Underpin Works is now an OpenAI Select Partner, helping businesses identify where AI can create meaningful commercial value and build practical AI systems using OpenAI technology.
For us, this is an important milestone.
But the partnership itself isn't the end goal.
The goal is to help more businesses move from experimenting with AI to actually using it inside the workflows that drive revenue, capacity, efficiency and growth.
Because access to AI is no longer the biggest challenge.
Knowing where it is worth using increasingly is.
What does becoming an OpenAI Select Partner mean for Underpin Works?
As an OpenAI Select Partner, Underpin Works is part of the OpenAI Partner Network.
Our work focuses on helping businesses identify valuable AI opportunities and turn them into working systems.
That can involve:
- identifying and prioritising AI use cases
- redesigning workflows around AI
- deploying and configuring repeatable AI systems
- building custom AI applications and agents
- connecting AI with existing business data and software
- introducing appropriate human review and approval
- measuring whether the implementation produces the intended business outcome
Our experience to date includes work with specialist recruitment agencies and marketing agencies, where we have explored and built workflows across business development, research, content, operations and client delivery.
But our approach isn't specific to those industries.
The underlying problems appear across sectors: valuable employee time being consumed by repetitive work, information fragmented across systems, processes that don't scale efficiently, and opportunities where AI can give people greater capacity.
As Underpin Works grows, we are expanding this work into other verticals where there is a strong business case for applying AI.
The industry may change.
The principle doesn't:
ROI first. Technology second.
Businesses don't have an AI access problem
A few years ago, simply gaining access to capable artificial intelligence was significant.
That has changed.
Businesses can now use AI to research information, analyse documents and data, generate content, reason through problems, interact with software and carry out increasingly complex workflows.
The list of things that can be automated is growing quickly.
But that creates another problem.
Which of them should be?
A workflow being technically possible to automate does not mean there is a strong business case for doing so.
A sophisticated AI agent might save very little money.
Meanwhile, a seemingly mundane three-minute task repeated hundreds of times across a team might consume days of productive capacity every month.
The technology doesn't tell you which problem matters more.
Understanding the business does.
We start with the business case, not the AI
Underpin Works was built around a simple principle:
Don't start with the technology. Start with where the business can move.
Before recommending an AI implementation, we look at questions such as:
- Where is valuable employee time currently being consumed?
- Which activities are limiting revenue or capacity?
- Where does work repeatedly get delayed?
- Which processes become expensive at scale?
- What knowledge does the business already possess but struggle to use?
- Where does human judgement remain essential?
- What would have to improve for an AI investment to generate a worthwhile return?
From there, we assess the potential value against the complexity of changing the workflow.
Sometimes that leads to an AI system.
Sometimes the process needs to be simplified first.
Sometimes existing software already solves the problem.
And sometimes the right answer is not to automate it.
Becoming an OpenAI Select Partner gives us an even stronger foundation for implementing the opportunities where the business case does hold.
From AI demonstrations to AI systems
There is an important difference between an impressive AI demonstration and a useful business system.
A demonstration has to work once.
A business system has to work repeatedly.
It needs the right context.
It has to work with existing data and tools.
It needs to handle exceptions.
Someone needs to know what happens when it gets something wrong.
Employees need to understand how it fits into their work.
And the output needs to be valuable enough for somebody to actually use.
This is often where AI projects become more difficult than they initially appear.
The model is only one component.
The workflow around it matters just as much.
What does AI implementation look like in practice?
The specific implementation depends on the business.
In a recruitment business, AI might help identify relevant hiring signals, research target accounts, connect opportunities with existing candidate knowledge and prepare better context for business development.
In a marketing business, it might help with research, SEO workflows, content operations, inbound enquiries, reporting or repetitive delivery processes.
In another industry, the valuable workflow may look completely different.
The important question isn't:
"Where can we add AI?"
It is:
"Where can AI meaningfully improve the economics, capacity or capabilities of this business?"
That is the question we want to answer before building anything.
AI for recruitment businesses
One area where Underpin Works has practical experience is AI for recruitment agencies.
Specialist recruiters often already possess valuable commercial information:
- candidate networks
- previous placements
- client relationships
- target accounts
- hiring activity
- years of market knowledge
The challenge is often turning that information into consistent action.
AI can potentially support workflows such as:
- monitoring relevant market and hiring signals
- researching prospective client accounts
- prioritising potential opportunities
- connecting opportunities with candidate knowledge already inside the agency
- preparing relevant business-development context
- reducing repetitive CRM and research work
- organising follow-up around approved opportunities
The objective isn't to automate the recruiter.
It is to reduce the operational work surrounding the areas where their judgement creates the most value.
The recruiter still decides who is worth contacting, why the conversation is relevant and how the relationship should develop.
AI for marketing businesses
We have also worked on AI workflows for marketing agencies.
There are increasingly powerful applications for AI across:
- research
- SEO
- content operations
- lead qualification
- inbound enquiries
- reporting
- repetitive delivery workflows
But simply producing more content or adding AI to every process doesn't necessarily improve agency economics.
The more useful question is:
Where can AI increase capacity, protect margin or improve the quality and consistency of delivery?
That might mean building a system that takes an approved SEO brief through research and towards a review-ready draft, as our SERP-to-Draft Content Engine does.
It could mean researching and qualifying inbound opportunities before they reach a salesperson.
It could mean automating repetitive operational steps around an existing service.
Or it might involve a workflow that is entirely specific to how that particular business operates.
Again, the objective is not AI adoption for its own sake.
It is improving the business.
The same approach applies beyond agencies
Recruitment and marketing are areas where we have practical experience, but the underlying opportunity is much broader.
Across industries, businesses have processes where:
- people repeatedly search for information
- employees copy information between systems
- knowledge is difficult to access
- valuable data isn't being used effectively
- administrative work limits the capacity of skilled employees
- customers or prospects wait unnecessarily
- decisions require information to be gathered and prepared first
- processes become increasingly expensive as the business grows
These are the types of problems we are interested in.
The specific technology might involve language models, AI agents, workflow automation, internal applications, retrieval systems, integrations or a combination of approaches.
But the technology comes after the problem.
As Underpin Works expands into additional industries, we intend to keep applying the same principle:
Find the commercial bottleneck first. Then decide whether AI belongs in the solution.
Automate the repetition. Preserve the judgement.
This becomes particularly important as AI becomes capable of doing more.
We don't believe every workflow should become fully autonomous.
In many businesses, the people involved are valuable precisely because of their judgement, relationships and accumulated knowledge.
A recruiter understands whether a candidate genuinely creates a reason to approach a client.
A marketing strategist knows when a technically correct piece of content still isn't good enough.
A salesperson understands nuances in an account that might never appear in the CRM.
A domain expert often recognises an exception long before a generic system would.
The better opportunity is often to design AI around those people.
Let the system perform more of the:
- research
- preparation
- organisation
- repetitive processing
- information retrieval
- administrative work
Keep people responsible for:
- judgement
- exceptions
- relationships
- strategic decisions
- approval where it matters
Automate the repetition. Preserve the judgement.
What kind of AI work does Underpin Works do?
Underpin Works helps businesses move from identifying an AI opportunity to deploying a working system.
There are broadly two ways we work: deployable AI systems and custom AI development.
Deployable AI systems
Where a valuable workflow occurs repeatedly across similar businesses, we can build the underlying system once and configure it around the individual company's processes, data and tools.
This reduces the amount of custom development required while still allowing the workflow to reflect how that business operates.
Custom AI development
Where the problem is genuinely specific to the business, we work backwards from the desired outcome.
We identify the opportunity, establish the business case and then design and build the appropriate system.
That can include:
- custom AI applications
- AI agents
- internal tools
- workflow automation
- knowledge and retrieval systems
- integrations with existing business software
- human-in-the-loop AI workflows
In either case, the principle remains the same:
Identify → Prove → Build → Improve.
What changes now that Underpin Works is an OpenAI Select Partner?
Our underlying philosophy doesn't change.
The partnership strengthens our ability to help organisations understand and implement OpenAI technology while remaining focused on the commercial outcome behind the implementation.
OpenAI's technology will continue to evolve.
Models will become more capable.
Agents will be able to complete increasingly complex work.
And the number of potential applications inside businesses will continue to increase.
That makes prioritisation more important, not less.
The businesses that benefit most from AI won't necessarily be those that adopt the most AI.
They will be the businesses that become good at identifying where AI genuinely changes the economics or capabilities of their organisation and then implementing it properly.
That's the work we want Underpin Works to do.
ROI first. Technology second.
Frequently asked questions
Is Underpin Works an OpenAI partner?
Yes. Underpin Works is an OpenAI Select Partner and part of the OpenAI Partner Network.
What does Underpin Works do?
Underpin Works helps businesses identify high-value AI opportunities and turn them into working systems.
This can include AI opportunity assessment, workflow design, deployable AI systems, custom AI development, AI agents, internal tools and workflow automation.
Which industries does Underpin Works work with?
Underpin Works works with businesses looking to apply AI to meaningful operational and commercial problems.
Our experience includes work with specialist recruitment agencies and marketing agencies, and we are expanding into other industries where there is a clear business case for AI implementation.
We are particularly interested in workflows where AI can increase capacity, reduce repetitive work, improve how information is used or enable a business to operate more effectively.
Does Underpin Works only work with recruitment and marketing agencies?
No.
Recruitment and marketing are industries where Underpin Works has practical experience, but the company's approach is not limited to those sectors.
We work from the business problem backwards and are expanding into other verticals where AI can create meaningful commercial value.
Does Underpin Works build custom OpenAI solutions?
Yes.
Where an existing deployable system isn't appropriate, Underpin Works can design and build custom AI applications, agents, internal tools and workflow automations around a specific business problem.
Does Underpin Works fully automate business processes?
Not necessarily.
Underpin Works focuses on automating repetitive work while preserving human judgement where it matters.
The appropriate level of automation depends on the workflow, business value, risk and decisions involved.