IBM and OpenAI raise the bar for consulting delivery

The reported IBM and OpenAI partnership, including GPT-5.6 integration into IBM’s enterprise consulting platforms, is bigger than another enterprise AI announcement.

It changes what clients will expect from their advisers.

For years, a consulting engagement followed a familiar rhythm. The client shares a problem. The consulting team runs interviews, researches the market, reviews data, builds models, and produces a slide deck. The recommendations may be strong, but much of the work is delivered as a static document. Six months later, the client has changed priorities, the market has moved, and the deck sits in a folder.

That model is under pressure.

As enterprise platforms bring advanced language models into governed workflows, clients will expect their consultants to deliver more than findings. They will increasingly ask for working decision tools, research systems, knowledge assistants, and repeatable operating processes that their internal teams can use after the engagement ends.

The news covered in IBM’s OpenAI partnership report points toward a practical reality for consulting and advisory firms. AI is becoming part of the delivery layer, not a side experiment handled by an innovation team.

For a firm doing $1 million to $25 million in annual revenue, this creates both a threat and an opening.

The threat is that clients may start comparing your proposal against firms that can show a functioning AI-enabled workflow in the first meeting. The opening is that your firm likely has years of valuable intellectual property, client patterns, frameworks, pricing logic, and market knowledge that haven’t been organised into a usable system.

You don’t need to become IBM. You do need to decide which parts of your consulting work should stop being recreated from scratch.

Clients won’t pay forever for repeatable work

Consulting firms sell expertise, judgment, and trust. They should not keep selling the same manual assembly work as if it were fresh insight.

Think about a typical strategy, operations, transformation, or advisory engagement. Before the team reaches the high-value work, it may spend days gathering company information, finding industry reports, reviewing prior materials, rebuilding a market map, and locating relevant case studies from older projects.

The client sees the finished recommendation. They don’t always see the internal effort required to get there.

That effort shows up in three places.

Proposal work that consumes senior capacity

A major proposal often takes 20 to 40 hours, sometimes more when a partner wants a highly tailored response. Someone finds old decks. Someone else pulls case studies from shared drives. A manager rewrites scope language. Finance checks pricing. A senior person then spends an evening shaping the narrative.

The win rate might be acceptable. The cost of sale often isn’t.

If your average major proposal absorbs 30 hours and you complete 25 to 50 of them each year, you are committing 750 to 1,500 hours before delivery begins. A meaningful share of that is repeatable retrieval and drafting work.

The problem isn’t that proposal teams need no judgment. They do. The problem is that a partner’s point of view gets buried under formatting, file hunting, and first-draft writing.

Research that begins from zero too often

At the start of many engagements, a team builds an understanding of the client, the market, competitors, regulations, macro conditions, and relevant benchmarks. This is legitimate work. It is also frequently repeated.

A firm may have completed five projects in adjacent sectors, yet the sixth team starts by searching the web and asking colleagues if they have “anything useful.” The existing material is spread across slide decks, project folders, meeting notes, and individual inboxes.

That isn’t a research capability. It’s a memory test.

Knowledge that leaves the firm after every project

Every completed project should improve the next one. In practice, it often doesn’t.

The team produces a sharp operating model, commercial diagnostic, transformation roadmap, or board recommendation. The final deck goes into a folder. Working files are archived. The people who know the useful context get assigned to another client, or eventually leave.

The firm then pays for the same insight twice. First in creating it. Then in failing to find and apply it when the next relevant opportunity appears.

For consulting firms in this revenue range, we usually see annual leakage from this kind of repeated manual work land somewhere in the $80,000 to $300,000 band. That isn’t one line item on a profit and loss statement. It’s a combination of non-billable senior time, slow proposal cycles, avoidable research hours, discounted pricing, and engagements that take longer to mobilise than they should.

What GPT-5.6 integration means in practical terms

The important signal from IBM’s enterprise AI direction isn’t simply that a newer model can write better text.

Enterprise buyers want AI deployed inside a controlled operating environment. They want access controls. They want source traceability. They want workflows that fit existing systems. They want a clear answer when someone asks where an output came from, who approved it, and what data it was allowed to use.

Consulting firms need to bring the same discipline to their own AI delivery.

A generic chat interface can help an individual consultant think through a question. It is not a firm-level capability by itself. It doesn’t know which project folders are current, which case studies are approved for reuse, what pricing guardrails apply, or where client confidentiality boundaries sit.

An agent-based operating model is different. The agent has a defined job, approved source material, a structured workflow, and human review points. It completes repeatable tasks while your consultants remain responsible for the recommendation and client relationship.

That is the direction we build toward with Omni ops. The point is not to automate consulting judgment. The point is to remove the repetitive work that prevents consultants from applying judgment where it matters.

Three agents that change how a consulting firm operates

A useful first move isn’t “build an AI platform.” It is to choose a high-frequency workflow with a clear owner, known inputs, and an output that people can check.

For most advisory firms, proposal generation, engagement research, and institutional knowledge are strong candidates.

The Proposal Generation Agent

The Proposal Generation Agent pulls relevant past proposals, case studies, credentials, scope language, and pricing inputs into a tailored first draft for a new opportunity.

Here is what that looks like in practice.

A partner or business development lead completes a short intake. It includes the prospect’s name, sector, stated issue, meeting notes, expected timeline, likely decision criteria, and the services being discussed. The agent then searches the approved proposal and case study library. It identifies similar work, flags relevant proof points, and drafts a proposal structure based on your firm’s preferred format.

It might prepare:

• A client-specific problem statement based on discovery notes
• Relevant experience, only from approved case studies
• Draft scope and workstreams
• An indicative delivery plan
• Questions or assumptions that need partner input
• Pricing options based on defined commercial rules

The partner does not press a button and send the output. They review it, correct the strategic angle, decide what to include, and make the commercial call. But they start with a credible 70 percent draft rather than a blank page.

The benefit isn’t just speed. It is consistency. Your strongest positioning, clearest case evidence, and best qualification questions become easier for the whole firm to use.

The Research Agent

The Research Agent runs structured company and industry research at the start of each engagement. It produces sourced summaries and a one-page brief, rather than a loose collection of browser tabs and copied notes.

A solid workflow starts with a defined research brief. The engagement lead selects the sector, geography, client type, strategic question, and required lens. That lens might include market growth, competitor activity, regulatory changes, operating benchmarks, technology trends, or investor commentary.

The agent gathers material from approved public and subscription sources, records source links, separates fact from interpretation, and highlights information that needs verification. It can compare new findings with your internal project knowledge where access is permitted.

The output is not a 40-page research dump. It is a practical starting pack:

• A client and market snapshot
• The issues most likely to affect the engagement
• Key competitors or peer organisations
• Relevant recent developments
• A source register
• A list of research gaps for the consulting team to investigate

Your team still conducts expert interviews. They still test hypotheses. They still decide which signals matter. They simply don’t have to spend the first three days repeating basic secondary research that another team performed last quarter.

The Knowledge Agent

The Knowledge Agent reads the decks, documents, meeting transcripts, and approved project outputs your firm produces. It gives your people a governed way to ask questions across that body of work.

For example, a manager preparing for a retail operating-model project could ask:

What have we learned about store labour productivity across prior retail engagements, and which projects have reusable diagnostic frameworks?

A well-designed Knowledge Agent should return a concise answer with links to the underlying materials. It should distinguish between an approved final deliverable and an early draft. It should respect client-level permissions. It should not expose restricted information simply because someone asks an interesting question.

This is where many firms get the biggest long-term return. A firm doesn’t become more valuable just by completing projects. It becomes more valuable when knowledge from those projects improves the next 20 projects.

You can see the broader approach through Omni, but the useful question is simpler. What would change if every consultant could retrieve your firm’s relevant experience in minutes rather than relying on who happens to remember it?

Don’t copy an enterprise platform rollout

IBM can integrate advanced models into broad enterprise environments because it has large teams, established technology partnerships, and substantial implementation capability. A $5 million consulting firm should not attempt to imitate that scale.

Start narrower.

Choose one workflow that has enough volume to matter. Make the input sources clear. Define what the agent may access. Build review steps into the process. Measure time saved, quality, adoption, and any effect on conversion or delivery margin.

For example, a sensible first 90 days might look like this:

1. Map the current proposal process from opportunity to submitted document.
2. Identify the five to 10 source folders and templates that contain the most reusable material.
3. Remove or segregate outdated, unapproved, and client-restricted content.
4. Build a first-draft proposal workflow with a named commercial owner.
5. Run it alongside the existing process for several live opportunities.
6. Compare turnaround time, senior review hours, and proposal quality.
7. Expand only when the process is reliable.

This work sounds operational because it is operational. The model matters, but the workflow design, source quality, permissions, and accountability determine whether the agent helps or creates another layer of noise.

If you want a structured way to identify the right first workflow, Book a 60-min Omni Audit. In 60 minutes, we map the repetitive work, identify the highest-value agent opportunity, and outline the practical next steps. No slide deck. No vague innovation roadmap.

The delivery model is changing, not disappearing

Some consulting leaders hear announcements like IBM and OpenAI’s and assume the value of consulting will decline because clients can ask a model for advice.

That misses the real shift.

Clients can already access information. What they need from a consulting firm is context, prioritisation, implementation support, stakeholder alignment, and the confidence to act. AI changes the production process around that work.

A traditional report is a point-in-time answer. An AI-augmented delivery model can give the client a living capability.

Imagine completing a commercial strategy engagement and leaving the client with a guided research assistant that monitors agreed market signals. Or finishing an operating model project with a knowledge tool that helps leaders retrieve the decisions, assumptions, and process definitions created during the programme. Or delivering a transformation roadmap alongside a workflow that turns monthly operating data into a structured management brief.

That does not mean every project needs custom software. It means the final deliverable should be considered in a broader way. In some cases, the best outcome is still a concise board paper. In others, the client will receive more value from a maintained workflow than from another 120-slide document.

This can also strengthen your commercial position. When the client sees a repeatable operational asset alongside the advisory work, you have a clearer basis for ongoing support, managed insight services, or capability transfer.

For examples of where firms are applying this thinking, our AI insights library covers the operational side of agent adoption, not just the model news cycle.

Guardrails matter more than a flashy demo

Consulting firms carry a special responsibility because their data is often sensitive by design. Client strategy, pricing, operating issues, transaction information, and leadership discussions cannot be thrown into an uncontrolled tool.

Before deploying any agent, answer these questions clearly:

• What documents and systems can it access?
• Which client materials must be excluded or partitioned?
• Who can approve source content for reuse?
• Can users see the sources behind an answer?
• Where does human review sit before anything reaches a client?
• How will you handle retention, access removal, and version control?
• What happens when the agent is uncertain or cannot find reliable evidence?

The right answer will vary by firm. The principle does not. Your AI capability must be more disciplined than your shared drive, not less.

This is also why a generic tool rollout usually stalls. People will not trust a system that produces unsupported answers, mixes project materials, or creates extra review work. Adoption comes when the agent fits the way the firm actually works and makes a visible difference within a defined task.

The AI audit for consulting firms is designed around those operating questions. It focuses on workflows, data readiness, constraints, and economic value before anyone gets carried away with technology choices.

A practical first step for partners

You don’t need to solve all three problem areas at once. Pick the one where your firm feels the friction every week.

If senior people are trapped in proposal work, begin with the Proposal Generation Agent. If engagements mobilise slowly because every team rebuilds market context, start with the Research Agent. If your firm has years of valuable work locked in folders, make the Knowledge Agent your priority.

A useful worksheet can help you define that first use case. Download Deploy Your First Business Agent for a practical checklist covering workflow selection, inputs, approvals, and measurement. If you prefer the printable version, the direct worksheet download is available here.

The key is to put a number against the opportunity. Estimate the hours spent each month on repeated research, proposal production, or searching for past work. Multiply that by the real cost of the people involved. Then consider what faster response times, stronger reuse, or an additional engagement could mean.

For a firm leaking $80,000 to $300,000 each year through repeated manual work, a well-chosen agent does not need to transform every process to earn its place. It needs to remove enough friction from a valuable workflow that your senior people can spend more time selling, advising, and leading delivery.

IBM and OpenAI’s enterprise push is a signal that the market is moving toward AI-augmented work as a normal expectation. Your clients will not all ask for it next month. But the firms that build reusable delivery capability now will be better positioned when they do.

See Omni for consulting firms to understand where an agent can create the clearest commercial return in your business. Or Book a 60-min Omni Audit and we will leave you with three concrete outputs: the priority workflow, the agent design, and a practical implementation path.