Claude for Professional Services: How Consulting Firms Are Using AI

If you run a consulting or advisory firm in ANZ, the May 2026 announcements from KPMG, PwC, and Deloitte have reset the floor of what client expectations look like. Claude is no longer a tool that one curious partner trials between engagements. It is the operating layer that the Big Four are rolling out to several hundred thousand staff at once. This guide gives you the practical answer on what consulting firms are actually doing with Claude, what Claude Cowork is, and where ANZ mid market firms should start.

What is Claude for Professional Services?

Claude for Professional Services is the deployment pattern Anthropic and its delivery partners use for consulting, advisory, audit, and legal services firms. It is not a single product sold off the shelf. It is a configuration of the Claude model family, the Claude Cowork collaboration product released in May 2026, and a set of agent templates targeted at the workflows that professional services firms actually run (Source: Anthropic product announcements, May 2026).

The distinction matters. Buying Claude API credits and giving partners a login does not deploy AI into a consulting firm. What deploys it is the activation work: connecting Claude to the firm's engagement knowledge base, building agents for proposal drafting and deliverable QA, configuring conflict checking, and producing the governance evidence that partners can sign off on. The model is the easy part. The activation is the work.

For ANZ mid market firms, the practical implication is that the Big Four announcements are a forcing function, not a competitive death sentence. Mid market firms do not need 276,000 licences. They need one production agent in 10 business days, then the next one, then the next one. Speed of deployment matters more than scale of licence count.

How are KPMG, PwC, and Deloitte using Claude?

KPMG announced on 19 May 2026 that 276,000 employees globally now have access to Claude across the firm (Source: KPMG global announcement, 19 May 2026). The deployment focuses on audit quality, tax research, advisory engagement intelligence, and the firm's internal knowledge platform. PwC and Deloitte have signed comparable enterprise alliances with Anthropic during the same May 2026 window, with similar workforce wide rollouts under way.

The headline workflows the Big Four are deploying fall into four buckets. Engagement intelligence: pulling the firm's prior engagement deliverables, methodology assets, and subject matter expert content into the agent context so consultants stop rebuilding work from scratch on every new engagement. Proposal drafting: generating first drafts of capability statements, methodology sections, and pricing rationale from the firm's win library. Deliverable quality assurance: running structured QA passes against draft client deliverables for consistency, methodology adherence, and risk language. Client research: assembling target client briefings ahead of pursuit meetings or relationship reviews.

The reason these workflows go first is straightforward. They are high volume, they are evidence rich (the firm already owns the knowledge), and the time saving per consultant is measurable inside a single billing cycle. Audit and tax research workflows follow once the base activation is in place and the firm has a clear governance pattern for regulated work.

What is Claude Cowork and when does it make sense?

Claude Cowork is Anthropic's collaboration product released in May 2026, targeted at how consulting and advisory teams actually work together on client engagements (Source: Anthropic product announcement, May 2026). It is not a chat interface bolted onto an existing model. It is a workspace where engagement teams co author with Claude, where artefacts are versioned alongside chat context, and where the model retains engagement specific context across sessions without exposing it to other engagements.

The practical use case is engagement workspaces. A team running a strategy engagement for a client creates a Cowork space for that engagement. The space holds the engagement charter, the interview notes, the analysis artefacts, the draft deliverables, and the Claude context. Team members work in the space across the duration of the engagement. When the engagement closes, the workspace is archived with appropriate retention. Conflict boundaries are enforced at the workspace level, not at the model level.

For ANZ mid market firms, Cowork makes sense when the firm runs concurrent multi week engagements with cross functional teams and needs a clean way to keep engagement context separate. For firms running mostly single consultant short engagements, the base Claude deployment with disciplined prompt patterns usually does the job for less governance overhead.

Where do ANZ mid market firms start?

The starting point that pays back fastest in ANZ mid market consulting is engagement intelligence. Most firms have a folder structure of past engagement deliverables, methodology assets, and credentials slides that nobody can find when they need them. Connecting Claude to that knowledge base, with appropriate access controls, gives every consultant a research analyst for the cost of one activation sprint.

The second workflow that consistently pays back is proposal drafting. ANZ mid market consulting firms typically write 50 to 200 proposals per year. A proposal drafting agent that pulls relevant past credentials, methodology language, and pricing benchmarks from the firm's win library produces first drafts in 30 minutes that previously took half a day. The proposal lead spends time on the client specific value proposition, not on assembly.

The third workflow is deliverable QA. A QA agent configured to check the firm's deliverable standards (structure, methodology citation, risk language, consistency of recommendations) catches the issues that partners currently catch on the final review. The result is fewer rework cycles and a more confident partner review session.

How does conflict checking and client confidentiality work?

Conflict boundaries are enforced through the deployment, not through the model. Claude does not natively know which engagements your firm is running or which clients are in conflict. A properly structured activation configures the engagement knowledge base with metadata that the agent respects: engagement code, client identity, conflict status, and team membership. The agent retrieves only the context the consultant is authorised to see for the engagement they are working on.

For client confidential content, the activation defaults to Anthropic's enterprise data handling baseline: client content submitted through the enterprise API is not used to train future Anthropic models (Source: Anthropic Enterprise Terms, May 2026). On top of that baseline, the deployment adds three layers. First, data residency is configured to Australian AWS infrastructure (Sydney region) so client content does not transit offshore in normal operation. Second, an audit trail captures every client document submitted to the model and every output generated. Third, access controls map to the firm's existing engagement team membership so consultants do not accidentally see content from engagements they are not staffed on.

How long until our first agent is delivering client work?

The standard activation timeline is 10 business days from kickoff to a production agent. Day 1 to 3 covers knowledge foundation: indexing the firm's engagement library, methodology assets, and credentials, plus configuring access controls. Days 4 to 7 cover the agent build and integration testing, including the first workflow (typically engagement intelligence or proposal drafting). Days 8 to 10 cover testing against historical engagements, partner review of outputs, governance documentation, and go live sign off.

Firms with complex knowledge management environments or bespoke proposal systems occasionally need an additional week for connector setup. This is identified at scope definition, not discovered mid sprint. The deliverable at the end of the sprint is a production agent doing real work, not a report about what the firm should build next.

What does this mean for our talent model?

The honest answer is that the entry level consultant role is being rewritten in real time. The work that historically built consulting competence (desk research, document synthesis, deck assembly, first draft proposal writing) is the work that AI does fastest and most reliably. Firms that respond by reducing graduate intake risk having no senior consultants in five years.

The firms that are getting this right are restructuring the graduate experience around supervised AI augmented work from week one. Graduates learn to brief Claude on engagement context, review and refine outputs, identify where the model is wrong, and apply judgement to recommendations. The competency being built is judgement and client intuition, faster than the previous model produced it. Hours per engagement drop. Engagement margins improve. Graduate progression accelerates rather than stalls.

Mid market firms have an advantage here. The deployment pattern is simpler, the change management is faster, and the graduate cohort is small enough to retrain in one cycle. The Big Four have to coordinate change across hundreds of thousands of people. A 40 partner firm can reset the operating model in a quarter.

What should partners do this quarter?

The forcing function in Q2 2026 is client expectation. Procurement teams in ANZ corporates and government are starting to ask consulting firms how they use AI in delivery. The firms that have a credible answer (a production agent, a governance framework, a partner who can talk through the deployment in technical detail) win the pitches. The firms that have a slide deck about AI strategy do not.

The practical action is to identify one engagement workflow that consumes the most senior time today, scope a 10 day activation against it, and have the production agent in front of clients in Q3. The second action is to brief the partnership on what the Big Four announcements mean and what your firm's position is going to be. Both conversations are clocked, not optional.

To scope a Claude activation for your firm, visit davidandgoliath.ai/claude-activation/professional-services or book a scoping call at davidandgoliath.ai/claude-activation/start.

Sources: KPMG global announcement, 19 May 2026. Anthropic product announcements, May 2026 (Claude Cowork). Anthropic Enterprise Terms, May 2026. PwC and Deloitte enterprise alliance announcements, May 2026.