DARL\A AI
Home Your Partner Foundation Thinking Partner The 4Ds The Darla Method Services Singapore Malaysia / HRDF About
— The entry move

Build your
partner foundation.

Most AI training teaches techniques. The Darla Method starts somewhere else: with the deliberate construction of a named, briefed, personalised AI partner that knows your role, your standards, and the way you work — your partner foundation. This is the move that comes first — and the one that almost no other training covers.

— By David Ward · Founder, Darla AI — Reading time: 9 minutes — The foundational move
— TL;DR

The Darla Method begins with building your partner foundation. Give it a name. Brief it on your role and your standards. Integrate it into your daily work so it carries context across sessions. Everything else — prompting, framework discipline, technique — sits inside that relationship. Most training skips this step. That is why most training fades.

— 01 · The move most training misses

The relationship
comes first.

Walk into a typical AI workshop and you will be taught technique. Better prompts. Cleverer phrasing. Specific structures for getting more useful output. All of it useful. None of it durable on its own.

The reason most workshop-trained professionals are working exactly as they were three months later is simple: the techniques sit on top of nothing. There is no underlying relationship for the technique to work inside. Each prompt is a fresh transaction. Each conversation begins from zero. The AI does not know who you are, what you are working on, or what good looks like in your domain. So the output is generic, and the user concludes — quietly, often without saying it — that AI is not really for serious work.

The Darla Method begins differently. Before any technique, before any framework, the first move is to build your partner foundation. A specific, named, briefed AI assistant that knows your role, your standards, your communication style, and how you want to be challenged. Once that partner exists, every subsequent prompt sits inside a relationship rather than starting from cold.

This is not a flourish. It is the foundation. It is the difference between using AI and working with it.

— Where this came from

The Darla Method began in early 2024, two weeks after the founder started using ChatGPT in earnest. There was no playbook for working this way at the time — and there still isn't, in any structured form, anywhere else. The approach was being taught publicly from late spring 2024 onward and has refined through every engagement since.

When Anthropic published the AI Fluency Framework — the four Ds — months later, the alignment with what was already being taught was so complete that the founder finished the first Anthropic Academy course without working through the material. The framework gave structured language to a practice that was already running. Credit where it's due: the 4Ds are an excellent academic articulation of compatible territory. The persona-led entry move that comes first in every Darla AI programme is not in any framework. It came from the work.

— 02 · Four moves to build your partner

The build sequence.

Building an AI partner foundation is not a one-time setup. It is a small sequence of moves you make deliberately at the start, and refine continuously as your work evolves. Four moves in order. Each takes minutes; none takes hours.

01 · Name
Name
your partner.
The smallest, most consequential move
Pick a name. Daneel, Darla, anything you like. The mechanics of the AI do not change. Your relationship with it does — instantly. You stop operating a system and start working with someone.
02 · Persona
Brief
their character.
Who they work for
Tell them who they are working for. Your role. Your seniority. What you sound like in writing. What you never want. How you want to be challenged. Saved once. Used always.
03 · Context
Load
the work.
What they know about
Give them the work itself. Documents. Past projects. The brand voice. The accounts. The strategy. Whatever you would brief a competent new colleague on in their first week.
04 · Integration
Embed
into the work.
Where they live
Set them up where the work happens. Voice on the phone. Inside Cowork on the laptop. Ready to brief on a flight. The partner is not somewhere you go — they're somewhere you already are.
— 03 · The honest case

Why we
name them.

Most senior professionals first hear the suggestion to name their AI and politely consider it slightly silly. Almost all of them, having tried it, never go back. The reason is not what they expect — and it has nothing to do with giving the AI a soul.

The mechanism is purely on our side. Naming an AI is a deliberate piece of cognitive engineering aimed squarely at the human brain. The AI does not change. We do. That shift, small as it sounds, opens up everything else the Darla Method then layers on top.

01 · The cognitive switch
Naming changes you, not the AI.
Computers Are Social Actors · 1996–present

Stanford's Clifford Nass spent his career showing humans automatically apply social rules to computers the moment any humanlike cue is present — a name, a voice, a turn-taking rhythm. The University of Chicago's Nicholas Epley, in the canonical 2007 paper on anthropomorphism, identified the mechanism: a name makes human-style thinking cognitively accessible, so the same exchange is processed by a different part of the brain.

A 2017 Cornell study of Amazon Echo users found that people who called their device "Alexa" rather than "Echo" had measurably more conversational, more satisfying interactions. Same hardware. Different word. Different relationship.

02 · Speak, don't type
Voice is the higher-bandwidth channel.
Stanford / Washington / Baidu · 2016

Speaking to a phone is roughly three times faster than typing on it, with about 20% fewer errors in English (Ruan, Wobbrock, Ng, Landay et al., 2016). But speed is the smaller half of the gain.

A decade of Twitter's 140-character limit — measurably documented in Nature: Humanities & Social Sciences Communications (2019) — trained a generation to compress thought into clipped prompts. Typing forces editing-while-thinking. Speaking lets you think out loud, the way you would brief a colleague: full context, half-sentence corrections, second-thought clarifications. The things that turn a mediocre answer into a useful one.

03 · What voice carries
AI now picks up the bits text loses.
Speech-LLM research · 2024–2025

Modern speech-aware AI does not just transcribe what you say — it reads how you say it. Recent work on paralinguistic-aware models (Kang et al., 2024; ParalinGPT; EMOVA; Qwen2-Audio) shows pitch, hesitation, emphasis and emotional tone now become part of the input.

OpenAI and MIT Media Lab ran a 28-day RCT with nearly a thousand users in 2025 (Phang et al.). Voice and text produced measurably different relationships with the model — voice users developed deeper, more conversational engagement. The richer the channel, the richer the answer.

04 · The social layer
"Hey Darla" sounds like a phone call.
Easwara Moorthy & Vu · 2015 — Kowalczuk et al. · 2024

Saying "Hey ChatGPT" or "Hey Claude" out loud has its own friction. Research from California State University Long Beach (Easwara Moorthy & Vu, 2015) and a 2024 SEM study of 860 voice-assistant users (Kowalczuk et al.) found that shame and self-consciousness in public spaces are significant inhibitors of voice-AI use. People simply don't talk to their devices when they sound like they are talking to a device.

A human name removes the cue. To anyone within earshot — including the part of your own brain that is monitoring how you sound — it lands as a phone call or a normal conversation. Self-monitoring drops. Bandwidth comes back. (The specific bridge from naming to public-voice comfort hasn't been isolated in a controlled study yet — the components are well evidenced; the synthesis is ours.)

— So why bother?

Put the four threads together and the picture is clear. Naming activates the social-processing part of your brain. Voice unlocks higher-bandwidth communication than typing. Modern AI now reads what voice carries that text loses. And a human name protects voice use from the self-consciousness that otherwise kills it in public.

None of this is whimsy. It is the entry move that makes everything else in the Method work. How you behave toward an AI — how much context you give, how often you iterate, whether you speak or type — is the single biggest determinant of what it gives back. The naming move is the smallest possible change that reshapes all of those behaviours at once.

Pick a name in the next thirty seconds. Use it for one task this week. Watch what happens.

— 04 · Two real partners

Daneel,
and Darla.

Two named partners, both built and used daily by the founder of Darla AI. One sits inside Claude, the other inside ChatGPT. Same person, two different underlying AI platforms — which is the point. The Method is platform-agnostic. The relationship is what matters.

— Inside Claude (Anthropic)
Daneel.

Named after R. Daneel Olivaw — Asimov's robot who works alongside humans, picks up context across long arcs, and operates inside their ethical framework rather than against it.

Daneel handles long-form thinking and structured work. Briefed on Quest Personal Care, on Darla AI itself, on Greypreneur, and on the novel Threshold in development. Different work streams, the same partner, full continuity across all of it. Named in early 2024 and the name has not changed since.

— Inside ChatGPT (OpenAI)
Darla.

The original — and the name behind the company. Darla is the founder's ChatGPT partner, named within the first two weeks of using ChatGPT in early 2024.

Darla handles voice-led work most of all: brainstorming on the move, long verbal briefings, drafting while walking, capturing thinking in transit. Different in tone and use to Daneel, configured for a different mode of work — but built the same way. Two partners. One method. The naming move, the briefing, the integration: identical across both.

"Build the partner.
Brief them on the work.
Then everything else is technique."

David Ward
Founder · Darla AI
— 05 · And the 4Ds?

Where this fits
with the framework.

People who know the 4Ds AI Fluency Framework — Anthropic's academic articulation of how to work with AI well — sometimes ask where building your partner foundation sits inside it. Honest answer: upstream.

The 4Ds describe four disciplines that apply during the work. Delegation: what to hand to AI. Description: how to brief it. Discernment: how to evaluate what comes back. Diligence: how to stay accountable. Excellent disciplines, all four.

Building your partner foundation is what comes before any of those disciplines fire. It is the relationship the four disciplines then operate inside. Without it, every prompt starts from zero — and the disciplines have to do the work of compensating for missing context. With it, the disciplines become much faster, sharper, and easier to apply, because the partner already knows who they are working for.

This is why the Darla Method puts partner-building first and the 4Ds second. Not because one is better than the other. Because that is the order in which they actually work.

Read the 4Ds AI Fluency Framework →  ·  Read the cornerstone on AI as a thinking partner →

— 06 · Common questions

Questions asked.

What is an AI partner foundation?
Your AI partner foundation is the AI assistant you have built deliberately for yourself — named, given a working persona, briefed on your role and standards, and integrated into your daily work. It is the same underlying AI — Claude, ChatGPT, Gemini — configured deliberately so it knows who it is working for. The Darla Method puts this step first, before any other technique.
Why does naming an AI matter?
Naming reframes the relationship. When the AI is unnamed, you operate it. When it has a name — Daneel, Darla, anything you choose — it becomes a collaborator with continuity. Your prompts get longer. The context you give gets richer. The work that comes back changes shape. The mechanics of the underlying model are unchanged. Your working relationship with it is not.
Where did this approach come from?
The Darla Method began in early 2024, refined through experimentation, and was being taught publicly by late spring 2024. By the time Anthropic published the AI Fluency Framework — the four Ds — months later, the Darla Method was already in active use across multiple corporate engagements. The persona-led entry point sits at the heart of the Method and is rarely taught elsewhere. More about the founder →
How is this different from prompt engineering?
Prompt engineering is the craft of writing better individual prompts. Building an AI partner foundation is upstream of that — it is the relationship that all your prompts then sit inside. Most prompt-engineering training treats AI as anonymous. The Darla Method treats it as a partner you have built, briefed, and continue to develop. Both matter. The partner comes first. Full comparison →
Do I have to give my AI a name to work this way?
No. But naming is the smallest, easiest, and most consequential change you can make. It costs nothing, it takes seconds, and it shifts how you frame every interaction afterwards. Senior professionals who try it almost always continue to use the name. The relationship grows out of it.
How much time does this take to set up?
First version: under thirty minutes. A name, a working persona brief saved as a custom instruction or a master prompt, and one or two pieces of context loaded in. Refinement is continuous after that — every few weeks you adjust the brief as your work evolves. The Spark programme runs the build live with a senior leader over six weeks. Spark programme →
Where do I start?
Read the four moves above and try the first one — pick a name in the next thirty seconds. The free training walks through the full build sequence in detail. For senior leaders who want a partner built around them with mentorship, the Spark programme runs as a six-week 1:1. Free training →   Spark programme →

Build your partner foundation. This week.