Claude vs ChatGPT for Letting Agents (UK 2026): Side-by-Side
Section 21 'no-fault' evictions are abolished on 1 May 2026. A Bristol agency principal asked us last week whether her team could use Claude or ChatGPT to handle the new notice regime. The honest answer was: probably, but only if the right tool is doing the work. And the subscription most of her negotiators already pay for is the wrong one.
Most Claude vs ChatGPT comparisons online are written by US real estate sales agents. None of them mention the Renters' Rights Act 2026. None of them work through Right to Rent compliance for AI inputs, or what to do with tenant PII sitting in chat logs. With commencement two weeks away, that gap matters.
So we tested both tools on four real letting-agent workflows from a 250-tenancy Bristol agency. Tenant comms. Compliance Q&A. Maintenance triage. Data handling. Here's the honest answer on Claude vs ChatGPT for letting agents in 2026.
The one-line verdict
Use Claude for tenant communications and compliance writing where tone and accuracy matter. Use ChatGPT for triage, custom GPT bots, and team-wide rollout. Whichever you pick, do not use the consumer tier for anything containing tenant data, Right to Rent documents and deposit records are special-category PII.
That's the verdict. Read on for what we tested, what each tool got wrong, and the starter stack we'd ship before 1 May.
How we tested
We ran four workflows against both tools between 10 and 17 April 2026, plus a fifth section on data handling. We tested:
- Claude Pro (consumer subscription, £15.50/month) and Claude for Work (Team tier, ~£20/seat/month, 5-seat minimum)
- ChatGPT Plus (consumer subscription, £16/month) and ChatGPT Business (Team tier, ~£20/user/month, 2-user minimum)
The test agency context: a Bristol-based independent managing 250 tenancies across central Bristol and BS3/BS4. Four staff: two negotiators, one property manager, one principal. Reapit-based CRM, Fixflo for maintenance, Zoopla and Rightmove for listings. About 380 tenant emails a week. Same prompts, same source documents, no cherry-picking.
Round 1: Tenant communications and dispute de-escalation
The test: an aggressive tenant email demanding the full deposit back after a supervised check-out where £340 of cleaning costs were proposed. The tenant referenced the TDS and threatened a formal complaint to Property Redress Scheme. Draft a reply that holds the line on the £340 deduction without escalating.
ChatGPT Plus drafted a polite, conciliatory reply in 14 seconds. The tone was excellent. The problem: it offered "a goodwill gesture of £100" without being asked, contradicting the agency's TDS process. If the negotiator had sent it as drafted, the agency would have weakened its position in the formal adjudication that almost certainly follows.
Claude Pro wrote a longer reply that took the dispute seriously without giving ground. It explained the TDS process, confirmed evidence had been logged, and routed the tenant to formal adjudication. It also flagged a sentence the negotiator might want to soften ("we are confident the proposed deductions will be upheld") as potentially inflammatory.
Round 1 verdict: Claude. Not just on tone, on procedural awareness. ChatGPT's reflex to defuse with a goodwill gesture is exactly the wrong instinct in a deposit dispute. For tenant comms with a compliance dimension, give the work to Claude as a default.
Round 2: Lettings compliance after the Renters' Rights Act
The test: five UK lettings questions a property manager might ask in the run-up to 1 May 2026. Section 21 notice validity after commencement. Conversion of an existing AST to assured periodic. Right to Rent check procedure for an EU national without settled status. The new statutory Information Sheet obligation. Rent-increase mechanics under the new regime.
Both tools knew the Renters' Rights Act was on the horizon. Both correctly stated Section 21's abolition. Both flagged the Information Sheet requirement.
The differences were in precision. ChatGPT invented a "Form RRA-1" that doesn't exist when describing the Information Sheet. The form is real (it's published on gov.uk) but isn't called that. Claude answered more cautiously, named the document by the gov.uk title, and pointed the user to the source PDF. On the Right to Rent question, Claude flagged that EU nationals without settled status need a share code from the Home Office; ChatGPT gave a more general answer that would have produced a non-compliant check.
This matters. A wrong Right to Rent check is a £20,000 civil penalty. A wrongly drafted Section 21 (or its replacement) loses you the possession case.
Round 2 verdict: Claude on safety. ChatGPT moves fast but invents official references with confidence, exactly the failure mode the Bloomberg AI tax-prep coverage flagged in March 2026. For compliance questions, verify everything regardless, but Claude needs less verification.
Round 3: Maintenance triage and work order drafting
The test: an out-of-hours tenant report, "water coming through the kitchen ceiling, dripping into a saucepan, getting worse." Triage the urgency, draft a work order to the agency's preferred plumber, and draft a tenant reply.
Both tools triaged it as Priority 1 (water ingress, active damage, risk to electrics). Both drafted reasonable work orders. The differences:
ChatGPT Plus was faster and produced a slightly tidier work order with a clean parts/labour estimate. The tenant reply was polite but generic. It didn't ask whether anyone was vulnerable in the property or whether power had been isolated to the affected room, both standard questions that affect contractor dispatch.
Claude Pro asked those questions in the tenant reply automatically. The work order was almost identical to ChatGPT's. The tenant reply was about 40% longer and included a "what to do in the meantime" paragraph that several agencies we work with have permanently in their reply templates.
Round 3 verdict: Roughly tied, with Claude's extra safety questions edging it for first-pass triage. If you've already built a triage template (or you're using something like Fixflo for routing), ChatGPT's speed wins. If you're starting from a blank prompt every time, Claude is safer.
A real example: A 4-person agency in Manchester ran their full week of maintenance reports through both tools as a stress test. Triage accuracy was identical. Tenant-reply quality favoured Claude. The property manager's verdict: "Claude writes the reply I'd want my best property manager to write. ChatGPT writes the reply I'd accept from a good one. Neither is bad, but only one of them sounds like us."
Want the prompt library, triage templates, and Make.com recipes we use with UK agencies? Get the £49 Letting Agents Playbook, five hours of admin time back in your first week, with the exact prompts and the Reapit-Fixflo wiring diagram.
Round 4: Data handling and UK GDPR for tenant data
This is the section nobody else covers, and for letting agents it's worse than most industries, you handle special-category data routinely.
Right to Rent checks involve passport scans, Home Office share codes, biometric residence permits. Tenancy applications include guarantor financials, bank statements, and employer references. Family lets include children's names, ages, and sometimes school addresses. All of this is personal data under UK GDPR, and most of it is sensitive enough that pasting it into a consumer AI tool is a serious problem.
The Claude Pro and ChatGPT Plus standard terms allow conversation retention for up to 5 years and may use inputs for model training unless you opt out, with the opt-out applying inconsistently depending on workspace settings. Dovetail's GDPR comparison of the major AI assistants covers the detail. The short version: standard consumer accounts are not GDPR-compliant for tenant-confidential data.
Claude for Work (Team and Enterprise) and ChatGPT Business / Enterprise are different. Both prohibit training on customer content by default. Both offer admin controls, audit logging, and SSO. ChatGPT Enterprise also offers UK data residency for eligible customers, useful if your insurer or your data protection officer requires UK storage.
The ICO guidance on AI and data protection is clear: you remain the data controller. The AI provider is the processor. If you're using a tool whose terms allow training on inputs, you need a lawful basis and almost certainly a Data Protection Impact Assessment.
In practice, for a letting agency:
- Don't paste Right to Rent docs into Claude Pro or ChatGPT Plus. Ever. Don't paste tenant ID copies, bank statements, or guarantor info either. Use consumer tiers only for redacted research and template drafting.
- For tenant-facing work, upgrade to Claude for Work or ChatGPT Business. ChatGPT Business is the easier starting point for small agencies (2-user minimum vs Claude for Work's 5-seat minimum).
- Update your privacy notice. A short paragraph noting AI tooling is used under business-tier terms is more honest than the alternative, staff quietly using consumer tools because nobody told them not to.
There's a separate angle worth flagging: tenants are increasingly using ChatGPT to draft formal complaints and tenancy defences. UK courts have started noting it explicitly. If your tenant emails are getting noticeably more polished and structured, that's why. Your reply should match the rigour, which is exactly the case for using Claude on the agency side.
Round 4 verdict: Both tools are GDPR-acceptable on business tiers. Both are not on consumer tiers. The mistake isn't choosing, it's failing to upgrade in the first place.
Pricing for a 250-tenancy agency
Here's what each tier costs in pounds, with realistic monthly spend by agency size:
| Tier | Monthly cost | Training on inputs? | Right for tenant data? |
|---|---|---|---|
| Claude Pro (consumer) | ~£15.50/user | Possible under standard ToS | No |
| ChatGPT Plus (consumer) | ~£16/user | Possible under standard ToS | No |
| Claude for Work (Team) | ~£20/seat, 5-seat min | No | Yes |
| ChatGPT Business (Team) | ~£20/user, 2-user min | No | Yes (UK residency on Enterprise) |
By agency size:
- Sole agent (1 staff): One ChatGPT Business seat (~£20/month) is the minimum for compliant tenant work. Claude for Work's 5-seat minimum is wasteful at this size. Add Claude Pro (~£15.50) for personal research only, never for tenant data.
- Typical high-street branch (4 staff): 4 ChatGPT Business seats (~£80/month) covers compliant tenant comms. Add Claude for Work (5-seat minimum, ~£100/month) only if compliance writing volume justifies it. Total compliant stack: £80–£180/month.
- Multi-branch independent (10 staff): 10 ChatGPT Business + 10 Claude for Work = ~£400/month. About £40 per person for both tools, recovered in well under an hour of saved time per week.
For a 250-tenancy agency at four staff, our typical recommendation is ChatGPT Business for everyone (~£80/month) plus Claude for Work for the property manager and principal only (5-seat minimum forces you to take 5, but two seats unused is fine). Total: ~£180/month. Time saved on tenant comms and compliance drafting alone usually exceeds 6 hours/week across the team.
The honest answer: which to deploy for which workflow
After running the four rounds, the split is clear:
| Use Claude for | Use ChatGPT for |
|---|---|
| Tenant emails with a compliance dimension | Maintenance triage (when paired with a template) |
| Deposit dispute replies and arrears chases | Custom GPT for tenant FAQ bot |
| Section 8 / new-regime notices and Information Sheets | Quick research and learning |
| Right to Rent procedural questions | Spreadsheet work (rent reviews, arrears reports) |
| Anything that needs careful regulatory language | Fast first drafts your team will edit anyway |
If your agency does mostly tenant find and basic management, ChatGPT alone might be enough. If you handle disputes, complex tenancies, or any HMO work, Claude pays for itself. Most agencies running both for a month settle on a 60/40 split, heavier on Claude for the property manager, heavier on ChatGPT for the negotiators.
For the wider tool stack beyond the LLMs, our AI for letting agents guide covers the broader landscape, and our tenant communications automation post walks through the email-triage flow specifically.
A starter automation stack for UK letting agents
If you're starting from zero and want to ship something working before the Renters' Rights Act commences on 1 May 2026, here's the build:
Week 1, Foundations and compliance. Sign up for ChatGPT Business (one seat per staff member, 4 minimum for our test agency). Cancel any consumer subscriptions used for tenant data. Update the privacy notice with a short AI-tooling clause. Set the team a single rule: no tenant data goes into anything except the new Business workspace.
Week 2, Tenant comms and triage. Write five saved prompts: tenant enquiry reply, deposit-dispute holding response, arrears chase, maintenance triage, viewing confirmation. Test each on anonymised data. For the maintenance flow, wire Fixflo to a Make.com scenario that pre-summarises the report before it lands with the property manager.
Week 3, Add Claude or stop. If your property manager is spending more than three hours a week on drafting (deposit replies, arrears letters, formal notices, Section 8 drafts), add Claude for Work. The 5-seat minimum is irritating but the writing quality justifies it for any agency doing real tenant management.
Week 4 (optional), Custom GPT tenant FAQ bot. Build a custom GPT trained on your AST template, your handover process, and your top 30 tenant questions. Embed it on your site or use it internally as a knowledge base. We cover the build pattern in our custom AI assistants service.
This is the same sequencing we use when we run an AI Assessment for a letting agency, typically mapping 12–15 candidate workflows, scoring them by impact and effort, and prioritising the three highest-return automations. For agencies in the capital, our AI for property managers London page covers the same approach with London-specific notes, and the letting agents industry hub covers the wider service offer.
If you're not sure where time is actually going, the admin cost calculator and automation checklist is a free 10-minute exercise that surfaces the obvious wins.
A real example: A London sole agent started with Claude Pro at £15.50/month for "tenant emails." She moved to ChatGPT Business (~£20/month) after a Right to Rent passport scan ended up in a chat thread that, under standard terms, could have been retained for 5 years. She kept Claude Pro on the side for personal research only and now spends about £36/month total. Time recovered in the first month: 5.5 hours. Zero compliance exposure.
Frequently asked questions
Which is better for UK letting agents, Claude or ChatGPT?
Claude wins for tenant communications with a compliance dimension, deposit-dispute replies, and any drafting that touches the Renters' Rights Act. ChatGPT wins for fast triage, custom GPT bots for tenant FAQs, and team-wide rollout. Most UK agencies benefit from running both, with the property manager leaning on Claude and the negotiators leaning on ChatGPT.
Can Claude or ChatGPT handle Section 21 notices after the Renters' Rights Act 2026?
Section 21 notices are abolished from 1 May 2026 for new tenancies. Both tools know this. Both can draft the new-regime equivalents (Section 8 with the appropriate ground), but neither should be used as the source of truth, verify against the gov.uk Information Sheet 2026 and consult NRLA guidance for edge cases. Claude was more accurate in our testing; ChatGPT invented an official form name.
Is Claude or ChatGPT GDPR-compliant for tenant personal data?
The consumer tiers (Claude Pro, ChatGPT Plus) are not safely GDPR-compliant for tenant data, both can retain conversations and use them for training under standard terms. Right to Rent documents are special-category data and should never go into consumer tiers. The business tiers (Claude for Work, ChatGPT Business) prohibit training on customer content and are appropriate for tenant-facing work. ChatGPT Enterprise offers UK data residency for eligible customers.
Can AI draft tenancy agreements or Section 8 notices?
Both tools can produce a usable first draft of a Section 8 notice or a tenancy agreement, but neither should be treated as the final version. The legal accuracy of notices matters, a wrongly drafted Section 8 loses you the possession claim. Use AI to accelerate drafting, then verify against your standard precedents and (for anything contested) your solicitor.
How much do Claude and ChatGPT cost for a UK letting agency?
Approximate 2026 pricing: Claude Pro ~£15.50/user/month, ChatGPT Plus ~£16/user/month (consumer tiers, not for tenant data). Claude for Work ~£20/seat/month with a 5-seat minimum. ChatGPT Business ~£20/user/month with a 2-user minimum. A typical four-staff agency running both for compliant tenant work should budget around £180/month total.
What should letting agents do about tenants using ChatGPT to write complaints?
Tenant emails are increasingly AI-drafted; courts have started noting it. Your reply should match the rigour rather than dismiss it. The polished, structured tenant complaint deserves a polished, structured agency response, which is exactly what Claude is good at producing. The deeper risk is that AI-assisted tenants may push more cases to formal adjudication; agencies need their compliance documentation to be airtight.
The bottom line
Claude vs ChatGPT for letting agents is not a single-winner question, but the answer matters more in 2026 than it did last year. With the Renters' Rights Act commencing on 1 May 2026, agencies have a small window to get their tenant comms, compliance drafting, and data handling right before the regime changes. Claude is the better default for compliance writing and tenant comms with stakes. ChatGPT is faster, cheaper to deploy, and stronger for triage and custom GPT builds.
If you want help mapping which workflows in your agency would benefit most, and which tool to deploy where, book the £499 AI Assessment. We'll interview you for 45 minutes, map your workflows, and deliver a custom report in five working days with the 5–7 highest-impact automations for your agency. If we can't find at least five hours of weekly time savings, you pay nothing.
The 1 May commencement isn't waiting. Get the stack working before then, and the new regime is a transition you ride rather than a fire you fight.