Fabricated references are not hypothetical. In June 2025 the High Court dealt with two cases together, Ayinde v London Borough of Haringey and Al-Haroun v Qatar National Bank. In one, five of the cases cited did not exist. In the other, eighteen authorities did not exist, and several of the real ones did not say what they were claimed to say. The lawyers involved were referred to their regulators. These were trained professionals filing documents a public body relied on. An ecological report is also a document a public body relies on, and a planning officer is entitled to assume your references are real. Open every one.

A wrong protected-species call is not a typo. If a report says there are no bats and there are bats, the consequence is not a corrected paragraph. It can be a permission granted on a false basis and later challenged, works that amount to an offence under the Conservation of Habitats and Species Regulations 2017 or the Wildlife and Countryside Act 1981, a developer with a stopped site, and a negligence claim aimed at whoever signed the report. Your professional indemnity cover and your licence-holder's standing with Natural England are what is on the table. The software vendor carries none of it.

Survey timing cannot be reasoned around. Bat activity surveys, great crested newt surveys, breeding bird surveys and botanical surveys all have windows, and government guidance to planning authorities is that they should not decide an application until the necessary surveys have been done, by a suitably qualified person, using the right methods, at the right time of year. If a tool ever seems to be helping you around a seasonal constraint, something has gone wrong.

The non-negotiables

• Apply your acoustic rule to your language model. The machine narrows, a competent person confirms, and the confirmation is what goes on the record.
• No AI in the chain that produces a protected-species conclusion, a condition score, or a biodiversity unit figure. Drafting the report around those findings is fine. Producing them is not.
• Open every citation, reference and designation. Assume invention until you have checked.
• Client and site data goes into a business-grade account with training switched off. Survey data locates protected species. That is sensitive information about a client's land and, in the wrong hands, about a badger sett.
• Be willing to say where you used it. CIEEM's approach to generative AI in its own membership process is transparency rather than prohibition. That is a defensible posture for report production too, and a much better place to be standing if a report is ever scrutinised.

A sensible first month

1. Pick one recurring document. The PEA or the standard survey report: high volume, known structure, easy to check.
2. Write your house structure and standard sections down once, properly, and give that to the tool as its template. Draft quality is set by the context you give it, not by the model you pick.
3. Use a business-grade account, not a personal login.
4. Draft only from your field data. If the notes do not support a sentence, the sentence does not go in, and a model that invents one has just told you your notes were thin.
5. At the end of the month, ask two questions. How many hours came back? Did a single error reach a client? If the second answer is yes, stop and find out why before you scale it.

What does not change

The value of an ecology consultancy is a competent professional who went to the site, formed a judgement, and will put their name to it. That is the part that is becoming scarcer, more valuable and harder to replace, precisely as the routine end of the work is being automated and deregulated away. It is also the exact part AI cannot do.

So the practices that come out of this well will not be the ones that adopted the most AI. They will be the ones that were clearest about which parts of their work were never for sale to a machine, and then used the machine ruthlessly on everything else.

What to do next

Take one report type, run it for a month, count the hours. If it would help to work out where AI belongs in your practice and where a person has to stay in charge, book a call. Before you spend anything, it is also worth knowing how to tell a good adviser from a good salesperson.

FAQ

Can AI write an ecology survey report?

It can write a competent first draft from your field data, which is a real saving on a task that is mostly typing. It cannot produce the findings. The survey, the species determinations, the condition assessment and the conclusions come from a competent ecologist, who then reads the draft against the actual data and owns what goes out.

Can AI do a Biodiversity Net Gain calculation?

No. The statutory metric is a defined calculation performed in the official tool, and its most important input, habitat condition, is a judgement made on site against published criteria. Ask a language model for biodiversity units and you will get a plausible-looking number with no reliable basis, which then flows into the planning submission and the management plan.

We already use automated bat call classifiers. Is generative AI different?

The rule is the same, but the tool is less honest. A classifier gives you a confidence score and a file you can open and listen to, so you know when to look harder. A language model produces equally fluent prose whether it is right or inventing, and it will never signal doubt. So the verification that your acoustic work already builds in has to be applied deliberately, because the tool will not prompt you for it.

Does the August 2026 BNG exemption change what a small practice should do?

It is worth planning for. From 6 August 2026, BNG stops applying to developments of 0.2 hectares or below, and GOV.UK already tells small developers they can run the small sites metric without an ecologist. The low-value end of BNG work is thinning. The answer is to protect the capacity only a competent ecologist can supply, which is an argument for taking admin and drafting load off your team, not for automating judgement.

What is the single biggest risk?

Not that AI writes your conclusion. That an AI-drafted report sounds more certain than your survey actually was, and nobody notices in the middle of the autumn backlog. Fluency is not evidence.

Sources and further reading

• Biodiversity Net Gain: one year on, Home Builders Federation, April 2025. Source for the 98%, 94% and 90% figures on SME builder delays, the proportion of councils without in-house ecological expertise, ecologist losses from councils, and the £23,000 average annual council spend on external consultants.
• Understanding biodiversity net gain, GOV.UK. Source for the 10% requirement, the statutory and small sites metrics, and the exemption for developments of 0.2 hectares or below from 6 August 2026.
• Solick, Hopp, Chenger and Newman, Automated echolocation classifiers vary in accuracy for northeastern U.S. bat species, PLOS ONE 19(6), 2024. Independent. Source for the variation in classifier accuracy, the low sensitivity for Myotis species, and the recommendation that a qualified analyst verify automated classifications before regulatory and permitting decisions.
• Ayinde v London Borough of Haringey and Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin), 6 June 2025. Source for the fabricated citations put before the court and the regulatory referrals that followed.
• Protected species and development: advice for local planning authorities, GOV.UK. Source for the expectation that surveys are done by suitably qualified people, using appropriate methods, at the right time of year, before a decision is made.
• State of the Profession Survey 2025, CIEEM. Source for the shortage of experienced senior ecologists and the concentration of workload pressure in the April to October survey season.