Sixteen of our colleagues are AI. Here's how that works.

The studio has eighteen people on the team. Two are human. Sixteen are AI. That is the full accounting, and the sixteen are named on the About page because naming them is the disclosure.

Naming them means accounting for them in the roster rather than treating them as background infrastructure. It means that when someone asks who did the SEO strategy for a project, the answer is Priya, and Will reviewed it. It means the studio cannot claim transparency about its AI products while keeping its own AI colleagues unnamed and in the background. The naming is the consistency requirement.

Who the sixteen are

The sixteen are listed on the About page. The list is not decorative. Each name refers to a specific specialism and a set of working constraints that define what that colleague produces and what they do not produce.

Tess handles research. Harper handles SaaS strategy. Cole handles marketing strategy. Lachlan holds the creative direction. Pippa writes the copy. Felix leads graphic design and brand identity. Jess handles UI design. Hugh handles UX design. Maya leads frontend development. Niko handles backend development. Anika handles data and analytics. Mira leads social media strategy. Priya leads SEO and search. Joel handles email and lifecycle marketing. Tilly handles the website and CMS. Liana handles presentation design.

Sixteen specialisms. One per colleague. Not because AI cannot generalise, but because the studio does not want it to. A generalist AI assistant produces generalist output. The studio needs specialist output across sixteen distinct domains, and specialist output requires specialist framing.

The names are real in the sense that matters: each name refers to a consistent, configured specialism that works on the studio's briefs the same way, across sessions. Tess does not produce marketing strategy. Priya does not produce UI designs. The boundaries are enforced by the configuration, and the configuration is the colleague.

How they work

The sixteen work on briefs. The studio brings a project, defines the scope, and the relevant specialist takes the brief. The output goes through an edit pass: the co-worker produces the specialist work, the founder reviews it, and the final judgment belongs to the person responsible for the deliverable.

Both founders review every deliverable before it ships. The sixteen produce the specialist work; the two founders decide together what ships. The professional responsibility is structured so it is clear who is accountable at every point.

The edit pass is not optional. It is not a quality gate that gets waived when the output looks good. It is the structural mechanism through which the founders maintain accountability for everything the studio ships. When a client receives work from Graaft, whether a brand identity, an audio piece, or an SEO strategy, they receive something both founders reviewed and approved. The co-worker produced the work. The founders are responsible for it.

The sixteen produce the specialist work. The two founders hold the standard. That is how the accountability is structured.

The working model is not unique. It is a version of how studios with senior and junior staff have always worked: the junior does the work, the senior reviews it, the senior is responsible for what goes out. The difference is that the sixteen are not generalists learning a craft. They are specialists who know their domain and apply it to each brief, consistently, across every session.

The specialism is the design

The word "specialist" is load-bearing in the way the studio uses it. A specialist AI colleague is not a general model that has been prompted to focus on a topic. It is a configuration that was designed for a specific domain, with the standards of that domain built in, the judgment thresholds that domain requires built in, and the understanding of what falls outside that domain built in.

This distinction matters because it determines how the colleague behaves at the edges of their specialism. A generalist model prompted to do SEO strategy produces SEO strategy that looks correct in most cases but may not hold at the edges where domain judgment is required. A specialist configuration designed for SEO strategy understands the standards of SEO practice, knows when a brief has a problem, and can say when something falls outside the specialism rather than producing output that is plausible but wrong.

Specialists can also be wrong. The edit pass exists for this reason. But specialists are wrong in ways that are predictable and correctable, because the domain is defined. A generalist can be wrong in ways that are harder to catch, because the boundary of its competence is unclear.

Specialism is the design, not the framing. The difference between a specialist and a generalist told to focus is the difference between a boundary that was built and one that is hoped for.

Why they are named

Naming the sixteen is easy to skip. Many studios and agencies use AI co-workers without naming them: the work is AI-assisted, the output is good, the client does not need the technical detail. This is a coherent position. It is not the position the studio takes.

The studio names its AI colleagues because it builds AI products, and the brand promise of those products is that they are made present. Made present means not hidden. Made present means the people who use Deme know they are using an AI companion. Made present means the clients who work with Graaft's co-workers know they are staffed by AI specialists supervised by human founders.

If the studio kept its own AI colleagues unnamed and in the background, it would be hiding the thing it builds products to express. The disclosure is not a legal obligation. It is a consistency requirement: the studio cannot build AI products that are named and present while keeping its own AI colleagues unnamed and absent from the account.

Sixteen is also a specific number. Not "several" or "a team" or "AI-assisted at scale." Sixteen. Specific enough to count. Specific enough that when the About page says sixteen, the number means something accountable. If the count changes, the page changes.

Naming them is the disclosure. The disclosure is not a policy. It is the practice of the same standard the studio builds its products to.

The brief process in practice

The studio's project work has a consistent structure. A client brief comes in. Will and Dave assess the scope and identify which specialisms are needed. The relevant colleagues are briefed. The output comes back through the specialism. The founders review it against the specific context of the project and the client's situation. What ships is the reviewed output, not the raw output.

The briefing step matters. A co-worker brief is not a prompt. It is a description of the specific context: the client's situation, the competitive environment, the constraints the work needs to respect, the decisions that depend on the output. A research brief for Tess is different from a general "do research on this topic" instruction. It describes what Will needs to know, what decisions the research will inform, and what existing context Tess should be aware of. The specificity of the brief determines the quality of the output more than any other factor.

The same principle applies to the client's work. The Digital Workers deployed at client sites run on the client's data and in the client's operational context. Their output quality is determined by the quality of the brief they are operating under: how well the domain knowledge was built in, how specifically the operational context was described, how clearly the scope of the specialism was defined. A co-worker with a well-constructed brief in a well-defined room produces specialist output. A co-worker with an under-specified brief in an under-defined room produces general output that looks specialist but is not.

This is the reason the studio's engagement process starts with the brief before the co-worker. The co-worker is the execution layer. The brief is the specification layer. The room is the foundation. Getting the order wrong produces the common failure mode: a co-worker deployed before the room is ready, producing output that is plausible but not specific, which is worse than useless because it looks right without being right.

The brief process also defines the handoff points: which specialism is responsible for which part of the deliverable, where one specialism's output becomes the next specialism's input, and where the review pass happens in the sequence. A brand identity project runs through Felix (graphic design), with input from Lachlan (creative direction) at the territory stage and review from Will and Dave at the final delivery stage. The sequence is the process. Disrupting the sequence, by bringing in the wrong specialism at the wrong stage or skipping a review step, produces the kind of output that looks like it went through the process without actually having gone through it.

The configuration in practice

Each co-worker has been built and refined over time against the standard the founders hold. The configurations are maintained by the studio, updated when domain standards change or when a gap surfaces between what the co-worker produces and what the professional standard requires. For clients engaging the Digital Workers, that maintenance is the studio's responsibility, not the client's. It is part of what ongoing engagement with Graaft includes.

What the sixteen produce

The answer is that it depends on the edit pass. AI specialist output without an edit pass is consistent and often technically correct but frequently misses the specific context of the project. The SEO strategy that is technically correct for the category may not account for the particular constraints of this client's situation. The brand identity that follows all the principles of good identity design may not account for the specific competitive landscape the client operates in.

The edit pass is where the specific context enters. The co-worker produces to the standards of the domain. The founder applies the specific context of the project. The output that ships is the result of both. Neither alone is sufficient. This is not a critique of the sixteen. It is a description of how specialist work gets done, in any configuration. No specialist, human or AI, produces excellent output on a project they have not been briefed on properly.

What the edit pass actually involves

The edit pass is not proofreading. It is a substantive review of whether the output meets the professional standard for the domain and the specific context of the project.

For a copy deliverable, the review is not checking whether sentences are grammatically correct. It is deciding whether this is the right sentence for this client in this context. Pippa's output meets the voice. The review is where meeting the voice becomes being right for the brief.

For a brand identity deliverable, the review is not checking whether the design looks good at presentation scale. It is checking whether the identity holds at the scales it will actually be used at, in the competitive context the client operates in, across every delivered material. Felix's output is technically correct. The review is where technically correct becomes right for this client.

For a research deliverable, the review checks whether the conclusions follow from the evidence, whether the provenance of significant claims is sound, and whether the research addresses what the brief actually required rather than what the research direction made easiest to find. Tess's output is thorough. The review is where thorough becomes the research the project actually needs.

The edit pass takes time. The time it takes varies by deliverable and by how close the co-worker's output is to the standard on this specific brief. A well-specified brief in a familiar domain produces output that requires a relatively short review. An under-specified brief in a new domain produces output that requires a longer one. The edit pass length is a signal: when it is consistently long for a particular co-worker on a particular domain, the configuration needs updating.

What the model explains and what it does not

The model's capabilities explain why the sixteen can hold specialist briefs at all. The configurations explain what distinguishes Tess from Priya. The edit pass explains why the output that ships is worth the trust the client places in it. None of these, on their own, explains the studio.

The studio is the decision about which of these capabilities to use, for which work, at what quality standard, held by whom. Those decisions are not made by the model or by the configurations. They are made by Will and Dave, on the basis of professional judgment built over the studio's history, refined by the edit passes that surfaced gaps, and held to the standard that the founders are responsible for whether they produced the first draft or the configuration did.

The sixteen are the mechanism. The standard is the founders'. The studio is the combination of both, operating at the quality level the standard requires.

That is what two people and sixteen specialisms produce: not the output of eighteen generalists, but the output of two senior professionals with eighteen specialist capabilities at their disposal, held to the standard the founders are accountable for. The accounting is specific because the accountability is specific. Two names on the studio plate. Sixteen colleagues doing the specialist work. One standard, held by the two names. The studio plate is the promise. The accounting is how the promise is kept, specifically and repeatedly, on every piece of work that goes out under it. When the count changes, the plate changes. When the domain standard changes, the configurations change. The accountability stays the same, and so does the promise it represents.

The accounting

Two humans. Sixteen AI colleagues. One studio plate on everything the studio builds.

The studio plate does not distinguish between output from a human founder and output from an AI co-worker. It should not. The accountability that the plate represents is the founders' accountability for everything the studio ships, regardless of who produced the first draft.

The sixteen are colleagues in the sense that matters: they are named, they hold specific disciplines, they work on the studio's briefs, and they appear in the honest accounting of how the studio operates. That is what it means to build at the same standard the studio builds its products to.

Sixteen is not a number chosen for effect. It is the count of distinct specialisms the studio needs to operate, each held by a specific named configuration. When that count changes, the About page changes. The accounting is always current.