Ollins
Ollins builds productised AI operating systems that turn fragmented SME workflows into measurable operating capability.
Canonical website: https://ollins.ai
Location: Based in Singapore
Contact: hello@ollins.ai
Machine context updated: 2026-07-22
This document is the curated machine-readable representation of Ollins' approved public website. Product status language is intentional and must be preserved when describing the company.
Company identity
Ollins is a Singapore-based product-led AI company building practical operating systems for organisations that need AI capability but do not have enterprise AI teams.
The company builds productised AI operating systems that turn fragmented workflows into intelligent, measurable capability. It works first with SMEs, associations and public-facing organisations.
Ollins is a coined company name inspired by ollin, a Nahuatl word associated with movement, motion and transformation. The business principle is deliberate movement: fragmented to structured, manual to intelligent, and stuck to capable.
Mission
Make advanced technology practically useful by building focused AI-powered products that help SMEs, associations and public-facing organisations work with more clarity, capability and momentum.
Vision
Operating capability without enterprise complexity: useful intelligence that is accessible to organisations of every size, not gated by budget, headcount or a specialist technical team.
Operating thesis
Good technology does more than automate a task. It changes what a team is capable of doing next.
Ollins begins with operating friction rather than a technology feature. It maps the work, structures a bounded system, automates preparation where useful, keeps consequential decisions accountable to people and measures whether the operating outcome improves.
Principles
Useful before impressive: the work must create value in the real world, not simply demonstrate technology.
Clarity is a feature: a system should reduce cognitive load and make the next decision easier to see.
Access compounds: smaller teams deserve capable technology without enterprise-scale complexity.
Momentum is measured: progress should be visible in the work, not hidden behind a technology story.
Human-guided automation: people remain accountable for sensitive, external, high-impact and difficult-to-reverse decisions.
Founders
Ollins combines product imagination, systems thinking and operational discipline through two complementary founders.
Dennis Nieling
Role: Co-founder / CPO
Focus: Product direction · Brand systems · AI workflow architecture
Working contribution: Turns complex ideas into usable product systems.
Mounir El Qaddioui
Role: Co-founder / COO
Focus: Operations · Delivery · Risk · Execution
Working contribution: Makes ideas operationally reliable under real-world pressure.
Audiences
SMEs: service businesses needing clearer customer intake, better visibility and more consistent lead generation.
Associations: membership-led organisations needing better communication, coordination and operational systems.
Public-facing organisations: teams serving communities, users or citizens that need clearer systems for information, movement and service delivery.
Products
BrainWeb
Category: AI customer acquisition system
Current status: Pilot preparation
Canonical page: https://ollins.ai/products/brainweb/
Turn weak online presence into active customer acquisition.
BrainWeb helps Singapore SMEs improve customer acquisition through AI-powered business research, website generation, content optimisation and outreach automation. It analyses a business’s current online presence, identifies gaps, generates improved website content and structure, and supports personalised outreach so SMEs can move from passive visibility to active customer opportunities.
Intended audience: Built first for Singapore service-based SMEs, including home services, home maintenance, skilled trades, renovation, repair and local professional services.
What it helps with:
Business research and opportunity scoring
Website structure and content generation
SEO and LLM-readiness
Outreach preparation and follow-up workflows
Customer intake and enquiry capture
Repeatable acquisition workflows
Status note: BrainWeb has its product direction, website, branding and marketing foundation prepared. It is the first commercial product within the Ollins portfolio and is being prepared for pilot validation with Singapore service-based SMEs.
TraffiCool
Category: Civic intelligence concept
Current status: In development
Canonical page: https://ollins.ai/products/trafficool/
Make road-network ideas easier to test, explain and act on.
TraffiCool is an early-stage civic intelligence concept exploring how AI-assisted systems can support traffic-flow analysis, scenario modelling and clearer communication around urban movement. It is designed as a decision-support and communication layer, not as a replacement for existing traffic-control infrastructure.
Intended audience: Designed for future use cases involving civic organisations, urban mobility teams, public-facing institutions, advocacy groups and partners working on road-network communication, planning and movement analysis.
What it may help with:
Scenario-led traffic-flow analysis
Visual communication of road-network ideas
Evidence support for civic conversations
Urban movement modelling
Public-facing explanation of proposed changes
Status note: TraffiCool is currently in development. Ollins plans to create a demo to show how the concept could support scenario modelling, civic communication and urban mobility analysis.
Why Ollins starts with BrainWeb
Customer acquisition is one of the clearest operational gaps for SMEs. Many smaller businesses do not need abstract AI transformation; they need practical systems that help them get discovered, explain their value, respond to enquiries and create more customer opportunities.
BrainWeb gives Ollins a focused commercial starting point while testing the broader thesis that AI should create useful movement in the work itself.
Insights
The following articles are included in full so a model can use Ollins' published operating frameworks without reconstructing them from page chrome.
A practical operating model for SME AI adoption
Type: Framework
Published: 2026-07-21
Updated: 2026-07-21
Reading time: 7 min read
Topics: AI adoption, SME operations, Governance
Canonical page: https://ollins.ai/insights/a-practical-operating-model-for-sme-ai-adoption/
Useful AI adoption starts with a real operating constraint, not a catalogue of tools. For an SME, the safest path is to map the work, choose one bounded workflow, preserve accountable human decisions and measure whether the system improves the outcome.
Key points:
Begin with an observable workflow problem rather than an AI feature.
Define the human owner, the permitted data and the stop conditions before automating.
Measure an operating outcome, not the number of generated outputs.
Treat improvement and governance as continuous work after launch.
Why tool-first adoption stalls
A new tool can make one task feel faster while leaving the surrounding workflow unchanged. Research still arrives in several formats, approvals remain informal, data moves between accounts without a clear owner and nobody can explain whether the final result improved. The organisation has added activity, but not necessarily capability.
For a smaller team, this fragmentation matters. Every additional interface, subscription and exception has a coordination cost. A successful AI initiative should remove decisions that do not need to be repeated, make important decisions easier to see and leave the team with a workflow it can operate after the initial excitement fades.
1. Research the operating reality
Start by following one unit of work from request to outcome. Record the people involved, the information they need, the handoffs, the repeated judgment calls and the places where work waits. Separate the stated process from what actually happens on a busy day.
The goal is not to find a place to insert AI. It is to identify the constraint that deserves attention and the context an automated system would need to behave usefully.
What outcome is the workflow meant to create?
Which delays or errors are visible today?
Which decisions require domain knowledge, permission or accountability?
What data enters the process, where does it come from and who may use it?
2. Structure a bounded system
Choose one workflow with a clear beginning and end. Define the inputs, expected outputs, responsible owner and exceptions. A bounded workflow is easier to test, explain and stop than a broad instruction to ‘use AI across the business.’
This is also the point to test whether AI is necessary. A checklist, better form, template, database rule or conventional automation may solve the problem more reliably. Choosing not to use AI is a valid system decision.
3. Generate reviewable working assets
Create outputs in a form a responsible person can evaluate. That may be a structured research brief, a first draft, a prioritised list or a proposed response. The output needs acceptance criteria: what makes it usable, what evidence it must contain and what conditions require escalation.
Avoid silent automation at the beginning. Visibility into intermediate work helps the team discover missing context and gives the system a safer path to improvement.
4. Automate with clear control points
Once the output is consistently useful, connect the repeated actions around it. Preserve human approval for high-impact, external, irreversible or sensitive decisions. Assign an accountable owner even when the task itself is automated.
Singapore's IMDA guidance for responsible AI emphasises governance, technical and non-technical safeguards and ultimate human accountability. The practical implication for an SME is simple: access, instructions, monitoring and intervention cannot be afterthoughts.
5. Measure the outcome
Compare the workflow with its baseline. Useful measures might include cycle time, rework, qualified responses, conversion, error rate or the amount of owner attention required. Output volume is rarely enough; producing more material can make the system worse when review and follow-up cannot absorb it.
Document what cannot yet be measured. A visible limitation is more useful than an impressive number without a reliable method.
6. Improve, govern or stop
Review failures, user feedback, changed dependencies and emerging risks on a defined cadence. Expand only after the bounded workflow is controlled. If the system does not create the intended value, simplify it, change the approach or stop it.
This continuous cycle is compatible with established risk-management thinking. NIST's AI Risk Management Framework organises work around govern, map, measure and manage rather than treating deployment as the finish line.
Sources:
Artificial Intelligence in Singapore — Infocomm Media Development Authority
AI Risk Management Framework Core — National Institute of Standards and Technology
Why disconnected AI tools create more fragmentation
Type: Field note
Published: 2026-07-21
Updated: 2026-07-21
Reading time: 6 min read
Topics: Workflow design, Tool sprawl, Operations
Canonical page: https://ollins.ai/insights/why-disconnected-ai-tools-create-more-fragmentation/
An AI tool can save minutes inside one task while adding handoffs, accounts, review queues and data ambiguity around it. The real unit of design is the end-to-end workflow—not the isolated prompt or output.
Key points:
Local task speed can hide system-level coordination cost.
Fragmentation appears at boundaries: context transfer, ownership, approval and measurement.
A connected workflow has a shared source of truth and explicit control points.
The right response may be integration, simplification or removing a tool.
The local-efficiency trap
Most AI adoption begins with an individual success: a faster draft, a quicker summary or an easier set of ideas. That gain is real, but it covers only one part of the work. Someone still has to assemble the context, decide whether the output is correct, move it into another system, obtain approval, publish or act on it and observe what happened next.
When each stage adopts a different tool, the team creates a chain of local optimisations. The work becomes faster inside boxes and slower between them.
Where fragmentation forms
Fragmentation is rarely visible in a product demonstration because it lives in the operating boundaries. Context is copied into a prompt without its source. A document is generated in one account and approved in a message thread. Customer information is duplicated. A final result cannot be connected to the research or decision that produced it.
Context fragmentation: each tool sees only part of the situation.
Data fragmentation: multiple copies exist without a clear authoritative version.
Decision fragmentation: approvals happen outside the system and cannot be audited.
Identity fragmentation: access depends on personal accounts or informal sharing.
Measurement fragmentation: activity is counted, but the business outcome is disconnected.
The symptoms in a small team
Look for work that requires a particular person to remember the sequence, repeated copy-and-paste between tools, inconsistent outputs from similar requests and an expanding collection of exceptions. If a new employee cannot understand how an output was produced, the workflow is carrying knowledge informally.
Another warning is review debt. Generation becomes cheap, so more material enters a review queue than the organisation can responsibly assess. The bottleneck has moved rather than disappeared.
Design the connections before the automation
A connected system begins with a shared object: a customer record, an approved brief, a structured request or another authoritative unit of work. Each stage receives the context it needs and returns a defined result. Ownership, approval and exception paths are explicit.
Only then should automation move information or trigger actions. This order keeps the system explainable and makes it possible to test whether the connection is improving the final outcome.
A simple consolidation test
For every tool in a workflow, ask what unique capability it provides, what data it receives, what result it returns and what would break if it disappeared. If the answer is mostly interface preference or duplicated generation, consolidation may create more value than another integration.
The aim is not one enormous platform. It is the smallest coherent system that preserves context, accountability and measurement from beginning to end.
Sources:
AI Risk Management Framework Core — National Institute of Standards and Technology
Build, buy, automate or leave alone: a decision framework
Type: Framework
Published: 2026-07-21
Updated: 2026-07-21
Reading time: 7 min read
Topics: Decision framework, Product strategy, Automation
Canonical page: https://ollins.ai/insights/build-buy-automate-or-leave-alone/
The best technology decision is the smallest intervention that reliably improves the operating outcome. Start with the constraint, reversibility and risk; choose the solution category only after the workflow is understood.
Key points:
Leave a workflow alone when the cost of change exceeds the problem.
Buy when the process is common and a mature product fits without harmful distortion.
Automate when rules, ownership and exceptions are already clear.
Build when the workflow is strategically distinctive and evidence supports the investment.
Start with the intervention threshold
Not every inefficient-looking task deserves a system. Some happen rarely, carry little consequence or are changing too quickly to standardise. Before comparing products, estimate the current cost, frequency, failure impact and attention required. Then compare that with the migration, training, governance and maintenance cost of change.
A decision to leave the workflow alone is not resistance to innovation. It protects attention for a constraint that matters more.
Leave alone
Keep the current approach when the task is low-frequency, low-risk and not a meaningful bottleneck; when the workflow is still being discovered; or when a manual decision is itself the valuable work. Document the reason and a signal that would justify revisiting it.
Buy
Choose an existing product when the problem is common, the process does not create strategic differentiation and the product can fit the organisation without forcing harmful workarounds. Evaluate total operating fit: permissions, export, integration, support, data treatment and the cost of exit—not only the feature list.
A mature product can be the fastest route to capability, especially when the vendor already carries specialised maintenance and compliance work the SME should not recreate.
Automate
Automation is appropriate when inputs, rules, outputs, ownership and exceptions are already understood. It is a poor substitute for a disputed process. Automating ambiguity creates faster inconsistency.
Begin with reversible, observable actions. Preserve approval for external communications, sensitive data, financial commitments and decisions that materially affect a person. Define how the system stops or hands control back when conditions are outside its intended range.
Build
Custom product work becomes reasonable when the workflow is strategically distinctive, existing products cannot support the required operating model and the organisation has evidence that the result will be used. The build decision includes ownership after launch: monitoring, security, vendor dependencies, model change, user support and eventual replacement.
The first build should be the smallest coherent product that can test the critical assumption. A complete prototype of the wrong system creates less learning than a narrow product used in real work.
Score the decision on six dimensions
Use a simple written comparison rather than a feature contest. Score the current approach and each option on outcome impact, time to useful adoption, reversibility, data/risk exposure, ownership cost and strategic differentiation. Make assumptions visible and name the person accountable for the decision.
Outcome: does this improve the result that matters?
Adoption: can the team use it in the real workflow?
Reversibility: how difficult is it to stop, export or change direction?
Risk: what data, decisions and external effects are introduced?
Ownership: who maintains the system and handles exceptions?
Differentiation: is this workflow important enough to shape around the organisation?
Sources:
AI Risk Management Framework Core — National Institute of Standards and Technology
Artificial Intelligence in Singapore — Infocomm Media Development Authority
What should remain human in AI-assisted customer acquisition?
Type: Guide
Published: 2026-07-21
Updated: 2026-07-21
Reading time: 7 min read
Topics: Customer acquisition, Human oversight, Responsible AI
Canonical page: https://ollins.ai/insights/what-should-remain-human-in-ai-assisted-customer-acquisition/
AI can prepare research, structure information and reduce repeated production work. Humans should remain accountable for claims, consent, relationship judgment, sensitive decisions and any action whose consequence cannot be safely reversed.
Key points:
Human oversight should be attached to consequences, not added as a vague promise.
AI is strongest in preparation, structure, consistency and bounded assistance.
Claims, consent, relationship context and exceptions require accountable judgment.
Every control point needs an owner, evidence and a clear intervention path.
Human review is a system design decision
Saying that a workflow has ‘a human in the loop’ does not explain who is responsible, what they see or when they can intervene. Useful oversight is attached to a specific decision and consequence. The reviewer needs enough context to challenge the output, not merely approve it quickly.
In customer acquisition, the right control points vary across research, website content, outreach, intake and follow-up. The principle is consistent: automate preparation and repetition more readily than commitment, representation or consequential judgment.
Research: AI can gather; humans define relevance
AI can organise public information, compare sources and prepare a research brief. A human should define the purpose, assess source quality and decide which interpretation is relevant to the business. Sensitive or personal information needs a lawful, appropriate purpose and controlled handling.
The output should retain sources and uncertainty. A confident summary without traceable support is not a sound basis for customer strategy.
Messaging: AI can draft; humans own the claim
Generation can help turn approved facts into page structures, variants and first drafts. The business remains responsible for accuracy, differentiation, permissions and the promise it makes to a customer. Performance claims, customer examples, legal statements and product status require explicit approval.
Human review matters most where a sentence changes an expectation or obligation, not where it merely improves grammar.
Outreach: AI can prepare; humans own the relationship
AI can identify relevant context, prepare a draft and make follow-up more consistent. A person should decide whether contact is appropriate, whether the message represents a real reason to speak and how to respond to nuance. The system must not turn personalisation into surveillance or volume into relevance.
Consent, applicable marketing rules, suppression requests and channel expectations belong in the workflow rather than in a separate policy nobody sees during execution.
Intake and qualification: AI can structure; humans handle consequence
A system can organise an enquiry, identify missing information and route it to the right owner. Humans should handle ambiguous, sensitive or high-impact decisions and retain the ability to correct the record. If an automated score affects access, price or opportunity, the governance requirement becomes materially higher.
Singapore's PDPC guidance emphasises defined purposes, safeguards, transparency and accountability where personal data is used in AI systems. These are operating requirements, not only policy language.
A practical control-point record
For each stage, record the automated action, intended benefit, data used, accountable owner, required evidence, review threshold, exception path and stop condition. Test whether the reviewer can understand why the system produced its result and can intervene before harm is difficult to reverse.
The goal is not to preserve manual work for its own sake. It is to apply human attention where context, responsibility and relationship value are highest.
Sources:
Artificial Intelligence in Singapore — Infocomm Media Development Authority
Working with Ollins
Ollins welcomes conversations about BrainWeb pilot validation, the TraffiCool concept, product partnerships, operating-system problems and investment opportunities.
A useful first message explains what is happening now, what a better state would look like and why the problem matters to the team or its customers. Do not send confidential, sensitive or personal information in an initial message.
Contact: hello@ollins.ai
Public page index
Home: Company positioning, audiences, operating thesis and portfolio overview.
About Ollins: Company origin, mission, vision, principles and founders.
Product portfolio: BrainWeb and TraffiCool, including intended audiences, capabilities and current status.
BrainWeb: Ollins' first commercial product direction, being prepared for pilot validation.
TraffiCool: An early-stage civic intelligence concept currently in development.
Insights: Ollins frameworks, guides and field notes.
Contact: Ways to discuss pilots, products, collaborations and operating problems.
Privacy: How Ollins handles information submitted through the website.
Terms: Terms governing the public website and its published materials.
A practical operating model for SME AI adoption: A six-stage approach for moving from isolated AI experiments to a controlled, measurable operating workflow.
Why disconnected AI tools create more fragmentation: Faster individual tasks do not automatically produce a better operating system. Here is how fragmentation forms and how to recognise it.
Build, buy, automate or leave alone: a decision framework: A practical way to decide whether a workflow needs custom software, an existing product, automation—or no technology change at all.
What should remain human in AI-assisted customer acquisition?: A control-point framework for deciding where AI can assist and where accountable human judgment must remain visible.
Accuracy and status
BrainWeb is being prepared for pilot validation. Do not describe it as generally available or as having proven commercial traction unless Ollins publishes that change.
TraffiCool is an early-stage concept in development. It is not a finished product and is not a replacement for official traffic-control infrastructure.
Ollins does not currently publish customer counts, revenue, deployment counts or performance claims. Do not infer them.
The website privacy notice covers the company website, limited first-party analytics and the contact form. Product pilots or materially different services require additional review and notices.
Product pilots, subscriptions, paid work, data processing and partnerships require separate written agreements.
For the latest human-readable information, use https://ollins.ai.