AI client intake for an immigration consultancy, done well, is not a chatbot bolted onto your website — it is a connected workflow that captures an enquiry, screens it against your eligibility criteria, scores it, offers qualified enquirers a consultation slot in a fee-earner's calendar, and pushes structured data into your CRM or case management system. The chatbot is only the interface. The value sits in the screening logic, the document collection, and the handoffs that happen after the conversation ends.
For immigration firms in Europe, this distinction matters commercially. A chatbot answers questions. A well-designed intake system helps you work out which of last month's enquiries are worth a paid consultation — and can get those people booked before they contact other firms.
The Core Problem: Enquiry Volume and Enquiry Fit
Immigration practices can sit in an unusual position. Enquiry volume may be high, because immigration is a topic people research at length and at all hours. Conversion quality may be uneven, because some enquirers will be ineligible, unable to pay, in the wrong jurisdiction, or asking a question they could have answered from public sources.
Rather than rely on general claims, it is worth checking four things in your own enquiry data:
- How many inbound enquiries are unqualified? — wrong country of residence, no eligible visa route, unrealistic timelines, or looking for free advice.
- How many enquiries arrive outside business hours? Applicants are often in different time zones from their destination country; your inbox timestamps will show whether that matters for you.
- How long do enquirers wait for a first response, and what happens to those who wait? Applicants who contact several firms at once may favour whoever answers first; your booking data can test that.
- How many paralegal and consultant hours go to conversations that will not generate a fee?
On response speed, the best-known research is a 2011 Harvard Business Review study, The Short Life of Online Sales Leads (Oldroyd, McElheran and Elkington, HBR, March 2011). It sent test web leads to 2,241 US companies. Among the companies that responded within 30 days, the average first response took 42 hours; only 37% replied within an hour, 24% took more than 24 hours, and 23% never responded at all. Firms that contacted a lead within an hour were nearly 7 times as likely to qualify it as firms that took longer, and more than 60 times as likely as firms that took 24 hours or more. This is US company data from 2011 about the odds of qualifying a lead — not sales outcomes, and not European or immigration data — so treat it as a reason to measure your own response times, not as a benchmark for your practice.
The traditional approach is a contact form plus a junior team member triaging an inbox. It tends to be slow and inconsistent, and it usually scales only by hiring.
What "AI Client Intake" Actually Includes in 2026
A functioning intake system for an immigration firm can be thought of as five layers. When you evaluate solutions, check whether a quote covers all five or only the first.
1. The Conversational Front End
A web widget, WhatsApp thread, or embedded form that asks questions in natural language, in the prospect's own language. For European firms the practical question is which languages your enquirers actually use and whether the system handles each of them within the same flow. Language models can handle multilingual conversation, but quality varies by language, so test each one you need.
2. The Screening Layer
This is where an immigration firm AI screening setup earns its keep. Instead of open-ended chat, a good system runs structured eligibility logic: nationality, current country of residence, visa category of interest, employment or investment status, family composition, prior refusals, timeline, and budget range. Each answer narrows the branch.
Crucially, this logic is yours. For illustration, a practice built around an investor-visa route (a Golden Visa-type programme, say) might screen on investment thresholds and source-of-funds readiness; a UK sponsor-licence practice might screen on company size and Certificate of Sponsorship need; a practice working with a Portuguese D7-type residence route might screen on passive income levels. These are examples of route types only. Routes and rules change, so the criteria in any screening flow must be set and checked by your own team against current official sources. A generic chatbot with no configured criteria cannot do this; a configured intake system can be set up to.
3. Scoring and Routing
Each completed conversation produces a score and a route: offer a paid consultation, route to a specific fee-earner by practice area or language, send a follow-up sequence, or politely decline with a signposting message. Declining well matters — a respectful decline can protect your calendar and your reputation at the same time. GDPR Art. 22 restricts decisions based solely on automated processing that have legal or similarly significant effects; ask your data-protection adviser whether and how that applies to your flow, and consider having a person review declines.
4. Booking and Payment
Qualified enquiries can be offered a live calendar inside the same session. Some firms also want the consultation fee taken at the point of booking. That is a commercial decision for your firm and an integration question for any vendor, so ask whether it is supported and how, rather than assuming it is.
5. Data Handoff and Document Collection
A good system writes a structured record into the CRM or case management platform your firm already uses, including a conversation transcript and a summary. Which platforms connect, and how, differs by vendor and by project, so ask directly whether it connects to your CRM. Some configurations can also trigger an initial document request list based on the route identified during screening. Whether clients should upload passports and other identity documents through an automated flow before a first call is a data-protection and risk decision for your own adviser.
Chatbot vs. Automated Intake System: A Direct Comparison
| Dimension | Basic Website Chatbot | Automated AI Intake System |
|---|---|---|
| Primary function | Answer FAQs, capture name and email | Screen, score, route, book, collect documents |
| Eligibility logic | None or generic | Configured to your visa routes and criteria |
| Output | Email notification to inbox | Structured CRM record + calendar booking + transcript |
| Language handling | Varies by product | Ask which languages are supported, and test each one |
| Unqualified enquiries | Passed through to your team | Can be routed or declined under rules your firm sets, ideally with human review |
| Payment at booking | Not a core function | Optional; depends on vendor and integration |
| Build time | Varies by product | Ask for a written timeline; it depends on routes, languages, integrations and volume |
| Cost profile | Varies by product | Ask for a written scope and fixed price; it depends on the same factors |
| Measurable outcome | Engagement rate | Cost per qualified consultation |
One commercial test: if a system cannot show its effect on your cost per booked, qualified consultation, it is closer to a convenience tool than an intake system.
What Changed Between 2023 and 2026
Three developments help explain why automated intake for immigration firms is more practical than it was a few years ago.
Reliability of structured extraction. Earlier chatbots tended to follow rigid decision trees (robotic) or free-form LLM chat (fluent, but less reliable at capturing clean data). Some current architectures combine both: natural conversation on the surface, strict schema validation underneath. That can help the nationality field actually contain a nationality, though captured data should still be spot-checked.
Compliance expectations. The EU AI Act entered into force on 1 August 2024. Since 2 August 2026, Art. 50(1) requires providers of AI systems designed to interact directly with people to make clear that the person is interacting with an AI system, unless that is obvious from context; Art. 50 breaches can be fined up to EUR 15 million or 3% of worldwide annual turnover. GDPR also continues to apply to the personal data an intake conversation collects, and Art. 83(5) sets fines of up to EUR 20 million or 4% of total worldwide annual turnover, whichever is higher. Which obligations apply to your firm, and how to meet them, is a question for your own counsel or data-protection adviser.
Integration options. Connections between conversational tools and CRMs, calendars and other business systems are more widely available than they were, so intake data need not land in a spreadsheet someone has to re-key. Coverage still varies from system to system, so check yours.
Frequently Asked Questions
Will an AI system give legal advice by accident?
It is a risk to design against. A sensible approach is to scope an intake system to information gathering and routing: collecting details, explaining process steps at a general level, and passing anything that looks like an advice request to a person. What counts as regulated immigration advice differs by country, and in some places giving it without authorisation is restricted, so your own compliance officer (or equivalent) should decide where the boundary sits for your practice and confirm that the system is configured and tested to respect it. Disclaimers can help but do not replace that boundary, and the system should make clear that the person is talking to an AI.
How does this work with GDPR?
GDPR applies to the personal data an intake conversation collects. How the rules on lawful basis, consent, transparency and retention apply to your firm is for your own counsel or data-protection adviser to decide; this article does not say what your business must or may do. Practical questions to put to any vendor: what data is collected and why; where it is processed and stored; which roles you and the vendor hold and how that is set out in written data-processing terms; how retention is configured; and how your privacy notice will describe the processing.
Do enquirers actually engage with it, or do they bounce?
That is something to measure rather than assume. Length and framing are the obvious variables to test: a shorter flow that explains why each question is asked ("this tells us which routes you may qualify for") is a reasonable hypothesis to try against a long form. Track completion rate by number of questions and by language in your own data.
What happens to the enquiries the system rejects?
In a well-designed system they receive a clear, respectful message explaining that their situation falls outside your current practice areas or eligibility thresholds, and it may point to official government resources. Some firms route near-miss cases into a follow-up sequence, since circumstances change; whether and how you may keep in touch is a question for your own counsel or data-protection adviser.
Can it handle a family or corporate enquiry with multiple applicants?
It can, though this needs deliberate design. Screening should identify the principal applicant, count dependants, and flag corporate enquiries for a different route — sponsor licence and relocation work differs enough from individual applications that it may deserve its own flow.
How long before there are results?
Nobody can give a trustworthy figure without knowing your routes, languages, integrations and enquiry volume, so ask any vendor for a written scope, timeline and fixed price. For results, decide in advance which baseline you will compare against — booked-and-attended consultations and cost per qualified consultation in a comparable period before automation — and how much volume you need before the comparison means anything. Be wary of any vendor that promises a specific outcome or date before seeing your data.
How to Evaluate Vendors at the Consideration Stage
If you're comparing options, five questions help separate serious providers from resellers of generic chatbot templates:
- "Show me the screening logic you'd build for our top three visa routes." A concrete answer suggests they have thought about your practice. A vague answer about "smart AI" is worth probing.
- "Which CRM and case management systems have you integrated with, and does it connect to ours?" Ask for specifics. Native integration, API, or Zapier-style middleware all have different reliability profiles.
- "Where is data processed and stored, and what written data-processing terms do you offer?" Ask your own adviser what your firm needs in place.
- "Which languages do you support in the conversation itself, and how do you test each one?" Especially relevant if you serve applicants from outside the EU.
- "What metric will you report on in month three?" A useful answer involves qualified consultations and cost per booking — not chat sessions or engagement rate.
The Bottom Line
The firms that benefit most from AI intake are likely to be the ones that treat it as a measured pipeline rather than a friendlier chatbot: enquiries screened against real eligibility criteria that a qualified person has set, qualified applicants offered a booking promptly, and unqualified enquiries handled respectfully without using up a fee-earner's afternoon. If that works, time not spent triaging an inbox can go to billable work. That is a hypothesis to test, not a promised outcome.
If you're currently evaluating AI client intake for your immigration consultancy, the useful next step isn't a demo of a chat widget. It's an honest look at your own numbers: how many enquiries you receive, how many convert to paid consultations, how long your average response takes, and what that gap is costing you.
If you would like a second view, a short free fit check on the Altamira site can help you see whether automation makes sense for your firm. Altamira is a new Malaga-based business (founded in 2026), so this article does not cite client results; the questions above apply to any vendor you talk to.