TL;DR: Atlas turns your client book from a static record into a queryable asset. You can ask plain-English questions across meetings, documents, and back office data from Intelliflo, Plannr, and Curo to surface clients with unused ISA allowance, pension transfer mentions, inheritance concerns, and overdue reviews. Every answer carries a source citation, so you can verify before acting and maintain an FCA-defensible audit trail. Atlas cuts the hours advisers typically spend on manual searches and spreadsheet exports, replacing them with plain-English queries that return cited answers in under two minutes. A 14-day free trial is available with no credit card required.

You know the clients who mentioned a pension transfer this year. You cannot find them without opening forty files. That gap costs you hours every week and surfaces at exactly the wrong moment: when a client calls with a question you know you discussed six months ago. 43.3% of UK advisers report paperwork and admin reduce time devoted to advice itself, and client book analysis sits squarely in that category. The data exists. The question is whether you can reach it without a manual audit.

AdvisoryAI built Atlas to change that. Atlas is our AI chat and intelligence layer that sits across your entire AdvisoryAI platform, letting you ask questions of your whole client book in plain English and receive cited answers drawn from your meeting transcripts, suitability reports, client documents, and back office data. Atlas remembers context across sessions, so conversations build on each other rather than starting from scratch. Think of it as a colleague who has read every file across the entire firm and can answer in under two minutes.

How AI Transforms Your Client Data into Insights

The shift is from static records to queryable data. Your client book holds years of meeting transcripts, fact-finds, and back office records, including the soft facts that matter most: client anxieties, family dynamics, health concerns mentioned in passing. The only way to answer a question like "which clients mentioned a pension transfer this year" has been memory or a spreadsheet export. Atlas reads across all of it and answers in prose with citations, so you get verifiable answers in under two minutes.

Advisers at Brooks Macdonald freed 6,000 hours annually across 60 advisers, with meeting write-up time reduced from 2.5 hours to a 30-minute review, and firms like Timothy James and Partners report a 50% reduction in post-meeting documentation time.

Querying Client Records in Plain English

You type a question the way you would ask a paraplanner. "Which clients mentioned a pension transfer in the last 12 months?" Atlas interprets your request, searches across Evie meeting transcripts, Emma-generated reports, uploaded documents, and synced back office data, then returns a cited answer. Adaptive Thinking, released in May 2026, shows each step as it happens: analysing the request, searching for a client, loading their profile. You can expand the thinking block to read the full reasoning behind any answer, and that reasoning persists with the conversation so older queries remain auditable.

How It Differs from Back Office Reporting

Your back office typically answers structured questions well. "Clients with review dates in Q4" is usually a standard report. But it typically cannot answer "which clients raised inheritance concerns in a meeting" because that information lives in unstructured transcript text, not a dropdown field. AdvisoryAI built Atlas to read the unstructured layer.

The distinction is clearest when you compare what each approach can answer:

How Atlas Retrieves Specific Client Data

Atlas reads synced back office data from Intelliflo, Plannr, and Curo, plus meeting transcripts, suitability reports, and uploaded documents. For Intelliflo, data is as current as your last manual sync. Plannr and Curo update automatically. Risk profile syncs from Intelliflo only, not Plannr or Curo. Atlas supports up to 50 synced documents per client, with in-chat uploads limited to 10 documents and 30MB per query.

When you ask a book-wide question, Atlas paginates through all matching records, not just the first result, so you receive the full set. It answers in prose, not tables or charts inside the chat, and it never exceeds your existing permissions: an adviser sees their own book, and cross-adviser meeting access is licence gated to Paraplanner and Paraplanner Plus, scoped by Team Management. Joint clients are not currently supported for Intelliflo or Plannr.

The time difference compounds across your working week. Each of these tasks currently means a manual export, a keyword search, or a spreadsheet build. Atlas replaces all five with a plain-English question.

Key Prompts for Deeper Client Book Analysis

The highest-value questions are specific. Vague queries return vague answers. These five categories surface the referral and review moments that manual audits miss.

Identifying Clients with Unused ISA Capacity

Prompt: "Which clients have not used their ISA allowance this tax year?"

Atlas scans back office data for ISA contribution records and cross-references meeting transcripts for any mention of additional savings capacity. The answer surfaces clients who may benefit from a pre-year-end review, creating a natural outreach trigger. This can be a referral opportunity: clients with unused allowance may know others in similar positions.

Tracing Recent Pension Transfer Requests

Prompt: "Which clients mentioned a pension transfer in the last 12 months?"

Pension transfers carry significant suitability and compliance weight. Atlas searches across meeting transcripts for mentions, including discussions where the topic was raised but not yet progressed to a formal recommendation, so you can review whether follow-up is needed. FCA file reviews look for personalised justification and a clear audit trail linking every recommendation to a specific client need.

Pinpointing Inheritance Tax Queries

Prompt: "Which clients raised inheritance tax concerns in meetings?"

IHT is typically not a structured field in your back office. It often appears in conversation: a client may mention a parent's estate, a property sale, or a gift to children. Atlas surfaces these mentions so you can prioritise estate planning conversations before the tax year end or before a client's circumstances change.

Flagging Clients Within 18 Months of Retirement

Prompt: "Which clients are within 18 months of their stated retirement age?"

Retirement transitions are among the highest-stakes advice moments. Atlas can help identify clients approaching their stated retirement date by searching across fact-finds and meeting transcripts, helping you schedule review meetings before the transition rather than after. This supports proactive client care at critical life stages.

Identifying Overdue Annual Client Reviews

Prompt: "Which clients have not had an annual review in the last 13 months?"

Consumer Duty requires firms to monitor and evidence that clients are receiving good outcomes from their ongoing service. Atlas can help scan review dates across your back office and meeting history to flag clients who may be overdue, helping you address potential bottlenecks proactively.

Verifying Answers with Direct Source Citations

Citations are not a nice-to-have. They are the difference between an answer you can act on and one you cannot. For advisers concerned about regulatory exposure from documentation gaps, source-traceability is what makes AI querying defensible rather than risky. Every Atlas response links back to the source meeting transcript, document, or back office record, so you can verify the context before making a client decision.

Verifying AI Answers with Source Records

When Atlas returns an answer, each statement carries a citation. You click through to the source: the meeting transcript paragraph, the fact-find section, or the back office field. You confirm the context matches what Atlas reported. This takes seconds, not hours, and it replaces blind trust with professional verification.

Verifying Answers Against Client Records

The recommended workflow is: review the Atlas answer, check the source citation, confirm against the back office record, then decide your action. Atlas does not replace your judgment. It surfaces the information, you decide what to do with it.

For a broader perspective on how AI fits alongside professional judgment in regulated advice, AdvisoryAI's CEO Alan Gurung discussed how AI and advisers work together in a conversation with Intelliflo's Nick Eatock. If Atlas flags a client as having unused ISA allowance, you open the cited back office record, confirm the contribution history, and then decide whether to reach out. The professional responsibility stays with you.

Ensuring FCA-Compliant Audit Trails

A query log with source citations creates a defensible record of how you identified and acted on client needs. FCA reviewers assess files against the Investment Advice Assessment Tool, looking for personalised justification, complete client profiling, and a clear audit trail. When you act on an Atlas query, you can document the query, the answer, the source citation, and your action. That chain supports the audit trail requirement without additional manual reconstruction.

Use AI to Detect Upcoming Review Meeting Triggers

Book-wide queries enable proactive client care. Instead of reacting to client calls, you can anticipate needs based on what your data already contains.

Using Data for Targeted Client Care

Query results feed directly into a prioritised outreach list. Clients with unused ISA capacity receive a targeted email before tax year end. Clients approaching retirement get a review meeting scheduled. You can filter the results, draft the outreach email from Atlas, and hand off to Outlook or Gmail with recipient and subject prefilled. The shift is from "who should I call this week" to "which clients need what, and when."

Uncovering Hidden Client Sales Triggers

Meeting sentiment and unstructured mentions surface opportunities that structured fields miss. For example, a client may mention a bonus in a meeting but have no updated income on file, and Atlas flags the discrepancy. Or a client references a property sale with no corresponding fact-find update. These are sales triggers that memory alone would lose. The demand exists: 62% of investors would welcome more help managing their investments, rising to 68% when reviewing them, so the constraint is your capacity to see it.

Surface Clients at Risk of Churn

Atlas can help flag clients with overdue reviews or changing engagement patterns. Retention is relational, but early warning is operational. Catching a disengaged client before they leave preserves the relationship and the referral potential that comes with it.

How to Search Your Full Client Records with AI

This is a three-step workflow you can follow today.

Step 1: Query Your Client Data Using AI

Start with a specific question. Use the prompts from the earlier section or adapt them to your book. The Atlas home page offers one-tap starter prompts covering common queries, or you can type your own. Book-wide prompts return data only for connected back offices: Intelliflo, Plannr, and Curo today.

Step 2: Validate AI Claims with Citations

Open the source for every answer. Confirm the context before acting. If the answer seems unexpected, review the reasoning behind it. If Atlas cannot find sufficient data to answer, it indicates this rather than generating unsupported responses.

Step 3: Automate Your Client Outreach

Atlas can draft outreach emails from query results. Sending opens Outlook or Gmail with recipient and subject prefilled, though the body may not carry across and often requires pasting. The adviser retains control of all client communication. Atlas Workflows, which would automate multi-step actions such as filtering clients and updating the back office, is on the roadmap and not yet live. The current capability is Atlas drafting a single outreach email and handing it off to Outlook or Gmail, not an automated workflow. Firms should confirm current availability directly with AdvisoryAI.

Meeting FCA Standards During Client Book Reviews

AI querying is not a compliance risk if the audit trail is sound. The FCA's concern is not whether you use AI, it is whether you can evidence good outcomes and maintain records that demonstrate how you reached your decisions.

Meeting FCA Consumer Duty Requirements

The FCA's Consumer Duty policy statement PS22/9 focuses on four outcomes: products and services, price and value, consumer understanding, and consumer support. Book-wide queries support all four by identifying underserved clients before they become complaints. The FCA will consider whether senior leaders are using all available data to deliver these outcomes effectively. If you are evaluating firm-wide ROI across a network, consolidator, or investment management firm, book-wide queries reduce the manual analysis time that currently pulls advisers away from client work. Atlas makes that data accessible.

How to Document AI-Led Queries

Consider recording key items in your client file note: the query you asked, the answer Atlas returned, the source citation, and the action you took. The reasoning behind every Atlas response persists with the conversation, so you can review how a query was resolved if your compliance process requires it. This is not additional paperwork, it is the record of work already done. Keep comprehensive records of all client interactions to demonstrate compliance with FCA Handbook requirements.

Keeping Client Data Secure in the UK

AdvisoryAI holds Cyber Essentials Plus certification, with ISO 27001 in progress. Annual independent penetration testing is conducted by Predatech (CREST-accredited), with continuous attack-surface monitoring through SecurityScorecard (Grade A). The platform carries £2m cyber insurance through AIG UK and maintains a formal incident-response plan with critical-alert activation within 15 minutes and customer notification within 1 hour. Data is processed on UK and EEA infrastructure, never on US servers, and client data is never used to train or fine-tune AI models.

Data transmission uses TLS 1.3 encryption, with AES-256 encryption at rest and AWS KMS key management. Audio recordings are retained for 12 months and transcripts for 36 months. The platform runs on multi-AZ infrastructure with 4-hour RTO and 15-minute RPO, with a 30-day data export window on termination. AdvisoryAI's AI Framework for Advice Firms sets out the full governance approach, including human-review checkpoints and incident management.

Request a demo to see how Atlas works with your client book and back office, or start a 14-day free trial with no credit card required.

FAQs

Can Atlas Query Across Multiple Advisers' Clients in One Search?

Atlas queries Evie meeting transcripts, Emma-generated reports, uploaded documents, and synced back office data from Intelliflo, Plannr, and Curo. Cross-adviser meeting access is licence gated to Paraplanner and Paraplanner Plus, scoped by Team Management. Atlas covers one adviser's meetings per question and cannot span several in one query.

How Accurate Are the Answers Atlas Provides?

Atlas grounds every answer in your firm's own data with source citations. Accuracy depends on the completeness of your firm's meeting records and back office data. If meeting notes are incomplete or back office data is not synced, answers will be incomplete.

What Happens if Atlas Can't Find an Answer?

Atlas says so rather than guessing. It does not generate plausible-sounding responses from general knowledge. If the data is not in your records, the answer reflects that.

How Long Does It Take to Query Your Whole Client Book?

Typically under two minutes for a book-wide query. The same analysis using manual spreadsheet exports and back office filtering typically takes significantly longer.

Does This Replace My Back Office Reporting?

No. Back office reporting typically handles structured questions well. Atlas is designed to add value on unstructured questions, such as meeting mentions and sentiment, that structured fields typically cannot capture.

How Does Atlas Handle Customisation for Different Firm Workflows?

Atlas queries your firm's own data: Evie meeting transcripts captured in your firm's template format, Emma reports built from your templates, and synced back office data from Intelliflo, Plannr, or Curo. The queries adapt to what exists in your records.

What Happens to My Data When I Query Atlas?

Your client data stays within your tenant. Atlas queries only the data you have permission to access, scoped by your role and Team Management settings. Data is never used to train or fine-tune AI models, and all queries and answers are logged for audit purposes.

Key Terms Glossary

Atlas: The AI chat and intelligence layer across the AdvisoryAI platform. Advisers ask plain-English questions and receive cited answers from meetings, documents, and back office data.

Adaptive Thinking: AdvisoryAI's May 2026 Atlas feature that shows the step-by-step reasoning behind every response, with live status updates and a persistent, auditable record.

Back office: The client management systems advisers use daily, including Intelliflo, Plannr, Curo, and Iress Xplan.

Consumer Duty: The FCA's 2023 regulation requiring firms to deliver good outcomes for retail customers across products and services, price and value, consumer understanding, and consumer support.

Suitability report: The document that records the adviser's recommendation and the reasons behind it, required for FCA compliance.

Fact-find: The structured record of a client's personal and financial circumstances, objectives, and attitude to risk.

LOA pack: The Letter of Authority and supporting documents used to gather information from product providers.

Annual review: The yearly meeting where adviser and client review progress, update the fact-find, and adjust the plan.

Audit trail: The documented record of advice decisions, client interactions, and compliance checks that demonstrates FCA adherence.

Source citation: The link in an Atlas answer that points back to the specific meeting transcript, document, or back office record supporting the statement.