One founder we work with had built his own CEO assistant and tools for his managers. His company still ran through the same layers, meetings and manual work. Personal productivity had moved; the organization had barely changed.
Codos works on that gap with the CEO, employees and engineers together: set the business goal, capture the company context and rebuild the workflows around it.
01Set the ambition
Choose a goal the current company cannot reach.
Double revenue per employee. Serve more locations without support growing in lockstep. These are example goals that require a different operating model.
Work backward from the goal: which tasks grow with every new account, and where does a customer need another person to move them forward? Choose a workflow with enough economic weight to matter and the access needed to change it.
02What Codos puts in place
A working layer across the company.
In one back-office diagnostic, dozens of AI-led interviews mapped hundreds of operations in about two weeks. Several queues crossed department boundaries, led by document drafting and report assembly.
The map ranked work by reported hours and automation readiness. Department heads prioritized contract drafting, refunds and payroll checks. Each candidate carried the reasoning and an employee’s description of the task.
- AI Interviewer → a map of real work
- Structured voice interviews capture the task, tools, time, variations and knowledge held in people’s heads. Reviewers get a diagnostic report and a ranked set of workflows to investigate and change.
- Company Brain → context agents can use
- A separate customer deployment connects the workplace sources you authorize. People, customers, projects, decisions and their relationships become available to employees and AI assistants, with access defined by your setup.
- Embedded engineers → workflows in production
- We connect the source systems, build the application or agent, define its checks and human review points, and test real cases with your teams. The deliverable is the working flow and the operating routine around it.
03A workflow rebuilt
The application becomes the work.
At a client, we moved document collection earlier and built self-service onboarding that populated the CRM from customer documents. Applications produced prefilled agreement drafts for human review; alerts linked employees to stalled customer records.
A later Codos case study reported roughly half the time from lead to customer ready to sell. A daily CRM report replaced funnel numbers collected by hand from chat. The useful unit of change was the full customer journey.
The application requests business documents earlier in the journey, before the team has to chase them after payment.
Document information populates the customer record. The agreement draft uses the relevant entity and commercial details.
A named-customer alert links to the record requiring attention. The employee helps resolve the case.
Review, signing and activation remain visible in the same customer journey. The process owner follows it through to service.
Illustrative walkthrough based on delivered onboarding work. Select each step to see the system, the team’s role and the measure.
Early reviews found staff still filling forms for customers and signing still causing delays. Form completion alone had not shortened activation; the complete journey needed attention.
04Change the speed of execution
Agents can build the systems, too.
In a client engagement, agents turned tickets into plans, code and reviewed changes. Accepted architecture and product documents constrained the work; automated reviews could block a change before it reached a person. People retained the merge gate.
Engineers maintained those constraints and reviewed the recorded agent sessions. In a separate QA pilot within the engagement, agents wrote and ran automated tests around work dominated by manual regression. Test output was evaluated alongside release quality and defects.
Build the workflow. Then improve the machinery that builds the next one.
05Make it how the company works
The leadership team has to experience the new company.
One client brought its leadership team into a hands-on workshop. Founders and executives rebuilt their own processes with engineers beside them. Every evening ended with a demo.
The AI Office carries that work into normal operations. Function leaders bring the workflows they are changing, show what people can now do, and resolve the decisions holding up the next build.
From the founder’s playbook: every function represented, leadership building alongside the team, a demo every evening.
06Make the economics real
Decide what the extra capacity is for.
Decide where released capacity will go: more customers served, hiring avoided or an expense removed. Measure the workflow’s volume, handling time, quality and running cost so finance can distinguish those outcomes.
07Build the next cycle
The first engagement should make the second build easier.
Onboarding context can help an authorized support agent understand the same customer. Exception records guide the next fix. Tested engineering checks protect the next automation. The first build leaves tools and knowledge the next team can reuse.
We begin with a scoped diagnostic, typically planned over a month. It ends with a decision about what to build first and the evidence against which that build will be judged.
- What you bring
- An executive sponsor, function owners and access to the relevant operating records: organisation structure, financial context, workflow volumes and the systems used to do the work.
- What you receive
- A workflow map, a ranked opportunity backlog and a baseline reviewed with finance. The implementation brief identifies the first workflow, its owner, required integrations and acceptance checks.
- How the next build is chosen
- Weigh economic value against variability, readiness and risk. Start where the team can reach a meaningful result, then reuse the context and tested components in the next workflow.
A few practical questions
Before we start.
What do we actually buy from Codos?
A scoped transformation engagement, supported by Company Brain and AI Interviewer where needed. Our engineers work with your teams to put automated workflows into production. The agreement defines deliverables and commercial terms.
Can you work with our existing AI and engineering teams?
Yes. We agree who owns the platform, integrations, process changes and daily operation. Your team’s business knowledge and technical constraints become inputs to the build; the engagement adds product, engineering and transformation capacity.
Which workflow should come first?
One tied to a material business goal, with usable evidence and a team able to change it. The diagnostic compares economics, volume, variability, technical readiness and risk, then produces a concrete first build.
What does leadership need to commit?
A goal, access to the relevant people and systems, and the authority to change how work gets done. The sponsor stays involved in the decisions about priorities, resources and the operating model.
