Today's AI Specialist: The Loom Personalisation Agent. The Agent That Suggests Where to Loom Without Filming Anyone.
Listen to this article
Narrated by Admin Pipeline Hub · 7 min read audio
Every new consultation lead on EnglishFluency.Online receives a Custom Conversation Pack within 24 hours of booking. The pack is built from the prospect's own diagnostic chat: their struggle moment, their desired outcome, their L1, their CEFR estimate. It is the most personalised artefact in the funnel. And for some prospects, but not all, it includes a personal video from me.
The agent that decides which prospects get a video, and what the video should contain, is the Loom Personalisation Agent.
It does not record anything. It writes a one-page brief: which pack warrants a Loom, which specific moment from the prospect's diagnostic to open with, and the three sentences the video needs to land. I record the video against that brief. The agent owns the brief, not the camera.
This is the build story of the Loom Personalisation Agent. Why a personalisation gate exists, what the brief actually contains, and the single rule that decides whether the video helps or hurts.
The problem the agent solves
There is a temptation in any pack-style funnel to give every prospect every possible touchpoint. Written pack, personal video, scheduled call, follow-up email, second pack. The instinct is that more is more. The instinct is wrong, and it is wrong in a specific way.
A generic personal video feels worse than no personal video. A founder who records a thirty-second "thanks for booking, here is what to expect" Loom that is identical for every prospect is signalling, very clearly, that the personalisation is performative. The prospect notices the lack of specificity instantly, and the rest of the pack suffers from the same suspicion.
The opposite is also wrong. A founder who tries to record a custom video for every prospect ends up either burning forty minutes a day on Looms or shipping low-energy videos at the end of long days when the production budget has run out. Both outcomes degrade the pack.
The Loom Personalisation Agent exists to solve this by gating. It decides which prospects warrant a Loom, and the gate is set high. Most packs do not get a video. The ones that do, get one that is worth recording.
What goes into the gate
The agent reads the diagnostic transcript, the pack content, and a small handful of prior Looms that worked, and applies four criteria.
First, is there a specific moment in the diagnostic that text cannot land? Some struggle moments are emotionally loaded. A prospect describing a meeting where they froze in front of their CEO, a candidate who failed an English interview they thought they would pass, a parent who is embarrassed about their English with their teenage child's friends. These moments carry weight that written text does not transmit at full strength. The agent flags them for video.
Second, is the prospect carrying a misconception the pack needs to correct? Some prospects arrive convinced their problem is grammar when their actual problem is fluency, or convinced they need C2 when C1 is what the role requires, or convinced fluency is mostly about vocabulary. Written correction is fine; video correction is better, because tone carries the reassurance that text cannot.
Third, is the prospect at a decision threshold? Some prospects are clearly going to book the consultation regardless of the pack; others are clearly not. The agent identifies the third group, the prospects sitting on the fence, where a personal video tips the decision. The criterion for this is heuristic, drawn from patterns the agent has been shown across past leads.
Fourth, is the scenario unusual enough that a generic answer would feel like a generic answer? Some prospects have unusual jobs, unusual L1s, unusual timeframes. A trauma surgeon preparing to present at an English-language conference. A Hungarian engineer relocating to Dublin in three weeks. A widow returning to professional English after eight years out of the workforce. These prospects can tell when the pack was clearly designed for them, and a video lands harder than text on the why.
The gate is set high. Most packs do not get a video. When the criteria do land, the agent recommends a Loom; when they land hard, the pack is surfaced at the top of the day's short Loom queue. Most days that queue is short on purpose.
The brief format
The brief the agent produces is one page, deliberately short. It anchors the video to a specific moment from the prospect's diagnostic (usually a phrase the prospect used to describe their struggle moment), names what the video has to do (correct a misconception, deliver reassurance only a person can offer, or reframe a wrong belief), and bridges back to the booked consultation slot so the video reads as a continuation of the pack rather than a separate piece of marketing.
That is the entire brief. No script. No scene-by-scene direction. A short structured prompt, and the rest is mine. The format was chosen after testing scripted versions; founders reading their own voice over a brief sound more natural than founders reading a script.
The rule that decides whether the video helps or hurts
There is one rule, and it overrides everything else.
If the video would have been recorded the same way for any prospect, do not record it.
This is the rule the agent applies in its gate, and it is the rule I apply when I read the brief. If the three landings in the brief are landings that would work for any pack, the brief has failed and the pack ships without the video. The Loom is the antithesis of generic. If it can be generic, it should not exist.
This is the rule that has saved me, in the first three months of running the pack pipeline, from recording approximately forty Looms that would have made the packs worse rather than better. The agent's job is to identify the prospects for whom the video adds something the pack alone cannot, and to refuse the rest.
TL;DR
The Loom Personalisation Agent decides which Custom Conversation Pack prospects warrant a personal video from the founder, and what the video should contain. It does not record anything. It writes a one-page brief that anchors the video to a specific moment from the prospect's diagnostic, names what the video must do, and bridges back to the booked consultation. The gate is set high; most packs do not get a Loom. The rule that overrides everything else: if the video would have been recorded the same way for any prospect, do not record it. Generic personalisation is worse than no personalisation.
The same allocation logic scales out. A 95-country direct-to-cart rollout for a healthcare brand required deciding which leads in which markets warranted which level of founder attention, on what gates, with what trigger conditions. The pack pipeline is the small version of an enterprise problem.
Language Analysis
Select a category above to highlight those words in the text.
Learning Materials
Key Vocabulary
personalisation
The act of tailoring something — content, a product, a message — to fit a specific individual rather than a general audience.
“The Loom Personalisation Agent decides which prospects warrant a video.”
performative
Done for show or appearance rather than out of genuine intent; signalling effort without substance.
“A generic personal video signals that the personalisation is performative.”
gate
To restrict access to something so it is given only when specific criteria are met.
“The agent gates Looms by four diagnostic-driven criteria.”
gate
A decision point or filter that controls what gets through to the next stage.
“The gate is set high; most packs do not get a video.”
brief
A short document instructing someone what to produce or address, without prescribing the exact words or steps.
“The agent writes a one-page brief, not a script.”
landing
In editorial and presentation usage, the key point a piece of communication must make stick with the audience.
“The brief is structured around three landings.”
threshold
The level at which something begins to take effect or a decision tips one way or the other.
“The agent identifies prospects at a decision threshold.”
prospect
A potential customer or client who has shown enough interest to be worth pursuing.
“Some prospects sit on the fence; a video tips the decision.”
misconception
A mistaken belief that the audience holds and that the communication needs to correct.
“Some prospects arrive with a misconception about what fluency means.”
reframe
To present an idea, problem, or experience inside a different frame so it is understood differently.
“Landing two reframes the misconception the prospect is carrying.”
SLA
Service Level Agreement: the promise about how quickly or to what standard a deliverable will be provided.
“The pack ships within the 24-hour SLA, with or without the video.”
recipient
The person who receives something — a message, a pack, a video.
“The recipient of a generic Loom notices the lack of specificity.”
generic vs specific
The contrast between language or content that could apply to anyone and language that addresses one person's situation.
“The rule pivots on the generic vs specific distinction: if it can be generic, it should not exist.”
queue
An ordered list of items waiting to be processed; in operations, what gets worked on next.
“High-priority Looms are surfaced at the top of the daily queue.”
recommendation
A suggestion, formally produced by a person or a system, on the action to take.
“If two criteria are met, the agent issues a Loom recommendation.”
Grammar Notes
Gerund as subject of the sentence
When the subject of a sentence is itself an action, English uses the gerund (the -ing form acting as a noun). 'Personalising every prospect' is the subject; 'feels' is the verb. This is the natural way to make an activity the topic of a statement, rather than forcing a longer noun phrase like 'the act of personalising every prospect'.
“'Personalising every prospect feels worse than personalising none.' (implicit in the post's argument that 'a generic personal video feels worse than no personal video')”
Common mistake: Switching to an infinitive after a sentence has begun with a gerund: 'Personalising every prospect feels worse than to personalise none' — gerund and infinitive must be parallel ('Personalising every prospect feels worse than personalising none').
Conditional 'if' clauses describing system rules
The zero conditional ('if + present simple, present simple') is used for rules that hold reliably: every time the condition is met, the result follows. It is the right voice for documenting a system rule — not 'would' (hypothetical), not 'might' (uncertain), but a flat statement that this is how the system behaves. The post uses zero conditional throughout when describing what the agent does under each criterion.
“'If the founder is not available to record, the pack ships without the Loom.'”
Common mistake: Mixing first conditional ('if the founder is not available, the pack will ship') with zero conditional. 'Will' adds a sense of futurity and prediction; for steady-state system rules, plain present simple is cleaner and more authoritative.
Rule-based imperative for engineering constraints
When stating a non-negotiable rule built into a system, English uses the bare imperative ('do not record it') rather than a softer modal ('you shouldn't record it' or 'it is best not to record it'). The imperative carries the weight of a constraint — not advice, not preference. The post uses this structure to frame the single overriding rule of the agent.
“'If the video would have been recorded the same way for any prospect, do not record it.'”
Common mistake: Hedging the rule with a modal: 'you probably shouldn't record it' or 'it might be better not to record it' undermines the rule's status. Engineering constraints are stated as imperatives because they are not optional.
Comprehension Questions
- 1.What are the four criteria the Loom Personalisation Agent applies, and how many must be met for a Loom to be recommended?
- 2.Why does the agent never block the pack on the video?
- 3.What is the single rule that overrides everything else, and why is it written as a bare imperative rather than a softer suggestion?
- 4.Why does the brief contain three landings rather than a full script?
- 5.Apply the four-criteria gate to a sales, marketing, or content personalisation decision in your own work. Pick one piece of optional personalised output you produce today (a video, a custom email, a tailored landing page) and describe how you would gate it.
Run your own diagnostic
Use the same Strategic Council I run my own decisions through. The assessment preview is free. The specific central human intelligence it is based on is verified in person during the call.
Start the free diagnostic →