Maybe* Enterprise
Execution you can trust inside real workflows
Defined work completed with clear rules and oversight.
Most organisations we speak to have already adopted AI. Very few have seen it complete real work inside their business stack. That gap is what we close.
A conversation with an expert AI Analyst.
Check submitted document against requirements
Requirements check, response to the submitter, and updated project record.
-
Document receivedUploaded to the project system. An authorised event starts the Task.Input Gateway
-
Requirements checkedThe document is compared against the project requirements.Agent, model provider A
-
Missing information identifiedEach gap is listed against the requirement it relates to.Agent, model provider A
-
Response preparedA reply to the submitter is drafted and held for review.Agent, model provider B
-
Approval givenThe accountable owner reviews the response and approves it.Person
-
Record updatedThe project record shows the outcome and every step taken.Agent, model provider B
An example Task in a project system. A person steps in where an exception or approval needs their judgement.
Trusted by start-ups, PLCs, and the UK Government.
Cyber Essentials certified
The outcome
The AI transformation you want, delivered one Task at a time
It is what the board is asking for, so we name it plainly. We deliver it with your team, one finished Task at a time, with the time saved shown by role and by Task. The board sees exactly what work AI is doing across the team, and where it is working well.
Repetitive work completes itself
Information and requests become completed work. The check is run. The response is prepared. The record is updated.
Work runs across your systems
Agents work inside your existing software, your workplace channels and the products you build on our API. Nothing gets replaced.
Judgement has space
Maybe* does not replace your thinking. It protects the time and space you need to do it.
Governance and control
Governed and auditable is not a feature. It is the buying criterion
The work is defined, the rules are clear, and every step can be shown. You need to understand what AI has done, and manage how it operates.
Scope
You decide which Agents are available, where they operate and which tools they can use.
Identity and access
Your existing identity and access systems stay part of the decision about what an Agent is permitted to see or change.
Boundaries
Context follows the work. It respects the boundaries between organisations, users, projects and resources.
Approval points
Human review applies where the workflow requires it. Proven write workflows run automatically within their agreed permissions.
Records
Records of activity and outcomes, with model usage and cost in view.
Limits and recovery
Limits on execution, and defined handling for failures or interrupted work.
Evaluation
Each workflow is checked against its intended result, within the agreed boundaries.
Autonomy
Set for each workflow. Expand it as evidence supports doing so. Restrict or disable a capability at any time.
The build finishes.The assurance does not.
A layer on top, not a replacement
We sit on top of what you already run. We reuse the logic already there. Nothing gets replaced. The Embedded Intelligence Layer is deployed into the infrastructure you provide, with defined responsibilities for identity, data, secrets, operation and support.
Your infrastructure. Your product. Continuously upgraded.
What runs
From incoming information to completed work
Maybe* Enterprise understands what is happening, decides what needs attention, coordinates specialist Agents and takes authorised action.
-
01
Receive work wherever it originates
The Input Gateway brings together requests and signals from connected conversations, email, documents, applications and scheduled activity. Your own software can also call Agents directly through the API. Work begins with an explicit request or an authorised event, such as a document being uploaded or a business record changing.
Input GatewayAPI -
02
Listen and judge what matters
Listening gives AI awareness of activity within an agreed scope. Judgement determines whether to remain silent, notify someone, propose an action, request approval or start work. You define what the system observes, what warrants intervention and how independently it operates.
ListeningJudgement -
03
Coordinate the work
Task Engine manages work that spans multiple steps, Agents and systems. It assigns specialist Agents, carries context between stages, manages dependencies and tracks progress towards an outcome. Agent Builder defines each Agent: its purpose, instructions, knowledge, tools and model.
Task EngineAgent Builder -
04
Use different models within the same Task
One Agent uses one model provider to analyse information. Another uses a different provider to complete the next stage. Model selection reflects the work, your requirements, quality and cost. The Task remains one coordinated piece of work.
Multiple model providers -
05
Take authorised action
Agents retrieve and interpret information. They also create and update records, prepare documents, submit requests and carry out connected business processes. A person becomes involved where an exception or approval requires their judgement.
Agreed permissionsHuman review
Where it runs
Three ways organisations put it to work
You choose the experience your users see, the processes AI supports and the authority it has.
Improve your own operations
Deploy Agents and coordinated Tasks across internal processes. Your teams see finished work inside the systems and channels they already use.
Add AI to the product you already sell
Software providers add Agent capabilities to their own applications. You keep your brand, your customer experience and your customer relationship. Our API and execution framework run underneath.
Build new applications
Developers combine platform capabilities into specialist products and services, without building the whole Agent execution and coordination layer themselves.
Adopt the capabilities your use case needs. Agents running inside your product deliver value independently of the wider Listening and Task Engine experience.
How it is built
One platform, three distinct layers
The platform provides the underlying capabilities. Each product or deployment determines how those capabilities are presented and used.
The choices made in the Maybe* product do not define what you adopt. Configuration separates reusable capability from the choices that are yours, so the same Agent framework supports different sector terminology, organisational structures, processes and user experiences.
Customer number one of the platform. The shared capabilities, combined into our own customer experience, with its own packaging, workflows and commercial offer.
The same foundations, applied to your operations or your software products. Your interfaces, identity systems, business rules and connected systems, supported by our deployment team.
Reusable capabilities for configuring and running Agents, connecting tools and knowledge, receiving work, coordinating Tasks and controlling execution. Developers use these capabilities to build their own applications and services.
Delivery
A reusable platform, and the team to deploy it around your business
Maybe* Enterprise includes the work needed to make the platform useful in a real business: understanding processes, connecting systems, configuring Agents, testing outcomes and supporting deployment.
Our team works alongside your product, engineering and operational teams. What is specific to you sits around shared platform capabilities, so each deployment starts where the last one finished.
- Completed work
- Reduced handling effort
- Better turnaround
- Dependable operation
Prove one thing, then earn the next
No proof, no scale.
-
Phase 0Task Audit
Scope the proof Task
One real Task, one measurable outcome, agreed with the accountable owner.
-
Phase 1Setup
Run with defined metrics
The Task runs inside the boundary. Governance and audit are on from the first run.
-
ScaleOngoing programme
Expand on evidence
Adjacent Tasks and Agents are added only on proven outcomes.
In deployment
A concrete enterprise foundation
One Maybe* Enterprise Client is a construction common data environment (CDE) with over 80,000 global clients. Its Maybe* deployment puts Agents inside highly regulated processes, under the clients brand, connected to the systems and permissions their customers already runs.
- Built on
- Our core API, Agent Builder and the framework for running Agents.
- Runs
- Within the clients environment across multiple data centres in multiple regions.
- Provides
- Its own product experience, connected to its own systems and permissions.
The first conversation is about one Task and one measurable outcome
You do not operate Maybe*. Your business does.
Everything else follows from that. Evaluate Your AI Roadmap is a conversation about where your organisation stands, which Tasks carry the most value and which route through Maybe* makes sense.