Seriun’s sell-out “Foundations for AI Success” at Crow Wood Hotel signalled a pragmatic, region-first approach to AI adoption that put governance, culture and stepwise pilots ahead of hype — and it offered Lancashire businesses a clear playbook for turning curiosity about generative AI into controlled, measurable change. The day-long launch of Seriun Intelligence, led by Seriun Directors Mark Edwards and Richard Lee, combined practical demonstrations (including Microsoft Copilot and Copilot Studio workflows), low-code Power Platform examples and hands-on AI “surgeries” that left delegates with concrete next steps rather than abstract promises.
Background / Overview
The event was designed as a roadmap for organisations at different starting points on the AI journey: auditing internal processes, building an innovation culture, and securing systems before scaling. Seriun presented AI adoption as a three-phase evolution — beginning with a preparatory audit (called
Pre‑Phase 1), moving through Phase I (“Getting Started with AI”), advancing to Phase II (building human-led agents and automation), and looking ahead to Phase III (deploying agents to run end‑to‑end workflows across the enterprise). Sessions emphasised the message that progress is earned through small, measurable pilots rather than single monumental investments.
Speakers from the local ecosystem reinforced this pragmatic theme. Clare Kay of Inturi Technology Solutions showcased Power Platform use cases — Power Apps, Power BI and Azure Functions — that reduce friction between business processes and reporting. Christine Lambe of East Lancashire Chamber of Commerce highlighted training and upskilling as critical to lasting transformation. The event closed with AI surgeries: small‑group clinics where businesses mapped readiness, identified quick wins, and received tailored advice from Seriun’s team.
Why this launch mattered for regional businesses
The Seriun Intelligence launch followed a model that has proven effective for SME-centric AI rollouts: bring platform demonstrations, local case studies and funding or skills partners together in one session so decision-makers leave with both inspiration and an execution path. That practical, regionally-tailored approach has repeatedly demonstrated higher pilot success rates than stand‑alone vendor marketing days because it addresses local constraints — skills, budgets and appetite for change — in tangible terms. Regional events that couple technical demos with governance and funding advice produce more durable outcomes for SMEs than talk-only conferences.
Key takeaways for Lancashire businesses were deliberately concrete:
- Start with process audits and measurable KPIs before deploying models.
- Use vendor tools to prototype faster but insist on portability and data export options.
- Build governance playbooks (data classification, DLP, logging) alongside pilots.
- Invest in short, practical training and hands-on mentoring to avoid “shadow AI” adoption.
These practical recommendations mirror accepted playbooks for moving AI pilots from curiosity to production, which emphasise staged testing, human verification and governance controls as mandatory elements of scale.
Foundations for AI Success: the Seriun framework
Pre‑Phase 1 — Audit, culture, security
Seriun positioned the first work as housekeeping: map where data and processes live, catalogue repetitive tasks, and identify obvious governance gaps. The objective is to create a bounded environment where early experiments won’t expose sensitive data or create downstream liabilities.
Why this matters: rapid pilots with no inventory frequently create blind spots — shadow AI and orphaned automation become audit risks if organisations can’t answer simple questions like “Which copilot has access to our sales data?” or “Who owns this agent?” The Seriun approach aligns with best practice frameworks that recommend early inventories, role‑based access and a named owner for each agent or bot.
Phase I — Getting Started (safe, practical tools)
Phase I focuses on accessible wins: Microsoft Copilot, safe prompting environments, sandboxed tenants and the development of basic AI policies. Seriun recommended:
- Testing Copilot in limited groups (e.g., Sales or Operations) with defined success metrics.
- Standardising safe prompting patterns to limit data leakage.
- Creating an “AI champion” network to surface early lessons and templates.
Copilot and similar tools can drive immediate time savings (meeting summarisation, draft generation) — but only when paired with policy and telemetry. Successful pilots track not just adoption but
quality metrics: how often outputs required human correction, number of DLP events, and measurable time saved. These are the KPIs that convert pilot momentum into budgeted production projects.
Phase II — Unlocking advanced AI (Copilot Studio, human‑led agents)
Phase II is where low‑code/no‑code agent building and human-in-the-loop automation begin to deliver scale. At the event Seriun demonstrated how Copilot Studio and Power Platform tooling let organisations create human‑led agents to automate repetitive, rule‑based tasks while keeping humans in supervisory loops.
- Copilot Studio provides a graphical environment to build agents that connect to internal knowledge and services — and can be published to Teams and Microsoft 365 channels for frontline use. Recent product updates extend agent capabilities to simulate actions in user interfaces, enabling automation where no API exists. These developments accelerate practical automation but also change the attack surface for organisations, which must treat agent permissions and publishing workflows as security‑sensitive operations.
- Power Platform tools — Power Apps, Power BI and Azure Functions — remain the pragmatic glue for low‑code solutions that integrate corporate data and deliver measurable business outcomes (logistics, reporting, case management). Microsoft’s own guidance highlights how Dataverse, Power Automate and Azure Functions work together to modernise applications and make AI‑driven copilots available in business apps.
Seriun’s emphasis was practical: build one reusable agent well, measure time saved and error rates, then replicate. Don’t automate entire decision chains without human oversight.
Phase III — Shaping the future of work (agents running workflows)
Phase III sketches a future in which agents orchestrate entire workflows, handing off to humans only for decisions that require judgment. Tools like Power Apps, Power Automate and broader Microsoft integrations make it technically possible to stitch agents into end‑to‑end processes — but governance, traceability and ownership become paramount.
- Organisations must implement audit trails for agent actions, define escalation rules, and require human sign‑off gates for high‑risk outputs (financial transactions, legal communication, personnel changes). The discipline of agent lifecycle management — including registration, owner attribution and periodic audits — is what separates transient pilots from durable automation programs.
Voices from the room: real use cases and practical training
Guest speakers added local credibility to the technical sessions. Clare Kay showed how low‑code solutions are being used today to:
- Streamline logistics workflows with Power Apps front ends connected to Dataverse and Azure Functions.
- Improve reporting with Power BI templates that deploy with automated configuration.
- Integrate disparate systems using custom connectors and serverless functions to remove manual reconciliation.
These use cases illustrate the pragmatic, measurable wins that small businesses can expect when they adopt a low‑code + serverless integration pattern. Microsoft’s Power Platform guidance validates this approach by describing Dataverse, connectors and Azure extensibility as foundational to modernisation work. Christine Lambe’s advice on training was equally pointed: short, scenario‑based upskilling — not long theoretical modules — accelerates adoption. Upskilling must be business‑led (frontline staff co-design pilots) and practical (how to prompt safely, when to escalate, how to validate outputs). These human measures are as important as the technical ones.
What the event did well
- Practicality over spectacle. The day prioritised real, testable use cases rather than aspirational keynotes. That lowers the barrier to a first pilot and helps attendees convert insight into action.
- Balanced mix of vendor tooling and local expertise. Demonstrating Microsoft Copilot alongside Power Platform case studies provides both the platform signal and the recipes for implementation.
- Hands‑on clinics. The AI surgeries model is an effective way to convert one‑day exposure into an actionable list of next steps — and the event generated immediate pilot ideas that local teams could follow up on.
- Regional ecosystem thinking. By linking technical demos to local skills and training support, the event reduced friction for SMEs that otherwise struggle to mobilise internal change. This ecosystem approach mirrors successful regional programmes that combine skills, funding and local implementers.
Risks and blind spots — what leaders must not ignore
The Seriun message — small, safe steps — is the right stance. But scaling AI in business still carries real risks that demand attention from the outset.
1) Agent publishing is an attack surface
Tools like Copilot Studio increase velocity for automation but can also be used maliciously. Recent security research by Datadog Security Labs shows how Copilot Studio agents can be misused for OAuth phishing (a technique dubbed “CoPhish”), enabling token theft and downstream account compromise. These findings are corroborated by security reporting and industry coverage — organisations must monitor agent creation, restrict application consent, and harden administrative roles to prevent privilege abuse. Practical steps:
- Restrict who can publish agents and require admin review for public or cross‑tenant links.
- Enforce least privilege on Entra (Azure AD) consent policies and disable unrestricted app creation for regular users.
- Monitor sign‑in and consent events for anomalous approvals.
2) Governance gaps lead to silent data leakage
Easy-to-build agents that connect to corporate knowledge can accidentally expose PII or sensitive IP if knowledge boundaries are not defined. Short workshops that stop at “how to use” and omit governance leave organisations exposed to exfiltration and regulatory risk. Effective pilots must include classification, DLP, and logging from day one.
3) Vendor lock‑in and portability
Rapid pilot success is appealing, but heavy investment in a single vendor’s agent format risks lock‑in. Organise pilots with exportable data, defined migration paths and contractual language about data use and portability.
4) Skills mismatch
Tool adoption without durable upskilling creates “fragile adoption” — employees use sanctioned solutions for a brief period, then revert to shadow tools or under‑utilise capabilities. Build ongoing mentoring and embed AI competence into role progression and performance frameworks.
Measured recommendations for Lancashire businesses
- Start with a two‑week AI readiness sprint.
- Inventory data sources and map sensitivity.
- Identify one low‑risk, high-frequency process (e.g., meeting summarisation, email triage).
- Define measurable KPIs: time saved, error rates, human override frequency.
- Set up a sandbox tenant and clear publishing controls.
- Use isolated environments or scoped Copilot Studio deployments to test agents.
- Require admin approval to publish agents beyond the pilot cohort.
- Create an AI charter and register agents.
- A short, plain‑language policy covering acceptable data, human sign‑off rules and escalation procedures.
- Maintain an agent registry with owner, last audit date and risk rating.
- Invest in short, scenario-based training.
- Two‑hour workshops for makers and a half‑day “champion” track for managers.
- Encourage template sharing and micro‑certifications for prompt design and verification.
- Pilot with measurement and a budget cap.
- Limit pilots to defined compute or calendar budgets to prevent runaway projects.
- Publish results and a decision gate: stop, iterate or scale.
These steps follow a proven pattern for converting small pilots into sustained capability without overextending budgets or exposing the organisation to unmanaged risk.
The Microsoft stack: powerful enabler, added responsibility
Seriun’s demonstrations — Copilot, Copilot Studio, and Power Platform integrations — reflect a broader reality: Microsoft’s ecosystem now provides end‑to‑end tooling for building copilots, agents and low‑code apps that integrate with corporate systems. Recent platform updates have added capabilities such as agent “computer use” (enabling UI automation) and deeper reasoning agents for analytics and research tasks, which accelerate what organisations can automate. At the same time, the more powerful these tools become, the more discipline organisations need for governance. Microsoft documentation and Power Platform guidance explicitly recommend controls — Dataverse security, connector governance and staged deployments — which must be implemented before broad rollouts.
What to expect next from Seriun and the local scene
Seriun confirmed plans for additional AI‑focused events to continue supporting the region’s business community. The launch was not a one‑off marketing push; it was a practical entry point into a longer regional AI capability building programme that emphasises skill-building, governance and funded follow‑through. This model increases the chance that pilots become institutionalised capabilities rather than short-lived experiments. Regional programmes that link training, funding signposting and vendor access typically demonstrate higher adoption durability and better ROI for SMEs.
Conclusion: the balanced path to AI value
The Seriun Intelligence launch at Crow Wood Hotel delivered a timely mix of optimism and caution. It showed Lancashire businesses that AI’s immediate value is real, but the route to sustained benefit is methodical: audit first, pilot with measurement, govern continuously, and scale only after proving outcomes.
The day’s strongest message — that success in AI comes from small, deliberate steps rather than single giant leaps — is the right one. For organisations ready to act, the practical recipe is clear:
- Pick one high‑value, low‑risk process and run a bounded pilot.
- Put governance, logging and human sign‑off rules in place before publishing agents.
- Invest in short, scenario-based training so staff can use tools safely and creatively.
- Treat agent publishing, consent and admin roles as critical security controls.
When those fundamentals are in place, tools like Microsoft Copilot, Copilot Studio and Power Platform become powerful enablers for productivity and automation. But as recent security research into agent-based OAuth phishing demonstrates, the same tools can amplify risk if they are not managed with the same rigour organisations apply to identity, data protection and application lifecycle management. Seriun’s launch has given Lancashire businesses an invitation: start small, measure hard, and govern strictly. The payoffs are real — time saved, cleaner processes and better reporting — but they arrive to organisations that treat AI as an ongoing capability rather than a one‑day event.
Source: Lancashire Business View
Seriun Launches Sell-out AI Event at Crow Wood Hotel | LBV Hub