Most consulting firms don’t have a utilisation problem first. They have an administration problem hiding inside their utilisation numbers.

A project manager spends Monday morning assembling a status update from Slack messages, a half-finished project plan, and a few notes from Friday’s client call. A partner rewrites that update before it goes to the steering committee. A consultant spends 45 minutes trying to find the latest version of a workstream deck. Someone else takes meeting notes, then spends another hour turning them into actions, decisions, and emails.

None of this is billable advisory work. Yet it often takes 30% to 40% of a consultant’s week.

For a consulting or advisory firm doing $1 million to $25 million in annual revenue, this is not a minor operational annoyance. It creates delivery drag, reduces capacity for client work, and pushes senior people into coordination work they should not own. Across firms in this range, we usually see annual leakage from fragmented project administration land somewhere between $80K and $300K.

The answer isn’t asking people to write better notes or create another SharePoint folder structure. You need a defined operating process, supported by an AI agent that can collect project signals, organise them, draft the right outputs, and route work to a human for approval.

This is the kind of work we assess through the AI audit for consulting firms. The goal is practical. Find the work that keeps recurring, map its inputs and handoffs, then build an agent that removes the manual assembly.

Where project admin consumes consulting time

Project administration rarely appears as one clean line item on a timesheet. It shows up in fragments throughout the day.

A consultant updates a project tracker after a client call. A manager asks the team to send their latest workstream status. Someone downloads an attachment from email because the file in the shared drive is not current. The engagement lead needs a view of risks before a client meeting, so three people scramble to create one.

Each step seems reasonable on its own. Together, they produce a lot of non-client-facing work.

The four areas below are where most firms lose time.

Status updates built from scratch

Many weekly status reports are assembled manually, even when the project has a clear plan, regular meeting cadence, and defined deliverables.

The usual process looks like this:

1. A project lead asks each workstream owner for updates.
2. Workstream owners send different formats through email, Teams, Slack, or a spreadsheet.
3. A project manager combines the inputs into a template.
4. Risks and dependencies are rewritten for the client audience.
5. A partner or director reviews and edits the final version.
6. The report is circulated, often after the meeting it was meant to support.

This is expensive because it repeats every week. It also creates a quality problem. If project status depends on who had time to write a useful update, leadership is seeing an incomplete picture.

An AI agent can pull from the project plan, task system, meeting notes, and prior reports. It can draft a status report in the firm’s agreed format, highlight late milestones, identify open decisions, and prepare a list of questions that need human input.

The project lead still owns the judgement. They should approve what goes to the client. But they shouldn’t have to assemble the first draft from six different sources.

File organisation and version confusion

Consulting projects produce a lot of material quickly. There are interview notes, analysis files, client data extracts, workshop outputs, issue logs, presentations, and draft recommendations.

Without an operating standard, the shared drive becomes a holding area rather than a working system. Teams create folders by habit. Files get saved locally. Decks appear with names like Final_v7_UseThisOne.pptx. A client asks for a document from three weeks ago, and a consultant has to search through email threads to find it.

The cost isn’t only the five or 10 minutes spent locating a file. Version confusion causes rework. People analyse outdated data. Teams prepare for meetings using an old deck. A director reviews the wrong document and asks for changes that have already been made elsewhere.

A project administration agent can monitor defined folders and channels, apply naming rules, tag documents by client, engagement, workstream, and document type, then identify likely duplicates or outdated versions. It can create a project index that tells the team where the approved current files sit.

That doesn’t mean giving an agent free access to move every file without controls. For most firms, the better design is for the agent to recommend classifications, flag issues, and perform only approved actions in designated workspaces.

Meeting notes that never become decisions

Client meetings are a major source of project information. They produce feedback, decisions, action items, risks, stakeholder concerns, and changes to scope.

Yet notes are often trapped in one person’s notebook, a Teams transcript, or an unstructured document that nobody reads again.

When the notes aren’t converted into a reliable project record, the same problems appear repeatedly:

• Actions are unclear or have no owner.
• Decisions get revisited because there is no agreed record.
• Client commitments are missed.
• New team members lack context.
• Partners spend time asking for a recap before every internal review.

A good agent workflow starts immediately after the meeting. It takes the transcript, meeting notes, or recording summary and produces a draft decision log, action register, open question list, and client follow-up email. It compares new commitments against the project plan and flags potential scope changes.

A human reviews the draft within a defined window, usually the same day. Once approved, the agent files the output in the project workspace and updates the relevant tracker.

That process turns a meeting from a one-off conversation into structured delivery data.

Client communication that gets rewritten too many times

Client communication is important. It should not sound automated, generic, or careless.

That doesn’t mean senior staff need to compose every routine update from a blank page.

Many client emails contain predictable elements. They confirm what happened, state what is needed next, restate deadlines, identify risks, and clarify ownership. The substance varies by project, but the communication structure is consistent.

An agent can prepare a draft based on the latest meeting, project status, and agreed communication templates. It can differentiate between a routine workstream update and a message that needs partner judgement because it involves scope, commercial risk, or a sensitive stakeholder issue.

The useful line is not “automate every email.” The useful line is “make the first draft and the project context available before a senior person starts writing.”

The operating model for a project admin agent

AI works best when the work is defined before the tool is selected.

If your team has no agreed project templates, no reliable place for actions, and no clear owner for client communications, an agent will surface the inconsistency. It won’t fix it on its own.

For consulting project admin, we typically design an agent around five stages.

1. Capture project signals

The agent needs access to the systems where project activity happens. This might include:

• Project plans and task boards
• Teams, Slack, or email channels
• SharePoint, Google Drive, or another document system
• Meeting transcripts and notes
• CRM records for client and engagement context
• Existing weekly report templates

Not every source needs to be connected on day one. Start with the few systems that hold the most reliable information.

For example, a transformation consultancy might connect its project plan, Teams meeting transcripts, and SharePoint document library first. That gives the agent enough context to draft updates, produce action logs, and improve file retrieval without creating a large integration project.

2. Apply a project-specific instruction set

The agent needs rules. Not broad statements like “make a useful summary.”

It needs instructions such as:

• Use the approved weekly status report format.
• Flag milestones due within 14 days with no confirmed owner.
• Treat unapproved scope changes as a risk, not a committed deliverable.
• Assign action owners only where the meeting record names one.
• Draft client emails in the engagement lead’s preferred style.
• Never send external communication without human approval.
• Save approved meeting outputs in the relevant client and workstream folder.

These instructions make the agent dependable. They also protect the client’s experience and your firm’s standards.

This is where Omni ops becomes more useful than a standalone chat tool. The aim is to put an agent into a repeatable business process, with the right information, rules, permissions, and review points.

3. Draft the work product

Once the inputs and rules are in place, the agent can generate work that normally takes project staff hours each week.

A typical weekly cycle might look like this:

• On Thursday afternoon, the agent reviews project activity and prepares workstream update prompts.
• On Friday, workstream leads confirm or amend the key points.
• On Monday morning, the agent drafts the internal delivery review pack and client status report.
• After the internal review, it prepares the client version and a follow-up communication.
• After each major meeting, it creates a decision record, actions, and a draft email for approval.

This is not hands-off project management. It is a faster preparation cycle, with fewer people chasing information.

4. Route exceptions to the right person

The agent should not treat every project issue equally.

A late internal task may only need a reminder. A client request that changes the scope, timeline, or fee needs escalation to the engagement lead. A risk involving a regulator, legal exposure, or a senior client relationship needs a different path again.

This is why agent design needs escalation rules. Your team should know what the agent can draft, what it can update, and what it must flag for a human decision.

The objective is to reduce administrative load without removing professional accountability.

5. Build reusable knowledge from the project

Every engagement should leave the firm smarter. In reality, much of the insight stays buried in slide decks, workshop notes, and individual inboxes.

That creates knowledge management debt. The firm pays for the same research, framing, and delivery lessons more than once.

A Knowledge Agent can read approved decks, documents, and meeting transcripts across your corpus. It can answer questions such as:

• How have we structured an operating model assessment for a similar client?
• What risks appeared in previous ERP implementation projects?
• Which case examples support this proposal theme?
• What did we recommend for the last client in this sector?

It should work from governed source material, with access rules that reflect client confidentiality. But done properly, it turns completed projects into accessible firm IP rather than archived files.

The role of research and proposal agents

Project administration is the immediate use case, but it connects to two other areas that usually consume expensive consulting time.

The first is research. New engagements often start with a familiar scramble. Consultants gather public reports, company information, competitor views, market data, and prior internal materials. Some research is essential. Some of it is repeated because the firm cannot quickly find what it already knows.

A Research Agent can run a structured starting process for each engagement. It gathers approved sources, creates summaries, records citations, and produces a one-page brief for the delivery team. That gives consultants a stronger starting point and reduces duplicated secondary research.

The second is proposal development. In many firms, partners and senior managers still spend 20 to 40 hours on a major proposal. The win rate might be acceptable, but the cost of sale becomes brutal when each response starts from a blank deck.

A Proposal Generation Agent can pull prior proposals, relevant case studies, team credentials, and approved pricing structures into a tailored first draft. Senior people still make the strategic calls. They set the offer, challenge the assumptions, and shape the story for the buyer. The agent removes the retrieval and first-assembly work.

These three agents work well together. Project admin creates structured delivery records. The Knowledge Agent makes those records reusable. The Research Agent and Proposal Generation Agent help turn the firm’s existing knowledge into faster commercial and delivery work.

You can find practical examples of how organisations are applying this thinking in our AI insights library, but the right starting point is always your own workflow, data, and team habits.

Put a dollar value on the admin problem

Consulting owners often see project admin as part of the cost of delivery. That is true. It is also a capacity decision.

Consider a 20-person firm where 12 people are regularly involved in delivery. If each person spends just five to eight hours a week on status collation, file searching, notes distribution, and routine client updates, that is 60 to 96 hours each week.

You won’t recover all of that time. Some coordination is necessary and good. But even recovering a portion can matter.

If the firm reduces this administrative effort by two to three hours per delivery person each week, it may create 24 to 36 hours of capacity. That can mean more client-facing preparation, less late-night catch-up, stronger project control, or room to serve another engagement without immediately adding headcount.

The financial impact depends on your billing rates, utilisation model, and how much recovered time becomes useful capacity. For firms in the $1 million to $25 million range, it is common for this category of leakage to reach the $80K to $300K annual band once you include senior review time, avoidable rework, and missed billable capacity.

Before buying software, calculate your own number:

1. List the recurring project admin tasks by role.
2. Estimate weekly time spent on each task.
3. Use loaded cost, not just salary.
4. Separate necessary project governance from repeated information chasing.
5. Identify where better preparation would release senior time.

That exercise often reveals that the largest opportunity is not the task that feels most irritating. It is the task repeated by the most expensive people.

If you want an outside view of that map, Book a 60-min Omni Audit. In 60 minutes, we identify the workflow, the data and system inputs, and the agent opportunity. You leave with three practical outputs, a prioritised use case, an initial operating design, and a view of the value at stake. No deck, no vague innovation discussion.

Start with one project workflow, not a firm-wide rollout

The best first deployment is usually a contained workflow with visible repetition and clear inputs.

For many consulting firms, that is the weekly project status process and post-meeting follow-up. It is frequent, easy to measure, and directly felt by consultants, project managers, and clients.

Choose one project type or one delivery team. Define the templates. Agree on the source systems. Set the review rules. Run it for four to six weeks. Measure how long status preparation takes before and after, how quickly actions are distributed, and how often the team needs to search for project material.

Then improve the workflow before expanding it.

If you need a practical way to define that first agent, use the Deploy Your First Business Agent worksheet. It helps you document the trigger, inputs, instructions, approval steps, and measures that make an agent useful in a real business process. You can also download the worksheet directly and work through it with your delivery lead.

The firms that get traction do not try to automate consulting judgement. They remove the repeated preparation work around that judgement.

Turn project admin into a controlled advantage

Your consultants were hired to solve client problems, run workshops, analyse difficult issues, and advise decision-makers. They were not hired to spend a third of their week reconstructing project status from scattered tools.

A well-designed project administration agent can produce cleaner status updates, faster meeting follow-up, more reliable project records, and less routine communication work. It also creates the foundation for reusable knowledge, faster research, and better proposal generation.

The important part is the operating design. You need the right workflow, systems, permissions, templates, human review points, and ownership model.

See Omni for consulting firms to understand how we approach that work. When you are ready to identify the first workflow worth fixing, Book my Omni Audit.