<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>ChrisMc</title>
    <link>https://chrismc.com/blog</link>
    <description>Twenty years of building ventures and putting technology to work inside real businesses. Free frameworks, books and hands-on help to make AI actually pay.</description>
    <language>en-GB</language>
    <atom:link href="https://chrismc.com/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Will AI replace accountants? The profession stays. Most of the people do not.</title>
      <link>https://chrismc.com/blog/will-ai-replace-accountants</link>
      <guid isPermaLink="true">https://chrismc.com/blog/will-ai-replace-accountants</guid>
      <pubDate>Thu, 19 Nov 2026 00:00:00 GMT</pubDate>
      <description>Will AI replace accountants? The profession stays. Most of the people do not. A typical 20-person firm scored live leaves eight roles. Here is the rubric.</description>
      <content:encoded><![CDATA[<p>Will AI replace accountants? If you mean whether the profession will exist in ten years, no: somebody has to sign the return and carry the liability. If you mean whether the people currently employed as accountants will still be employed as accountants, then for most of them the answer is also no. In the video below I score a typical 20-person firm with the same EmbedRight rubric I use on client work. Eight of the twenty roles survive. The only thing holding that arithmetic back is one partner meeting.</p><figure><iframe src="https://www.youtube-nocookie.com/embed/Wco_ocHCuao" title="Accountants Are Being Replaced. I Scored a Real Firm to See Who Survives" loading="lazy" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" referrerpolicy="strict-origin-when-cross-origin" allowfullscreen></iframe><figcaption><a href="https://www.youtube.com/@chrismcthrive">Chris McMenemy on YouTube</a></figcaption></figure><p><strong>Chapters</strong></p><ol><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=0s">0:00 Will AI replace accountants</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=71s">1:11 The canvas, run on a real firm</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=196s">3:16 Layer 1: the Value Map</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=240s">4:00 Layer 2: the Knowledge Map</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=307s">5:07 Layer 3: the Friction Map</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=388s">6:28 Layer 4: the Human Layer</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=507s">8:27 Phase 2: scoring it live</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=596s">9:56 The arithmetic: twenty to eight</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=653s">10:53 The pipeline problem</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=700s">11:40 The timeline</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=734s">12:14 Why this is happening now</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=879s">14:39 Friction Map to Threat Map</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=908s">15:08 Where your value is</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=1006s">16:46 The four paths</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=1255s">20:55 Your next 30 days</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=1348s">22:28 The hard part</a></li><li><a href="https://www.youtube.com/watch?v=Wco_ocHCuao&amp;t=1403s">23:23 Which one is yours</a></li></ol><h2>Will AI replace accountants, or only some of the jobs?</h2><p>It replaces the jobs that are mostly made of work a machine can already do, not the profession itself. Clients of a typical firm are buying three things: a return filed on time, a tax figure they can trust, and somebody to call when something changes. Only one of those is a document.</p><p>Knowing the rules is already a reference task. Judgement lives in about five heads in a firm of 20. The time bleeds out in chasing, re-keying, reconciliation, queries, prep and formatting. In the firm I scored, that was ten of the 20 people: the eight staff accountants on production, and two of the three on admin. The Human Layer is six people, none of them also in the friction layer.</p><p>The same pattern shows up in <a href="https://chrismc.com/blog/ai-in-law-firms">law firms</a>: automate the certainty the client is actually buying, and you remove the reason they were paying you.</p><h2>How do you score which jobs AI reaches first?</h2><p>With four questions, two points each, for a maximum of eight. Any job that scores six or more is one the software reaches first.</p><table><thead><tr><th>Question</th><th>Points if yes</th></tr></thead><tbody><tr><td>Does the work follow a predictable pattern?</td><td>2</td></tr><tr><td>Does it happen often enough to be worth changing?</td><td>2</td></tr><tr><td>Do written records show how it should be done?</td><td>2</td></tr><tr><td>If the output is occasionally wrong, are the consequences survivable?</td><td>2</td></tr></tbody></table><p>Document chasing scored eight: the same documents every quarter, perhaps 50 times a week, a checklist on file, and a wrong chase noticed before anything is filed. Reconciliation and re-keying also scored eight. Telling a client something they do not want to hear scored zero. The three highest-scoring rows were the job description of eight of the 20 people.</p><blockquote>The only thing standing between your firm and that arithmetic is somebody in your partner meeting realising it is possible, and deciding to go ahead.</blockquote><h2>What does the arithmetic look like in a 20-person firm?</h2><p>It is a halving, not a 10% efficiency story. The ten people whose week sits inside the Friction Map go. The two partners and three managers stay. You keep two of the four seniors, to review the output and hold the judgement, and one person on admin. That is eight out of 20, doing the same volume of work, or more, because the bottleneck was the prep.</p><p>Every good senior learned judgement by doing about 3,000 hours of that friction work. Take that away and, in nine years, the firm has no seniors of its own. That is a problem for the partnership, not a reason to wait if you are in the ten.</p><p>There is no date you can circle, and you do not get 18 months&#39; notice. BambooHR&#39;s May 2026 finance analysis found US senior-to-entry hiring at 3-to-1 (<a href="https://www.bamboohr.com/resources/data-at-work/data-stories/april-finance-2026">BambooHR</a>). KFF&#39;s 2026 survey of 1,937 employers put average family health cover at $28,303 a year, up 5% (<a href="https://www.kff.org/health-costs/2026-employer-health-benefits-survey/">KFF</a>). The gate is a partner meeting, not a laboratory breakthrough.</p><p>If you run the firm, the next step is <a href="https://chrismc.com/blog/how-to-implement-ai-in-an-accountancy-firm">how to implement AI in an accountancy firm</a> without automating the reason clients stay.</p><h2>What are the four paths that survive?</h2><p>&quot;Learn AI&quot; is not one of them. The four ThriveRight paths I walk through in the video are the Orchestrator, the Builder, the Creator and the Human. Three involve leaving.</p><ol><li><strong>The Orchestrator.</strong> Stay and run the systems, rather than be the person they replace. Almost everybody picks this, which is the problem: two seats, and 12 people who need one. The people who get these seats had already automated a piece of their own work, six months before anybody asked, and could show the partners what it saved.</li><li><strong>The Builder.</strong> Leave and own something physical. After enough sets of financials you often know which local businesses actually make money. Equipment rental, commercial cleaning, excavation, waste hauling, pest control: work that still requires someone to turn up.</li><li><strong>The Creator.</strong> Build an audience so your income does not depend on one employer&#39;s decision. As I say in the video, a search for AI tools for accountants on YouTube often surfaces accountants with audiences, not software companies.</li><li><strong>The Human.</strong> Go where the value is that it is a person and not a model: coaching, teaching, training, examining, CPA instruction or review.</li></ol><p>Hoping to be one of the two Orchestrator seats is the same thing as having no plan. If you do nothing, you become the person the system gets built on top of: the one asked, nicely, to document the process.</p><h2>What should you do in the next 30 days?</h2><p>Start the same way, whichever path you pick:</p><ol><li>For one week, write down your actual week: every task, honest hours. The sheet is for you alone, never for the firm.</li><li>Score each line with the four questions above.</li><li>Add up the hours that score six or more. That is the proportion of your job that is leaving.</li><li>Keep the hours that score low. That is what you build on.</li></ol><p>The gap between starting in 2026 and starting in 2029 is the difference between arriving early and arriving with everybody else.</p><h2>Questions people ask</h2><h3>Will AI replace all accountants?</h3><p>No. It replaces the work that scores six or more on the rubric. Someone still has to sign the return, carry the liability and have the conversations clients actually pay for.</p><h3>Which roles in a typical firm are most exposed?</h3><p>Staff production work and most of the admin and bookkeeping support. In the firm I scored, that was ten of the 20 people.</p><h3>If I want to stay, how do I get one of the remaining seats?</h3><p>Do not wait to be asked. Automate a piece of your own work on real client files, and show the partners what it saved.</p><h3>What if I run the firm?</h3><p>Fill in the free <a href="https://chrismc.com/embed">EmbedRight canvas</a> honestly, then use <a href="https://chrismc.com/blog/how-to-implement-ai-in-an-accountancy-firm">how to implement AI in an accountancy firm</a> to change the work without automating the advice.</p><h2>Where to start</h2><p>If you want your version of the second map without doing the 30 days on your own, take the <a href="https://chrismc.com/thrive-quiz">ThriveRight quiz</a>. It will tell you which of the four paths fits what you actually do all day. If you run the place, download the <a href="https://chrismc.com/embed">canvas</a> and fill it in with the people who do the work. Either way, pick one of the four. Not next quarter. Straight away.</p>]]></content:encoded>
      <dc:date>2026-11-19</dc:date>
      <author>media@chrismc.com (Chris McMenemy)</author>
    </item>
    <item>
      <title>How to implement AI in an accountancy firm: start with the work</title>
      <link>https://chrismc.com/blog/how-to-implement-ai-in-an-accountancy-firm</link>
      <guid isPermaLink="true">https://chrismc.com/blog/how-to-implement-ai-in-an-accountancy-firm</guid>
      <pubDate>Thu, 19 Nov 2026 00:00:00 GMT</pubDate>
      <description>How to implement AI in an accountancy firm: map the real work, score each task on impact and AI suitability, then automate the overhead and protect the advice.</description>
      <content:encoded><![CDATA[<p>How to implement AI in an accountancy firm comes down to one rule: start with the work, not the tool. Map what clients actually pay you for, where the know-how lives, where the hours leak out and what must stay human, then score every finding on impact against AI suitability and do the top three. In most practices those three are compliance overhead such as chasing records, capturing receipts and reconciling bank lines, not advice.</p><h2>Why does AI so often fail to stick in accountancy firms?</h2><p>It fails because it is bought as a tool, not designed into the work. A partner sees a demo, the practice buys seats, and six months later the old way is still the default because nobody decided which job the tool was for.</p><p>Accountancy has an extra twist. Many firms already pay for software with machine learning built in. Dext reads supplier, date, total and tax from receipts and invoices, and validates UK VAT numbers against HMRC (<a href="https://help.dext.com/en/articles/416691-what-information-does-dext-extract-from-costs-and-sales-documents">Dext</a>). Xero&#39;s JAX reconciles bank lines automatically only when it is highly confident, and suggests a match for the rest (<a href="https://www.xero.com/uk/accounting-software/reconcile-bank-transactions/">Xero</a>). The question is rarely &quot;which AI should we buy?&quot; It is &quot;which of our jobs should this take over, and which should it never touch?&quot;</p><p>The test I use is the same one I apply in legal: a human must be able to check the output faster than they could have produced it. A reconciliation suggestion passes. A tax planning recommendation, generated from a model&#39;s general knowledge, does not.</p><h2>How do you implement AI in an accountancy firm, step by step?</h2><p>Work through the four maps of the <a href="https://chrismc.com/frameworks/embedright">EmbedRight canvas</a> in order, then score:</p><ol><li><strong>The Value Map: what are clients actually paying for?</strong> Not the fee line. Year-end done without chasing, the same accountant every year, tax advice before it is too late. &quot;Quality&quot; does not count.</li><li><strong>The Knowledge Map: where does the expertise live?</strong> Name the place. If only one person knows how R&amp;D claims or payroll exceptions are handled, star it.</li><li><strong>The Friction Map: where do time and money leak?</strong> Count it: times a week, minutes each. Chasing client records, bank reconciliation queries, re-keying receipts, engagement letters by hand.</li><li><strong>The Human Layer: what must stay human?</strong> Telling a client they owe HMRC. The planning meeting before year-end. Spotting the worry behind a question.</li><li><strong>Score and rank.</strong> Impact from 1 to 5, multiplied by AI suitability from 0 to 8. Two points each for a predictable pattern, enough volume to justify building, written records of how it is done, and occasional imperfection being manageable. Keep the top three.</li></ol><blockquote>The client does not pay you for the reconciliation. They pay you for the phone call when the numbers look wrong.</blockquote><h2>Which accountancy tasks is AI actually good at?</h2><p>AI is good at high-volume, rule-shaped work where a qualified person can check the result quickly. It is poor at judgement, and at anything a client will treat as advice.</p><table><thead><tr><th>Task</th><th>AI suitability</th><th>Why</th><th>Where it often already lives</th></tr></thead><tbody><tr><td>Chasing client records</td><td>High</td><td>Predictable, frequent, easy to check</td><td>Practice management and email</td></tr><tr><td>Receipt and invoice capture and coding</td><td>High</td><td>Structured fields, learns from past coding</td><td>Dext and similar capture tools</td></tr><tr><td>Bank reconciliation</td><td>High</td><td>Rules and history; low-confidence lines left for review</td><td>Xero JAX and other ledgers&#39; matching</td></tr><tr><td>Long client email threads into file notes</td><td>Medium to high</td><td>Summary is quick to check against the thread</td><td>Copilot in Outlook can summarise a thread with citations to each email (<a href="https://support.microsoft.com/en-us/outlook/copilot-pages/summarize-an-email-thread-with-copilot-in-outlook">Microsoft</a>)</td></tr><tr><td>Engagement letters and routine letters</td><td>Medium</td><td>Works from your own templates, not from scratch</td><td>Document templates</td></tr><tr><td>Tax planning and advice</td><td>Low</td><td>Judgement, and liability sits with you</td><td>Keep it human</td></tr></tbody></table><h2>What does Making Tax Digital change?</h2><p>It raises the volume of the work AI handles best. Sole traders and landlords with qualifying income over £50,000 had to use Making Tax Digital for Income Tax from 6 April 2026, followed by those over £30,000 from April 2027 and over £20,000 from April 2028 (<a href="https://www.gov.uk/guidance/find-out-if-and-when-you-need-to-use-making-tax-digital-for-income-tax">GOV.UK</a>). Those clients keep digital records and send quarterly updates through compatible software.</p><p>For a practice, that means more of the chase, capture and reconciliation cycle happens four times a year rather than once. That strengthens the case for automating the overhead now.</p><h2>What should stay human in an accountancy practice?</h2><p>Judgement, bad news and anything that rests on trust. ICAEW&#39;s guidance applies the five fundamental ethics principles to generative AI, and warns specifically against automation bias, over-reliance on tool output, and putting confidential client data into tools without thought (<a href="https://www.icaew.com/technical/technology/artificial-intelligence/generative-ai-guide/ethics">ICAEW</a>). Its getting-started guide recommends involving legal, data protection and HR expertise early (<a href="https://www.icaew.com/technical/technology/artificial-intelligence/generative-ai-guide/getting-started">ICAEW</a>).</p><p>That lines up with the Human Layer. A practice that automates the phone call when a client&#39;s business is struggling has automated the reason the client stays.</p><p>If you are an accountant wondering what this means for your own job, see <a href="https://chrismc.com/blog/will-ai-replace-accountants">will AI replace accountants</a>.</p><h2>What does this look like on a real canvas?</h2><p>The EmbedRight page includes an illustrative canvas for a regional accountancy practice, not a client. Its top three were chasing client records (impact 4 × suitability 8 = 32), receipt capture and coding (3 × 8 = 24) and engagement letter drafts (3 × 6 = 18). Telling a client they owe HMRC, the planning meeting before year-end and the call when a business is struggling sat in the Human Layer, and stayed off the plan.</p><p>The same pattern holds in <a href="https://chrismc.com/blog/ai-in-law-firms">law firms</a>: the return is in removing non-professional work from expensive people, not in making the professional work faster.</p><h2>Questions people ask</h2><h3>Is our practice too small for this?</h3><p>No. Small practices often gain most, because there is no back office to absorb the chasing and re-keying. Start with tools you already pay for.</p><h3>Will AI replace our bookkeepers?</h3><p>It changes the role first. Data entry shrinks and reviewing, querying and client contact grow. Firms that cut first and redesign later tend to rehire.</p><h3>Is it safe to put client data into AI tools?</h3><p>Treat it as a professional obligation, not an IT question. Find out where data is processed, whether it is retained or used for training, and who is liable, and get the answers in writing.</p><h3>Where should we start?</h3><p>With the Friction Map. Count the hours on chasing, capture and reconciliation, then check whether your existing software can already take them on.</p><h2>Where to start</h2><p>Not with a tool. Download the free <a href="https://chrismc.com/frameworks/embedright">EmbedRight canvas</a>, give it an afternoon with your partners and practice manager, and keep the top three. If you would like a hand running it, <a href="https://chrismc.com/work-with-me#enquire">get in touch</a>.</p>]]></content:encoded>
      <dc:date>2026-11-19</dc:date>
      <author>media@chrismc.com (Chris McMenemy)</author>
    </item>
    <item>
      <title>AI in law firms: the four jobs it is actually good at</title>
      <link>https://chrismc.com/blog/ai-in-law-firms</link>
      <guid isPermaLink="true">https://chrismc.com/blog/ai-in-law-firms</guid>
      <pubDate>Mon, 02 Nov 2026 00:00:00 GMT</pubDate>
      <description>Most legal AI tools are sold on everything they can do. Four jobs are worth your money, and they share one property. A guide for small and mid-sized firms.</description>
      <content:encoded><![CDATA[<p>AI is reliably good at four jobs in a law firm: finding things in large volumes of documents, drafting documents the firm has already written a hundred times, turning unstructured material into structured records, and absorbing the administrative work fee earners should never have been doing. Everything else is a demonstration, not a business case.</p><h2>The test that separates the four from the rest</h2><p>Those four jobs have one property in common, and it is the only test worth applying to any legal AI tool you are shown.</p><blockquote>A human can check the output faster than they could have produced it.</blockquote><p>That is the whole thing. When checking is quicker than doing, AI makes a fee earner faster and the risk stays bounded, because a qualified person still signs it off and the error surfaces before it leaves the building. When checking takes as long as doing, you have bought nothing. And when the output cannot be checked at all without redoing the work from scratch, which is the case for anything that invents citations or quietly omits what it did not find, you have bought a liability and put your name on it.</p><p>Run that test over the brochure the next time someone demonstrates a tool to you. Most of what gets pitched to law firms fails it. The four below pass.</p><h2>1. Finding things in volume</h2><p>Disclosure, due diligence, bundle review, any exercise where the answer is somewhere inside twelve thousand documents and somebody has to go and get it.</p><p>This is the strongest case in legal, and it is strong because the job is pattern matching across more material than a person can hold. The model reads everything, flags what matches, and hands back a shortlist. A solicitor reads the shortlist. The checking is genuinely faster than the doing, by a wide margin, because reading two hundred flagged passages is not the same job as reading twelve thousand documents.</p><p>The constraint is that recall is never perfect and you must assume it is not. The tool is a first pass that narrows the field, not a guarantee that nothing was missed. Firms that treat it as the latter are the ones that get hurt. Scope it as triage, keep the human review on the shortlist, and sample what was rejected.</p><p>Worth knowing: this is also the category where the tooling is most mature and most expensive. You are often buying something your document management vendor already sells as a module.</p><h2>2. Drafting what you have drafted a hundred times</h2><p>Standard-form contracts. Leases. Letters of claim. Engagement letters. Section 21 notices. The documents where the structure is settled, the variations are known, and the firm has produced hundreds of near-identical versions already.</p><p>The value here is almost entirely in your own precedent bank, not in the model&#39;s general knowledge of law. A model drafting from first principles produces something that reads like a contract and is not one. A model drafting from your firm&#39;s precedents, with your clauses and your house positions, produces a first draft a fee earner can correct in ten minutes.</p><p>That distinction decides whether this works. If a tool cannot be pointed at your own precedents, it is doing the wrong version of this job.</p><p>The constraint is that it is useless for novel drafting, and novel drafting is where your margin is. This job takes the routine end of the work off the desk so the people who can do the other kind have time to do it. It does not do the other kind.</p><h2>3. Turning unstructured material into structured records</h2><p>This is the one nobody pitches and it is quietly the biggest time sink in a working practice.</p><p>Attendance notes into matter records. Email threads into a chronology. A scanned bundle into something searchable. A client&#39;s shoebox of correspondence into a dated index. Voice notes from a hearing into a file note.</p><p>Nobody sells this as artificial intelligence because it does not demonstrate well. It is also the work that fee earners complain about most, that gets done badly at seven in the evening, and that costs the firm real money in write-offs when it is not done at all.</p><p>The checking test passes comfortably. Reading a generated chronology against the source emails is far quicker than building one. And the failure mode is benign: a wrong date in a chronology gets caught by the person who knows the matter.</p><p>If you do one of these four first, do this one. It is the cheapest to implement, it does not touch advice, and it is the one your fee earners will thank you for.</p><h2>4. The work that was never legal work</h2><p>Time recording. File opening. Conflict check preparation. Client onboarding and the identity documents that go with it. Billing narratives. Chasing the things that need chasing.</p><p>None of this requires a law degree and most of it is done by people who have one. It is overhead that attached itself to fee earners because there was nobody else to do it and no system that could.</p><p>This is the least glamorous of the four and probably the highest return, because you are not making legal work faster, you are removing non-legal work from expensive people. The business case does not depend on the model being clever. It depends on it being adequate and cheap, which it now is.</p><h2>The four it is bad at, and why</h2><p>Worth being equally plain about the other side, because the firms that get burned are the ones who were told there was no other side.</p><p><strong>Legal research it cannot show its working for.</strong> The failure mode here is not that the model is wrong occasionally. It is that it is wrong in a way that looks exactly like being right. A fabricated citation is formatted identically to a real one. This is not hypothetical: in <em>Mata v. Avianca</em> in 2023, lawyers in New York were sanctioned after filing a brief containing case citations that did not exist, produced by ChatGPT. Research tools that retrieve and link to the actual source are a different proposition from a model answering from memory. Know which one you are buying.</p><p><strong>Judgement.</strong> Whether to settle. What this client can actually afford to risk. Whether the other side is bluffing. These are not information problems and more information does not solve them.</p><p><strong>Anything privileged going into a tool you do not control.</strong> Where the data goes, who trains on it, and what the contract says about retention are not IT questions, they are professional obligation questions. Get them answered in writing before anything client-related is typed into anything.</p><p><strong>Advice.</strong> Obvious, and still worth writing down, because the distance between a well-drafted document and advice is shorter than it looks once a tool is sitting in the workflow and people start leaning on it.</p><h2>What this looked like in practice</h2><p>I run the technical side of One Scriptum, which began as document automation for a single practice.</p><p>The thing worth taking from it is where we started, because it was not with a tool. It was with the data model, and with what that firm actually did differently from every other practice on the street. Those two questions in that order. A practice that has worked out what its clients are genuinely paying for can automate the rest with some confidence, because it knows what must not be touched. A practice that has not will automate the thing that made it worth instructing, and will not notice for a year.</p><p>That is the whole argument of <a href="https://chrismc.com/frameworks/embedright">EmbedRight</a> and <a href="https://chrismc.com/blog/how-to-implement-ai-in-an-accountancy-firm">it is not specific to legal</a>. It is just unusually visible in legal, where the difference between two firms on the same street is rarely the documents.</p><h2>Where to start</h2><p>Not with a tool. Four questions, in this order.</p><ol><li><strong>What are clients actually paying this firm for?</strong> Not what you bill for. What makes them instruct you rather than the firm next door.</li><li><strong>Where does the week actually go?</strong> Not the recorded time. The hour at seven in the evening that never gets recorded.</li><li><strong>Of that, what could a competent person check faster than they could do?</strong></li><li><strong>What must stay human,</strong> because it is the answer to question one?</li></ol><p>Those four questions are the first half of the EmbedRight canvas. The second half scores what you found on impact against how suitable it actually is for AI, and ranks it, so you come out with the two or three things worth doing rather than a list of forty you will never start.</p><p><strong>One firm&#39;s canvas, filled in</strong> An illustrative firm, not a client: fifteen people on a high street, doing conveyancing, private client and some civil litigation. Two findings scored but left off the plan: drafting from the precedent bank (4 × 4 = 16, tidy the precedents first) and legal research (3 × 2 = 6, leave alone).</p><p><strong>The EmbedRight canvas</strong> It is free and it takes an afternoon.</p><h2>Questions people ask</h2><h3>Is our firm too small for this?</h3><p>The third and fourth jobs on this list get better value at small scale, not worse, because a small firm has no back office to hide the overhead in. The first job, document review in volume, is the one that genuinely needs scale to justify the licence.</p><h3>Will this replace our paralegals?</h3><p>It changes what the role is before it changes how many of them you need. The parts that are reading and summarising get faster. The parts that are chasing, judging and dealing with people do not. Firms that cut first and work out the role afterwards tend to rehire.</p><h3>What about client confidentiality?</h3><p>Treat it as a contractual question, not a technical one. Where is the data processed, is it retained, is it used for training, and who is liable if it leaks. Get the answers in writing. A tool that cannot answer those four in writing is not a candidate.</p><h3>How much should we expect to spend?</h3><p>Less than you are quoted, if you start with the third or fourth job on this list. Both can usually be built on tooling the firm already pays for. The expensive category is document review in volume, and that is the one to do last unless disclosure is your main business.</p><h3>We tried an AI tool and it did not stick. What went wrong?</h3><p>Usually that it was bought as a tool rather than designed into a workflow, so using it stayed optional and the old way stayed available. The second most common reason is that it was pointed at the interesting work instead of the boring work.</p>]]></content:encoded>
      <dc:date>2026-11-19</dc:date>
      <author>media@chrismc.com (Chris McMenemy)</author>
    </item>
  </channel>
</rss>
