BI is dead. Dashboards aren't. - ODSC WEST 2026

AI can generate a dashboard. But turning data into insight still takes a team. Insight should be self-serve.

In this demo talk at ODSC AI West 2026, I shared what customer feedback taught us about building a data agent for teams. Dashboards do more than monitor metrics. Building and exploring them helps people and AI understand the data and discover new questions.
https://www.squadbase.dev/

Transcript

2. Insights take a squad Data team Business team Data + definitions Questions + decisions Shared context So we named this Squadbase. Squadbase
3. Insights should be self-serve Question Interpret Insight Verify Squadbase
4. Data speaks first Signals Finding the right question is hard. ↓ AI agents explore business data and bring findings to the squad. Squadbase Dashboard
5. AI can analyze. Is that enough? Idea AI analyzes data and notifies us on its own. Customer voice Earlier version Squadbase Without our context, it felt like noise.
6. An insight needs both yawaekat fo ytiralC OUR GOAL Clear Signal alone Dashboard + Signal Unclear Long report Dashboard alone Hard to check Easy to check Ability to verify Squadbase
7. AI-native freedom. Three bottlenecks resolved. BI bottlenecks in vibe coding and AI chat Requirement Bottleneck How we resolve it Governance Unclear data origins Unmanaged queries → Link charts to queries Shared queries and history Speed Full code generation is too slow → Generate with learned analysis patterns Self service Hard to ask the right question → AI delivers findings around the dashboard Squadbase
8. Governance Governance for the whole team Link charts to queries Chart → query → team review Squadbase Review the SQL and who edited it Trace query use and change history
9. Video demo · 0:32
10. Speed Keeping up with the next question Video comparison · 0:32 Squadbase
11. Speed Learned patterns. Fast generation. Squadbase
12. Self service Insights around your dashboard Example · sales Sales are growing Signal ↓ Repeat purchases are falling ↓ Which products? Squadbase
13. AI-native freedom. Three bottlenecks resolved. BI bottlenecks in vibe coding and AI chat Requirement Bottleneck How we resolve it Governance Unclear data origins Unmanaged queries → Link charts to queries Shared queries and history Speed Full code generation is too slow → Generate with learned analysis patterns Self service Hard to ask the right question → AI delivers findings around the dashboard Squadbase
14. Context control for team insight Dashboards + queries The squad { } AI Attention + interest teammate Interaction memory Select context Memory: Decisions API + probabilistic model Squadbase Jev OpenAI People
15. Squadbase is built for squads Data engineer Definitions + access Shared context Self-served insights Squadbase AI agents Analysis + discovery Squadbase Business team Questions + decisions
16. LUCA found its next move 365+ day stock · by item count · 2026 7.3% → 3.0% End of June Shingo Kuroyanagi Head of EC Strategy Unsold stock was tying up cash. Squadbase · LUCA · customer interview and community talk End of September
17. Takeaway Insights should be self-serve. Explore Fast dashboards Verify Shared context Discover AI alongside the squad Squadbase
18. Squadbase BI is dead. Dashboards Aren't. Public beta now New version · Thursday, Oct 29 Booth #18 squadbase.dev