[ TL;DR ]
CSM helps companies build AI into their operations. We applied that thinking to our own business.
The opportunity went beyond helping individuals complete tasks faster. We wanted to capture the knowledge behind our work, connect it to our business systems and use it to build workflows that agents could carry out within clear boundaries.
That meant bringing together our processes, standards, commercial context and team experience. We built agents around repeatable work and workspaces around the people responsible for directing, reviewing and improving it.
Our ambition is to build a company whose capabilities grow with its experience: useful knowledge is retained, successful methods become repeatable, and improvements can benefit the whole team.
[ The challenge ]
Valuable knowledge needs a way to become repeatable work
Preparing a proposal requires more than writing a document.
It requires understanding the customer, recognising what matters, defining the scope, applying commercial rules and knowing which assumptions need checking.
Research, content, tendering and delivery handovers have their own versions of that challenge. The finished output depends on a combination of information, experience and process.
At CSM, we wanted to make that combination available to the people and agents doing the work.
A folder of documents would provide reference material. A useful operating system needed to go further: connect the relevant information, explain how it should be used and organise the steps required to produce something the team could act on.
That became the foundation of the build.
[ What CSM built ]
A shared operating layer around the business
We brought together four connected parts.
Business information: the records and documents held across our existing tools.
Company knowledge: our objectives, processes, standards, commercial rules and examples of good work.
AI agents: workflows with defined responsibilities, access to relevant context and clear limits.
Team workspaces: shared places where people and agents contribute to a process, inspect outputs and manage decisions.
Omnia provides the company chat through which people can ask questions and access that context. Commercial Intelligence brings together the evidence leaders need to understand performance.
The system is designed around the responsibilities of each team. The underlying experience stays consistent, while the knowledge, agents and workspaces reflect the work being managed.
[ Building autonomous workflows ]
Turn an established method into a process an agent can carry out
Our starting point is the work itself.
What starts the process? What information does it require? Which steps are repeatable? What should the finished output contain? When should the workflow stop and involve a person?
Research provides a practical example.
An experienced team has a method for investigating a company or opportunity. It knows which questions matter, where to look, how to distinguish useful evidence from superficial information and what the next person needs from the findings.
We can capture that method in instructions, SOPs and examples, then use it to design a workflow:
01
Gather the relevant information from permitted sources.
02
Investigate the questions defined for the task.
03
Assess the findings against the company’s criteria.
04
Prepare an output with supporting evidence and unresolved questions.
05
Route exceptions or decisions to the person responsible.
The objective is for the agent to progress the repeatable steps without someone prompting every action.
People establish the standard, review performance and handle the situations that require judgement. Their corrections can then inform a reviewed improvement to the workflow.
This focus on process matters. McKinsey’s research found workflow redesign was the factor most strongly associated with reported earnings impact among the organisational attributes it tested. The evidence supports changing how work happens, alongside introducing AI tools. Source: McKinsey
[ Making company experience reusable ]
Build the team’s methods into the system
The knowledge behind a workflow determines how useful its output can be.
For CSM, that includes what makes an opportunity relevant, how we assess delivery requirements, how we communicate and what needs checking before work progresses.
The knowledge layer brings together company direction, operating guidance, market intelligence and role-specific context. Agents can draw on the parts relevant to their responsibilities.
This is also where the company can preserve useful experience. A recurring error can lead to a better check. A successful method can become a documented process. An approved example can help define the standard for future work.
Research offers evidence for the potential of this approach. A study of 5,179 customer-support workers found that an AI assistant increased issues resolved per hour by 14% on average, with larger gains among less experienced workers. The researchers found suggestive evidence that the system helped spread the practices of stronger performers. Those are findings from that study, rather than results claimed for CSM. Source: NBER, Generative AI at Work
For us, the principle is valuable: experience should help improve the company’s ability to deliver the next piece of work.
[ Connecting autonomous work to the team ]
Keep people, evidence and decisions in the same process
An agent completing a task is one part of a wider workflow.
A quote still needs commercial review. Content needs editorial judgement. A tender needs a bid decision. A delivery handover needs someone to accept responsibility for what follows.
Our workspaces bring those contributions together.
They show the work in progress, the people and agents involved, the information supporting the output and what needs to happen next.
Actions brings pending decisions into view. Activity provides a record of the work and reviews behind them.
This creates a clear place for human responsibility within a process that includes autonomous steps.
[ Commercial Intelligence ]
Use business context to decide what deserves attention
We also built the system around the questions leaders need to answer.
Which opportunities deserve investment? Where is work getting held up? What is changing in our commercial performance? What should we investigate before making another commitment?
Commercial Intelligence brings together the marketing, sales and financial picture. Company knowledge provides the context in which to interpret it.
For example, a campaign generating more opportunities at a lower cost may look like a reason to increase spending. The decision also depends on opportunity quality, delivery capacity and cash priorities.
Our aim is to make that assessment easier: bring the evidence together, show the relevant context and help the team identify the next question or action.
[ Why this can build enterprise value ]
Develop capabilities that remain with the company
Our view is that the long-term opportunity extends beyond the time saved on an individual task.
The company is investing in a reusable body of knowledge, connected processes and tested workflows. Research into AI and other general-purpose technologies identifies these complementary investments in processes, skills and business organisation as potentially valuable intangible assets. It also shows why the benefits can take time to emerge. Source: NBER, The Productivity J-Curve
For a business building this foundation, there are several routes to value.
More capacity from an established team
When agents reliably handle repeated preparation, the business may be able to process more work without increasing every supporting activity at the same rate.
The commercial test is the net improvement: output, quality and turnaround after accounting for review time, technology costs and maintenance. Workflow redesign is associated with stronger reported AI performance, but each implementation still needs to demonstrate its own economics. Source: McKinsey
Less dependence on knowledge held by particular individuals
Capturing methods, criteria and operating knowledge gives the company a way to make them available beyond the people who originally developed them.
We see that as a route towards better continuity and a more transferable business. It addresses a recognised acquisition concern: dependence on an owner or key people can put relationships and operations at risk when those people leave. Documentation and workflows contribute to addressing that risk; they do not remove it on their own. Source: BDC
Improvements that can be reused
A better research method, a stronger qualification check or a clearer handover process can become part of the next version of a workflow.
That is the asset we want to develop: a company’s accumulated experience expressed in processes that its people and systems can use. This is consistent with the research on the importance of organisational investment alongside the technology itself. Source: NBER, The Productivity J-Curve
An AI operating system does not automatically increase a company’s valuation. The enterprise-value case depends on whether it produces durable improvements in earning capacity, reliability and the transferability of the business.
Those are the outcomes we believe companies should build towards and measure.
[ Moving beyond individual AI use ]
The difference is how deeply AI is built into the business
ChatGPT and Copilot can be part of this foundation. ChatGPT supports connected company knowledge, and Microsoft Copilot Studio supports autonomous agents. The distinction is the depth of the implementation, rather than a claim that those platforms cannot support business-specific work. Source: OpenAI · Source: Microsoft Copilot Studio
For CSM, the work is in defining and connecting the operating system around the technology:
• Which company knowledge should inform each task?
• Which workflows can run autonomously, and within what boundaries?
• How will quality and performance be assessed?
• Where do people review, decide or intervene?
• How will the company maintain and improve the system?
Answering those questions turns access to AI into a practical business capability.
[ What this means for CSM ]
A foundation the company can keep developing
We built our operating system to bring company knowledge, autonomous workflows and human responsibility into a shared environment.
The next stage is to keep improving that foundation: strengthen the knowledge, refine the workflows and measure the effect on the work they support.
The opportunity we see is a business that becomes more capable as it learns. Its methods are captured, its repeated work is organised and its people can spend more attention on the decisions and improvements that move it forward.