March 2026 AI Dinner - Stop implementing AI. Start driving business value

AI Dinner Event Growth: Rapid expansion needs larger venues; tickets sell out within a week. Next dinner on April 28.

Community Engagement: WhatsApp group has 100-150 members for exclusive event details, increasing ticket sales and promoting attendance.

Operational Intelligence Platform: Autumn AI raised £250,000 SEIS investment, focusing on enterprise markets and automating business operations.

Strategic Shift: Targeting large organisations with secure AI needs; exclusive agreements for the UK’s largest AI data centre planned in Scotland.

AI Tooling and Frameworks: Microsoft Agent framework chosen for development; planning migration to UK sovereign data centre for data security.

Real-World AI Challenges: Internal resistance to AI adoption; data quality affects decision-making processes, highlighting the need for human insights.

Notes AI Dinner Event Growth and Community Engagement The AI Dinner is expanding rapidly, necessitating a move to larger venues and new event formats to accommodate growing demand.

Venue expansion and new event formats planned as the current room capacity of 25-30 people is insufficient, with tickets now selling out within a week, compared to months previously (06:02)

Plans include rotating venues across North London and hosting a Prosecco reception for 100 people, with hopes to secure sponsorship to make it free The new format separates the reception and dinner to accommodate different engagement levels and increase overall attendance The event will maintain a monthly schedule on the last Tuesday, with the next dinner on 28th April at a yet-to-be-announced venue Growing digital presence and community management through a WhatsApp group limited to attendees, currently with around 100-150 members (08:06)

The group shares event details and AI discussions but maintains exclusivity to ensure quality engagement Tickets for dinners are released in this group first, accelerating ticket sales Social media promotion is encouraged by organisers to maximise attendance and community reach (09:39) Speaker and workshop highlights enhance engagement and learning with recent showcases of AI-powered solutions and workshops like “Ship an app in a day” led by Eric, designed to empower participants to build AI applications quickly (13:01)

Upcoming speakers include Kim Fara presenting an AI-powered index of European optical opportunities and Dave Killeen from Pendo sharing insights on AI personal chief of staff systems (14:21) These sessions aim to demonstrate practical AI application and inspire attendees to adopt AI natively in their businesses Autumn AI Platform Development and Enterprise Adoption Paul and Jeanie shared a detailed journey of developing the Autumn AI operational intelligence platform, highlighting technical innovation and a pivot to enterprise markets.

Development of Autumn platform to automate business operations began with custom software expertise and evolved through building AI agents that understand and automate workflows using natural language (33:14)

Initial approach mimicked organisational structures with multiple agents but proved slow and bureaucratic Pivoted to a single dynamic agent model that loads only necessary tools and skills, improving speed and flexibility (54:53) Autumn integrates with tools like N8N for workflow automation and supports OAuth-based credential management for scalability and security (47:28) Successfully raised £250,000 SEIS investment and formalised business structure in early June to transition from prototype to commercial SaaS, pausing most client work to focus on Autumn (48:52)

Faced internal team scepticism regarding AI replacing jobs, addressed by focusing on augmenting employee effectiveness rather than headcount reduction (50:34) Built operational context layers and vector memory to enhance AI’s understanding and collaboration within teams Shifted strategic focus to large organisations and sovereign AI infrastructure driven by UK government and large enterprises’ need for data sovereignty and secure AI operations (59:09)

Signed exclusive agreements to provide Autumn as the operations layer for the UK’s largest AI data centre planned in Scotland, scaling to 50,000 GPUs (01:02:39) Currently engaging with clients including police, Ministry of Defence, NatWest Bank, and insurance, targeting teams with 400+ members and aiming for 700 licenses per month (01:00:40) Emphasised privacy with encrypted and redacted data accessible only in UI, addressing security concerns of enterprise clients (01:08:15) AI Tooling Choices and Technical Strategy Paul discussed the practical decisions around AI tooling, frameworks, and maintaining independence from large American cloud providers.

Chose Microsoft Agent framework over custom Python code after Microsoft released a feature-rich, .NET-based agent framework, enabling rapid switch in two days while aligning with team expertise (01:04:08)

Cursor AI tool used extensively for code generation and enforcing architecture patterns, improving development efficiency and quality (01:05:21) Despite using Microsoft tools, planned migration to UK sovereign data centre to avoid US cloud regulations and data access risks under the Cloud Act (01:05:50) Adopted multi-agent orchestration strategy using different AI models and tools for specialised tasks rather than relying on a single model (01:20:14)

Leveraged lighter “nano” models for routine utility tasks and top-tier models for complex needs, balancing cost and performance Emphasised open-source and self-hosted AI agents as strategic advantage for data control and flexibility Addressed market differentiation and competitive pressures by focusing on unique operational context and business intelligence features beyond general-purpose AI assistants (01:07:20)

Recognised large organisations’ reluctance to allow non-expert employees to build tools, reinforcing Autumn’s value as a horizontal operational layer serving vertical market specialists (01:08:00) Product Vision, Strategy, and Customer-Centric Design Florence provided a comprehensive framework for building AI products grounded in strong vision, customer understanding, and strategic alignment.

Vision and mission must be clear, stable, and inspiring to guide long-term strategy amid rapidly changing AI technologies (01:27:02)

Used examples like Spotify’s vision to become the world’s creator platform and their North Star metric of time spent listening to music to illustrate alignment Emphasised that strategy is the flexible “how” to reach the vision, requiring continuous adjustment based on market and technology changes (01:36:31) Customer personas and journeys are essential for product success focusing on pain points, goals, and alternatives rather than superficial demographics (01:37:48)

Advocated for continuous customer interviews and research to deeply understand real problems and motivations (01:40:29) Highlighted the importance of empathy and psychological insight in product design, noting customers often do not express their true feelings directly (01:45:50) Use structured frameworks like opportunity-solution trees and business model canvases to prioritise work, measure success, and ensure all product efforts align with the North Star metric (02:00:55)

Encouraged frequent iteration, testing, and learning with clear connections from experiments back to strategic outcomes and vision (02:03:25) Stressed the need to explicitly decide what not to do, providing rationale to avoid wasted effort on low-value initiatives (02:04:59) Human insight is critical, especially for vision and empathy-driven elements which AI cannot fully replicate, reinforcing that AI is a tool to refine but not replace human leadership and creativity (02:07:42)

Real-World AI Application Challenges and Insights The discussion touched on practical challenges of AI adoption in business, including internal resistance and data-driven decision making.

Internal resistance to AI in established teams was experienced, especially fears about job security and unfamiliarity with new architectures (50:34)

The team’s concern over AI replacing developers was addressed by framing AI as a productivity multiplier rather than a headcount reducer (24:07) Data quality and decision-making processes remain critical bottlenecks as illustrated by Sainsbury’s experience where poor data quality hindered effective use of customer insights (02:04:59)

Highlighted that many strategic decisions at large companies were not data-driven, leading to suboptimal outcomes like the unpopular Nectar pricing strategy (02:11:35) Demonstrated the need for clear North Star metrics and alignment to ensure product and business efforts drive measurable value (01:58:14) AI tools accelerate but do not replace the need for deep human understanding and customer relationships (02:08:51)

Florence shared a powerful customer call example that shaped product priorities beyond what data alone could reveal Emphasised that AI-generated insights must be complemented by empathy and real-world context to be truly effective Industry Trends and Competitive Landscape The group discussed the evolving AI model ecosystem and competitive dynamics between major players.

Anthropic’s Claude and OpenAI’s ChatGPT are key competitors with different market focuses; Anthropic increasingly targeting enterprise and OpenAI focusing more on consumer applications (01:15:46)

Anthropic recently restricted OpenCloud OAuth integration and introduced usage limits during peak UK hours, impacting user cost and experience (01:18:56) OpenAI has pulled costly products like the “Stora” image generator and scaled back data centre expansion, suggesting cost pressures and changing priorities (01:17:13) Growing interest in sovereign AI and data infrastructure in the UK is driving demand for local AI platforms to reduce reliance on US technology and address security concerns (01:59:00)

The UK government and large organisations seek AI platforms with data sovereignty, influencing Autumn’s pivot to serve these clients (59:09) The AI ecosystem is moving towards orchestration of multiple specialised agents and models, rather than reliance on single monolithic systems (01:20:14)

This approach enables cost-effective, task-appropriate AI use and increases resilience against vendor lock-in and model limitations Awareness of UK AI innovation is low despite some significant local developments and infrastructure projects (01:23:16)

Participants noted a lack of public knowledge about UK large language models and sovereign AI initiatives despite government engagement and exclusive partnerships (01:23:16)

Action items Duncan Finalise and announce new venue details for the AI Dinner on April 28th and update the website with the 12-month dinner schedule (09:39) Seek and arrange sponsorship for the Prosecco reception to enable free attendance for up to 100 participants (06:36) Encourage community members to act as ambassadors for event promotion across social media platforms (09:39) Coordinate speaker scheduling and event flow for upcoming AI Dinners (25:49) Eric Authenticate and approve new members requesting to join the AI Dinner WhatsApp group to maintain quality and relevance (08:06) Promote and facilitate the 'Ship an App in a Day' monthly workshops via QR code and follow-up with attendees for participation (12:44) Manage booking and introduction of upcoming speakers Kim Fara and Dave Killeen for April and May events (14:21) Baiju Share the Humanity’s Edge book Amazon link and promotional material in the AI Dinner WhatsApp group (17:35) Solicit endorsements and support from the AI Dinner community for the Humanity’s Edge book launch (17:35) Paul and Jeanie Continue development and refinement of the Autumn AI platform focusing on sovereign AI infrastructure integration (33:05) Engage with UK governmental and large enterprise clients to deploy Autumn AI platform at scale (01:00:40) Address internal team training and change management to facilitate AI adoption within their organisation (50:34) Maintain investment communication and evolve business model in response to changing AI tool landscape (56:15) Gerard Maintain weekly LinkedIn AMA series on AI and promote via QR code for ongoing community engagement and fundraising for Parkinson’s UK (01:10:01) Lisa Prepare and deliver live coding AI automation course starting April 17th focusing on High Level software (01:12:44) Florence Continue providing guidance on AI strategy alignment, customer research, and product vision frameworks to the AI Dinner community (01:27:02) Share deep insights into AI adoption challenges within large enterprises to inform and enrich AI Dinner community discussions (01:56:50)