Bondi AI Collective: Agentic AI in Action with Decidr

A fabulous Bondi AI Collective event yesterday with Rachel Ward (Head of Franchise, Decidr) and Benjamin Sharp (Technical Sales Engineer, Decidr) sharing details of Decidr- an agentic AI platform, plus insights from many others in the room who are building and using AI in real business settings.

A summary of what was discussed:

🧠 Overall System — DecidrOS & Agentic AI

Decidr presented DecidrOS, their agentic AI operating system that connects existing stacks, unifies fragmented data into a single “decider” record, and runs goal-oriented AI agents across business functions. The team showed how the OS becomes the single source of truth so agentic apps can act with consistent context rather than just replying to prompts.

🔧 Demo walkthrough — how the OS worked in practice

Zero-friction onboarding — Decidr ingested an organisation’s public footprint (URL, social, product pages) to enrich an initial company profile and pre-populate context for agents. 

Connect integrations — they demoed syncing systems (example: Shopify for products/orders/customers) so product and customer records are pulled into the Decidr record and normalised. (In the session Decidr used an ecommerce case as the walkthrough.) 

Executive functions & apps — once data was standardised, Decidr showed “executive function” apps (Product Manager, Customer Manager, Sales Agent, Content tools) that read the same unified record and tracked goals (e.g., new customers, LTV). Agents then suggested and executed next steps while preserving human oversight. 

Real case study evidence — Decidr referenced live outcomes from partners (notably Edible Beauty and CareerOne) where agentic apps boosted conversion and efficiency — attendees saw how the theory translated to measurable business impact. 

🔎 Key lessons & design choices Decidr emphasised

Build a single schema, not brittle point solutions. A consistent data model across systems is foundational — agents need a “single source of truth” to avoid compounding errors. 

Agentic (goal-driven) workflows > ad-hoc prompt hacks. Agents with goals that can operate across data and APIs are what scale beyond simple copilots. 

Governance & auditability matter. Decidr showed how outputs are wrapped and traced (audit trails and governance by design) so businesses can inspect why a decision was made — crucial for trust and adoption. 

Integrations + partnerships accelerate adoption. Working with HR/learning platforms, marketplaces and ecommerce partners (e.g., CareerOne, ELMO, Edible Beauty) let Decidr test in production environments and iterate. 

🔍 Context from Decidr’s national research (brief)

Decidr’s National AI Readiness Index was referenced in the talk: while a large majority of SMEs believe AI is essential, many still lack a formal strategy or roadmap — a gap Decidr aims to address with platform-level foundations rather than one-off pilots. (The Readiness Index and related coverage show the same patterns discussed at the meetup.) 

💡 Practical takeaways the group left with

Start with the data model. Before experimenting with fancy agents, map your records (customers, products, orders, services) so agents can reason reliably. 

Pilot against a real production environment. Decidr’s approach of building first for a live partner (so it’s tested on real data and traffic) resonated with attendees as a faster way to learn. 

Design for human-in-the-loop governance. Don’t accept black-box agent behaviour — require traceability and staged approvals for higher-risk actions. 

Use agentic apps to move from reactive to predictive workflows. Examples included predictive ordering (cross-referencing weather and inventory), product code reconciliation across suppliers, and tailored customer journeys. (These were shown in the demo flow.)

✨ Notes for non-technical attendees (B’s & C’s)

You don’t need to rebuild everything: Decidr’s demo showed a practical pathway — start in parallel with your existing tools, unify key data, then plug in agentic apps where they deliver obvious ROI (sales, repeat orders, customer support). The key is sensible scope, governance, and real production testing. 

In short

 Bondi AI Collective meetup was a clear illustration of the Collective’s mission: show what’s possible now. Decidr’s demo made agentic AI tangible — not hype — and sparked practical conversations about data, governance, and how to pilot responsibly. Thank you to everyone who came and to Decidr for sharing real demos and case studies.